Piano Music Sight Reading:
The Transfer of Expertise to Non-Musical Domains and the Effect of Visual and
Auditory Interference on Performance.
Patricia Jean Arthur
A thesis in fulfilment of the requirements for the degree of
Doctor of Philosophy
School of Optometry and Vision Science
Faculty of Science
March 2017 i PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet
Surname or Family name: ARTHUR
First name: Patricia Other name/s: Jean
Abbreviation for degree as given in the University calendar: PhD
School: Optometry and Vision Science Faculty: Science
Title: The Transfer of Expertise to Non-Musical Domains and the Effect of Visual and Auditory Interference on Performance.
Abstract 350 words maximum: (PLEASE TYPE) Music sight-reading is an extensively studied area of visual processing expertise. Past research has established that the eye movement (EM) patterns of expert and non-expert music sight-readers reflect those of text readers; that Working Memory Capacity (WMC) and that engagement in long periods of deliberate practice (DP) are all characteristic of sight-reading expertise. The effect of altering visual parameters (blur, contrast and size), and semantic information (spacing and punctuation), is well documented for text-reading, but less so for music sight-reading. Research into the transfer of such expertise into non-musical domains still has many unanswered questions; as are the effects that other sensory input might have on the performance of a music-related visual processing tasks. The thesis examines these issues in relation to expert and non-expert music sight-readers and, where possible, musicians and non-musicians. The key features of expertise were initially explored – speed of performance, WMC and amount of DP – and their relationship to a specific level of piano music sight-reading ability. These features were all found to be significantly superior in those who could successfully perform a 6th Grade AMEB sight-reading assessment piece. Having thereby assigned expertise, the effects on EMs were measured between the expert and non-expert sight-readers when the blur, contrast and size features of the score were manipulated, the metronome introduced and the notation visually disrupted. Sight-reading expertise was further explored in relation to non-musical forms of sensory processing: auditory tone frequency and modulation discrimination, visual and perceptual spans, visual working memory, word reading speed, basic visual search and visual search using music-like targets with congruent or non-congruent simultaneous auditory interference. The findings were discussed within the DP and innate abilities frameworks. In conclusion, expert music sight-readers frequently exhibit enhanced visual processing skills not only beyond those of non-musicians, but beyond those of non-expert music sight-readers in both related and non-related domains and show distinct effects in response to auditory interference. These are new aspects of the cognitive processing repertoire of expert music sight-readers and has implications for music educators and the use of music as remediation for other visual processing deficiencies.
Declaration relating to disposition of project thesis/dissertation
I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.
I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only).
………………………………………………………… ……….…02/11/2016…………… … Date Signature
The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.
FOR OFFICE USE ONLY Date of completion of requirements for Award: COPYRIGHT STATEMENT
‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright
Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.
I also authorise University Microfilms to use the 350-word abstract of my thesis in
Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'
Signed ......
Date ...... 24/03/2017......
AUTHENTICITY STATEMENT
‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’
Signed ......
Date ...... 24/03/2017......
ii ORIGINALITY STATEMENT
‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’
Signed ......
Date ...... 24/03/2017 ......
iii
DEDICATION
To all those who ever wondered if learning to read music
really did make you smarter……
iv ACKNOWLEDGMENTS
What comes to mind when you reflect upon writing a thesis? There will be a variety of reactions depending on whom you ask. Most recall the stress of deadlines, sleepless nights, writer’s block, computer glitches and formatting nightmares; not to mention the lack of social life! What is frequently lost is the recognition of the myriad of support structures that are integral to the process; without which, the job would never get done!
My ‘journey’ has not been one of the typical honours-student-to-PhD kind and most definitely not possible without such support. What hope do you have if you had not achieved stellar grades in your undergraduate degree some 30 years prior? Would anybody think your idea was worth pursuing? Where do you start when you have a research question that has been on your mind for decades? You head for your alma mater, the University of New south Wales School of Optometry and Vision Science.
There I found Sieu Khuu who took on the role of supervisor while Blanka Golebiowski went in to bat for me in order to obtain an Australian Postgraduate Award (APA). As I had not achieved honours in my undergraduate Optometry degree, I had to enrol in the
Research Masters programme and apply to transfer into the PhD programme after one year provided my work was of the appropriate standard - it was and here we are! My thesis had an identity crisis mid-stream, where I needed more musical guidance and
Diana Blom from Western Sydney University Music became involved. Diana replaced
Mei Boon who became a keen supporter from the sidelines and supplier of many a 1st year student subject. Other academics employed me as their Teaching Fellow: Barbara
Junghans, Maria Markoulli and Sieu Khuu; while others gave me teaching opportunities
v in their courses; Lisa Asper, Rosemary Paynter and Maitreyee Roy. The UNSW
Optometry Clinic also became a mainstay of experiences as well as income - Nicole
Grskovic always looking out for extra clinics for me to supervise and Kath Watt acting as my reader and providing fantastic feedback throughout my candidacy. Then there was
Brien Cheng – computer guru! Can’t forget my fellow-travellers, the other candidates from Vision Science – Eric Chung and Vanessa Honson. Eric always so willing and able to help with Matlab issues and participate in my experiments and Vanessa feeding me chocolate in the stair well when I was having a moment and the car trips home where we solved all the world’s problems!! A massive thank you to all.
Lastly, my family. Unfortunately, Mum did not make it to see me graduate as she passed away a few weeks ago and Dad would no doubt have bought me a tin of ‘Dr Pat’ pipe tobacco if he was here! My daughter, Robyn, had a stint crunching numbers for me and developed an understanding for my subject beyond the casual and has always been keen to know how it’s all going. My son Richard, with his philosophical bent, was always able to turn my mind away from my own bubble to ponder wider issues. Larfew guys!!!!
My husband, John. There is no way any of this would have been possible without his endless encouragement, tolerance and practical support – like feeding me and making cups of coffee so that I could continue to write uninterrupted. He is truly a rare and wonderful man that I love more than ever.
A PhD cannot happen in a vacuum. It can never be a solo effort. I am so very grateful to you all for carrying me to this place.
March, 2017 vi ABSTRACT
Music sight reading is an extensively studied area of expertise. Past research has established that the eye movement (EM) patterns of expert and non-expert music sight- readers reflect those of text readers; that Working Memory Capacity (WMC) and that engagement in long periods of deliberate practice (DP) are all characteristic of sight- reading expertise.
The effect of altering visual parameters (blur, contrast and size), and semantic information (spacing and punctuation), is well documented for text-reading, but less so for music sight-reading. Research into the transfer of such expertise into non-musical domains still has many unanswered questions; as are the effects that other sensory input might have on the performance of a music-related visual processing tasks. The thesis examines these issues in relation to expert and non-expert music sight-readers and, where possible, musicians and non-musicians.
The key features of expertise were initially explored – speed of performance, WMC and amount of DP – and their relationship to a specific level of piano music sight-reading ability. These features were all found to be significantly superior in those who could successfully perform a 6th Grade AMEB sight-reading assessment piece.
Having thereby assigned expertise, the effects on EMs were measured between the expert and non-expert sight-readers when the blur, contrast and size features of the score were manipulated, the metronome introduced and the notation visually disrupted. Sight- reading expertise was further explored in relation to non-musical forms of sensory processing: auditory tone frequency and modulation discrimination, visual and perceptual vii spans, visual working memory, word reading speed, basic visual search and visual search using music-like targets with congruent or non-congruent simultaneous auditory interference. The findings were discussed within the DP and innate abilities frameworks.
In conclusion, expert music sight-readers frequently exhibit enhanced visual processing skills not only beyond those of non-musicians, but beyond those of non-expert music sight-readers in both related and non-related domains and show distinct effects in response to auditory interference. These are new aspects of the cognitive processing repertoire of expert music sight-readers and has implications for music educators and the use of music as remediation for other visual processing deficiencies.
viii TABLE OF CONTENTS
Page
DEDICATION ...... iv
ACKNOWLEDGMENTS ...... v
ABSTRACT ...... vii
TABLE OF CONTENTS ...... ix
ABBREVIATIONS ...... xiv
LIST OF TABLES ...... xv
LIST OF FIGURES ...... xvi
CHAPTER 1 ...... 18
Literature Review ...... 18
Preamble ...... 18
Introduction ...... 19
Nature vs Nurture ...... 21
Nature: the evidence ...... 21
Nurture: the evidence...... 24
Nature + Nurture: the evidence ...... 25
Eye Movements ...... 30
Cognitive Processing ...... 38
Perceptual and Visual Spans ...... 38
Tone Frequency and Modulation Discrimination ...... 40
Rapid Automatized Naming ...... 42
Visual Search ...... 43 ix Thesis Aims ...... 44
Thesis Structure ...... 45
CHAPTER 2 Methods ...... 47
Eye movement studies ...... 47
Participants ...... 47
Stimulus ...... 48
CHAPTER 3 Sight Reading Expertise...... 54
Introduction ...... 54
Study 1: The Survey ...... 56
Hypothesis ...... 58
Method ...... 59
Results ...... 59
Discussion ...... 63
Conclusions ...... 68
Study 2: Speed of Performance ...... 68
Hypothesis ...... 68
Methods ...... 69
Results ...... 69
Discussion ...... 70
Conclusion ...... 72
Study 3: Working Memory Capacity (WMC) ...... 72
Background ...... 72
Hypothesis ...... 75
Methods ...... 75
Results ...... 77 x Discussion ...... 78
Conclusions ...... 78
CHAPTER 4 Eye Movements and Music Sight Reading Expertise ...... 80
Study 1: The Effect of Blur and Size on EMs when Sight Reading Music ...... 80
Introduction ...... 81
Hypothesis ...... 82
Methods ...... 82
Results ...... 83
Discussion ...... 89
Conclusion ...... 97
Study 2: The Effect of a Visually Disrupted Score on EMs when Sight Reading ...... 97
Background ...... 98
Hypothesis ...... 101
Methods ...... 101
Results ...... 102
Discussion ...... 106
Conclusion ...... 109
Study 3: The Effect of Imposed Time-Keeping on EMs when Sight Reading ...... 109
Introduction ...... 110
Background ...... 111
Hypothesis ...... 112
Methods ...... 112
Results ...... 113
Discussion ...... 117
Conclusion ...... 120 xi Expertise, Music Sight Reading and Eye Movements – a Summary ...... 121
CHAPTER 5 Cognitive Processing and Music Sight Reading Expertise ...... 125
Background ...... 125
Study 1: Working memory Capacity (WMC) ...... 126
Background ...... 126
Hypothesis ...... 127
Methods ...... 127
Results ...... 128
Discussion ...... 130
Conclusions ...... 132
Study 2: The Perceptual and Visual Spans ...... 132
Introduction ...... 133
Hypothesis ...... 135
Methods ...... 135
Procedure ...... 137
Results ...... 138
Discussion ...... 142
Conclusions ...... 147
Study 3: Rapid Automatized Naming of Words - WordRAN ...... 148
Background ...... 148
Hypothesis ...... 149
Methods ...... 150
Results ...... 151
Discussion ...... 152
Conclusions ...... 154 xii Study 4: Frequency Discrimination and Modulation ...... 154
Background ...... 154
Hypothesis ...... 156
Methods ...... 156
Results ...... 158
Discussion ...... 160
Conclusion ...... 162
Study 5: Visual Search of Musical and Non-Musical Targets ...... 163
Background ...... 163
Hypothesis ...... 165
Methods ...... 165
Results ...... 168
Discussion ...... 178
Conclusion ...... 184
CHAPTER 6 General Discussion ...... 186
Eye Movement Studies ...... 188
Cognitive Processing ...... 195
Solving for ‘X’ ...... 198
Conclusions ...... 207
APPENDICES ...... 209
APPENDIX 1 MUSICAL EXPERIENCES SURVEY ...... 209
APPENDIX 2 RECRUITMENT POSTER 1 ...... 210
APPENDIX 3 RECRUITMENT POSTER 2 ...... 211
BIBLIOGRAPHY ...... 212
xiii ABBREVIATIONS
AMEB Australian Music Examination Board
AMus Associate Diploma in Music
CPS Critical Print Size
DP Deliberate Practice
EMs Eye Movements
FD Frequency Discrimination
FM Frequency Modulation
MAR Minimum Angle of Resolution
MM Metronome Marking
RAN Rapid Automatized Naming
VPL Visual Perceptual Learning
VSTM Visual Short Term Memory
WMC Working Memory Capacity
WordRAN Random Automatized Naming of Words
xiv LIST OF TABLES
Table # Page #
1. Eye Movement Research Methodologies 36
2. Survey p-values between Expert and Non-Expert Sight Readers 60
3. 3a Correlation between Survey Categories – Experts 64
3b Correlation between Survey Categories – Non-Experts 65
4. Summary of significant effects in EM Studies 123
5. Summary of Processing Features 1 131
6. Summary of Processing Features 2 146
7. Summary of Processing Features 3 153
8. Summary of Processing Features 4 161
9. Summary p-values for sensitivity in visual search 176
10. Summary of Processing Features 5 183
xv
LIST OF FIGURES
Figure # Page #
1. A Sample Music Excerpt 47
2. Experimental Setup – Eye Movement Data Collection 49
3. EM Results – Size vs against Time 70
4. WMC Model 73
5. Screen Shot of WMC Task 76
6. WMC Results – Expert vs Non-Expert Sight Readers 77
7. EM Results – Blur and Size vs Expertise 84
8. Music Score - Temporal Aspects of Music Notation 100
9. Music Score – Normal vs Disrupted Spacing 102
10. EM Results – Normal vs Disrupted Spacing 103
11. EM Results – Metronome vs No Metronome 115
12. WMC Results – Musicians vs Non-Musicians 128
13. WMC – Non-Musicians vs Non-Expert vs Expert Sight Readers 129
xvi 14. Visual Spans – Experimental Design 137
15. Visual Spans – Musicians vs Non-Musicians 139
16. Visual Spans - Non-Musicians vs Non-Expert vs Expert Sight Readers 141
17. WordRAN Results 151
18. Sample Sound Waves 154
19. FM Results 157
20. FD Results 158
21. Visual Search Targets 165
22. Visual Search Experimental Design 166
23. Visual Search Results – Landolt C 169
24. Visual Search Results – Musical Targets vs Time 171
25. Visual Search Results – Musical Targets vs d’ 173
26. Visual Search – All Targets vs d’ 174
xvii CHAPTER 1
Literature Review
Preamble
This thesis investigates the visual processing capabilities of music sight readers. Sight reading is a subset of music skills where prompt performance of a piece of music is required when read directly from notated music, known as the score, with little to no prior experience of the piece to be played. This skill is a high priority for repetiteurs, accompanists and a very useful skill-set for piano teachers, performers and musicians generally. Musicians exhibit a vast range of ability in sight reading. Past research has shown that some musicians attain ‘expert’ status in this domain. Expertise in music sight reading has been shown in the patterns of (EMs that are employed when reading notated music (Kinsler & Carpenter, 1995) and that these patterns resemble those utilised by expert readers of text (Sloboda, 1977; Wurtz, Mueri, & Wiesendanger,
2009). Similarly, the EMs of unskilled or non-expert music sight readers have also been shown to resemble those of novice text readers. However, conclusions relating to expertise differences are problematic due to inconsistencies in the definition of what constitutes a sight reading ‘expert.’
Recent fMRI studies have shown that the neural processing networks involved with expertise in facial recognition, the fusiform face area, are also involved in object recognition in those who have specific object expertise (McGugin, Gatenby, Gore, &
Gautier, 2012), like recognition of car makes and models. However, little is known about how such expertise might transfer to other domains. For example, do those car experts also have superior visual processing skills in relation to numbers, letters or
18 words for example? In the present context of music sight reading, if music sight reading expertise was to be clearly delineated, then testing visual processing expertise using non-musical targets could be carried out to explore these relationships. In addition, comparisons between musicians and non-musicians could also then be examined.
Of greater interest, however, might be differences that music sight reading experience might confer upon non-expert sight readers compared with non-musicians. This can only be investigated once an objective and reliable method of assigning music sight readers into expert or non-expert groups has been found. This thesis firstly considered a definition of sight reading expertise and confirmed its validity. It is from this position that comparisons were able to be made between non-musicians and expert and non- expert sight readers in relation to visual processing abilities in non-musical domains.
The balance of Chapter 1 outlines the concepts of expertise in general terms. After a discussion of the nature vs nurture debate in relation to expertise, EMs are discussed in greater detail in relation to music sight reading, previous research and musicians’ brain structures. Cognitive features of expertise are then outlined, concluding with thesis aims and thesis structure.
Introduction
Is an expert the result of ‘practice makes perfect’; were they ‘born that way’ or a bit of both? The literature is divided on this issue. Supporters of the ‘practice makes perfect’ or deliberate practice (DP) philosophy, as championed by Anders Ericsson (Ericsson,
Krampe & Tesch-Romer, 1993) agree with the authors’ assertion on page 3 of a more recent paper that expertise is “superior reproducible performance, generally emerging only after extended periods of deliberate practice that result in subsequent physiological
19 adaptations and complex cognitive mechanisms” (Ericsson, Roring & Nandagopal,
2007). The amount of DP required is approximately 10,000 hours or 10 years and innate features of IQ, such as working memory capacity (WMC), lose significance as the amount of DP increases (Kulasegaram, Grierson, & Norman, 2013 ). On the other hand, it has been argued that innate ability is required which is observable in an individual, usually at a young age, but then perfected by DP, for example, (Ackman, 2014;
Campitelli & Gobet, 2011; Hambrick et al., 2014; Kulasegaram et al., 2013 ; Lombardo
& Deaner, 2014; Meinz & Hambrick, 2010; Plomin, Shakeshaft, McMillan, &
Trzaskowski, 2014).
Difficulties arise, however, when defining expertise in the musical domain (Farrington-
Darby & Wilson, 2006) and particularly so in relation to prodigious musical talent
(Ruthsatz & Urbach, 2012). One would rightly question whether or not there is more than one category of expertise in the musical domain and that this needs to be more precisely defined in order to substantiate any claims made by or comparisons made between studies. For example, how can the ‘reproducibly superior performance’ of music as a result of DP be compared with the improvisations of a child prodigy or musical savant? According to Simonton (2014, p68), creativity must have utility, be original and have the element of surprise or novelty (Simonton, 2007). By that account, the former cannot be said to be creative, as they are reproducing someone else’s work, often as closely as possible to the original. An adult with expertise in improvisation might be different again and worthy of consideration in this discussion. This is not the topic of this thesis – music sight reading expertise is under investigation and it is not intended to imply that one type of musical expertise is superior to another.
20 Perhaps the assertion that ‘deliberate practice is necessary but not sufficient’ to develop music sight reading expertise is more appropriate. The involvement of nature vs nurture in the process of developing expertise will further expand this idea.
Nature vs Nurture
The nature versus nurture debate is not new. Individual differences in performance were explained in terms of innate abilities as long ago as Plato or by environmental influences according to Aristotle (Ruthsatz, et. al, 2014). Neither extreme position is sustainable (Ackman, 2014) and a more measured consideration of both influences is the current position (Plomin et al., 2014). Nevertheless, the debate has not subsided and evidence gained from the use of new technologies, such as fMRI, has broadened our understanding as much as it has sustained the debate.
Nature: the evidence
Evidence for innate capabilities being exclusively responsible for expertise has been drawn from a variety of disciplines including sport, the arts, languages and chess. The existence of child prodigies is difficult to explain any other way as, by definition, their age precludes them from having had time to develop their skills through extensive practice. The innate talents of a child prodigy have been defined as being genetically transmitted, evident from an early age, predictive of adult ability, rare and (relatively) domain specific (Howe, Davidson, & Sloboda, 1998).
A recent survey of child prodigies found a number of genetic characteristics that were common within the group (Ruthsatz & Urbach, 2012). Of the 8 children studied, 3 were diagnosed with autism, 50% had relatives with autism and all scored as high as autism sufferers in the attention to detail category of Autism-Spectrum Quotient testing. All
21 had enhanced working memory capacity (WMC) and raised Full Scale Intelligence
Quotient, with the average being in the gifted range. All were recognised as gifted well before the age of 10. While the mean IQ may have been in the gifted range, 2 of the 7 children measured (28.6%) had full scale IQ scores barely above average: 108 and 112
(p 421) whereas their WMC was consistently in the 99th percentile. Of this group, only 2 were over 18 years of age and both sustained successful careers outside of their original domain of expertise: one moved from puzzles to music to cooking/catering and the other from music to visual art. As the other participants were less than 18 years of age, it will be some time before it could be confidently stated that early talent was predictive of their adult abilities, but it does not appear that expertise is necessarily domain specific. Certainly the connection between prodigiousness, autism and its prevalence in such families is profound and worthy of further investigation by those suitably qualified. High full scale IQ was not an absolute prerequisite, but one subset was –
WMC – which is covered in more detail later in this review.
Superior sporting achievements are claimed to be a result of innate abilities. For example, in a recent analysis of elite sprinters, many of their characteristics coincided with Howe’s features of child prodigies (Lombardo & Deaner, 2014). However, consideration of DP confounded the assertion that elite sportspeople were the result of talent alone. Lombardo found that in order to achieve world class status, DP was required, but was as little as 3 or 7.5 years. Sprinters have been known to transfer to other sports, but they generally involve elements similar to their original area of expertise: transferring from sprinting to bobsled or to various codes of football. As with any scientific research, methodological assumptions need to be valid and clearly identified for sensible results to occur. For example, height was found to be unrelated to
22 the NBA’s top point scorers and this led to the conclusion that height was not required to be a top scorer in the NBA (Detterman, 2014). The author pointed out that such an obviously absurd conclusion was reached because the methodology was flawed - the analysis ignores that fact that this cohort already had a median height 3.4 SD higher than the general population. Therefore, the natural endowment of height is required for an individual to excel at professional basketball along with requisite DP.
Chess expertise also claims innate talent to be necessary for expert performance. In young players, full scale IQ is higher but is not a representative trait in adult players
(Campitelli & Gobet, 2011). The innate feature that is consistently significant in chess players is WMC, even when DP has been factored out (D. Hambrick et al., 2014; Meinz
& Hambrick, 2010). Similarly, expertise has been extensively researched in the field of music sight reading. WMC, apart from being the principal intrinsic feature of child prodigies as previously discussed, was also found to be predictive of sight reading skill
(Meinz & Hambrick, 2010). If WMC has been found to be key to the development of expertise in chess and music sight reading, perhaps a difference in brain structure can be found.
Studies of musicians’ brains have found certain structures to be different when compared with non-musicians. There is a difference in the size of the superior parietal region of the brain in musicians (Gaser & Schlaug, 2003). This area is responsible for integration of multimodal sensory information and is important in the sight reading of music: visual input to motor output with auditory feedback. Also, musicians have a greater degree of bilateral neural connectivity than non-musicians, in that learning to play an instrument involving both hands appears to facilitate extra myelination of fibres
23 in the corpus callosum, leading to greater equality in the speed of transfer and balance between hemispheres (Patston et al., 2007).
Nurture: the evidence.
The principal advocate for the role of nurture in the form of DP that results in expert performance is Anders Ericsson. Much of the present nature vs nurture debate stemmed from his 1993 paper entitled ‘The role of deliberate practice in the acquisition of expert performance’ and revolved around the claim that repeatable expert performance in any domain can only be achieved after extensive amounts of DP (Ericsson et al., 1993).
DP is described as consisting of activities prescribed and modified utilising feedback with the intent of improving performance; as distinct from doing activities for work or even for enjoyment. His most compelling argument in favour of DP alone for attaining expertise is that advanced training techniques, particularly in sports, have caused an increase in performance levels beyond those attained without training and due to innate abilities. As a result, experts develop abilities that allow for more efficient processing without relying on short term memory and the physical speed of executing a movement.
There is the ability to anticipate what is ahead, plan according to clues within the terrain and execute economical movements to achieve the desired goal. This process is known as ‘chunking’ and results in the characteristic speed of expert performance and is a direct result of prolific and structured domain knowledge.
Supporters of the DP model do not accept that innate ability plays a part in the development of expertise. What is said to be ‘talent’ is really a child who is more advanced or successful or demonstrates more enjoyment than their peers in a particular activity and is then targeted for DP - but they are not advanced relative to an adult.
24 They also claim that variations in the amount and quality of DP are sufficient to account for individual differences (Lombardo & Deaner, 2014) and early introduction of DP is advantageous (Campitelli & Gobet, 2011; Wai, 2014).
Nature + Nurture: the evidence
Studying twins could potentially substantiate the claim of innate ability or DP in so far as there have inbuilt controls for nature (identical DNA in monozygotic twins and different in dizygotic twins) and nurture (mostly brought up in the same environment).
Recent research tested 12 year old twins for expertise in reading (Plomin et al., 2014).
Every possible combination of twin type and gender was examined, 4955 twin pairs in all, with the result that over half of the difference in expertise was due to genetic factors.
However, the study does not take into account or try to measure the extent that personality traits - such as motivation - might factor into the analysis. This has been shown to affect the ability to sustain DP (I. Gauthier et al., 2014; Simonton, 2014; Wai,
2014) and DP is certainly a requirement in learning to read. The results of the Plomin et al. study suggest that some combination of nature and nurture is necessary for the development of processing expertise.
The majority of experts in the field accept that both nature and nurture need to be part of the equation (Campitelli & Gobet, 2011; Detterman, 2014; I. Gauthier & Bukach, 2007;
Howe et al., 1998; Kulasegaram et al., 2013 ; Lombardo & Deaner, 2014; Meinz &
Hambrick, 2010; Simonton, 2007, 2014; Wai, 2014). Difficulties arise when different amounts of DP, innate ability and intelligence have been found to exist in varying ratios for different domains as stated earlier. A model known as ‘Summation Theory’ has been proposed to take into account all aspects that influence expert performance (Ruthsatz,
2014; Ruthsatz, Ruthsatz, & Ruthsatz-Stephens, 2014). According to this theory,
25 performance or Y’ = Xg + Xds + Xp; where Xg is general intelligence, Xds is domain specific skills and Xp is deliberate practice. In this way, expertise in most fields can be accounted for. A prodigy will lack Xp, but makes up for it in Xg and/or Xds and what a sportsperson might lack in Xg would be bolstered by Xds and Xp. Although somewhat simplistic, the Summation Theory is an attempt to acknowledge that expertise is a multifactorial entity.
Considering the difference in brain structures that have been found to differ between musicians and non-musicians, are musicians born that way? A recent study considered whether children who expressed an interest in music had measurable differences in specific brain structures that are different in the adult musician brain and it was found that there were none (A. Norton et al., 2005) 2005). Their goal is now to develop a longitudinal study to assess whether a relationship exists between brain structure and those children who seek music study and later become musicians or show evidence of giftedness. A positive correlation here would bolster the ‘nature’ argument for being the basis of expertise. Norton et al. suggest that a randomised control of musical intervention would be the only way to confirm the ‘nurture’ position.
Working Memory Capacity, (WMC), as previously mentioned, features extensively throughout the expertise literature. It is a measure of an individual’s ability to recall information following the presentation of supplementary material. It can be tested in the verbal domain, for example by recalling the last word of each sentence after a number of sentences are read out. Non-verbal WMC can be assessed by recalling the order of numbers tapped out from a display of numbers. As discussed earlier, it is significantly related to child prodigy, expert chess players and is predictive of music sight reading ability. A recent study proposed that both DP and WMC were co-predictors of expertise
26 by analysing the results from studies that link expertise with WMC, indicating that both nature and nurture are necessary for expertise to develop (Kulasegaram et al., 2013 ).
Expertise in facial recognition deserves special attention as it examines an area in which sighted people can be considered to have visual expertise of long standing (I. Gauthier
& Bukach, 2007; McKone, Crookes, Jeffery, & Dilks, 2012; Tso, Au, & Hsiao, 2014).
The viewing of faces has been shown to activate an area of the brain called the fusiform gyrus and has become known as the Fusiform Face Area or FFA (Kanwisher,
McDermott, & Chun, 1997). Whether or not the FFA is responsible for facial recognition only or is an area involved in holistic visual processing expertise more generally is controversial and research is continuing into new aspects of the functioning of this brain area. For example, the aforementioned authors have claimed that the FFA is only for facial recognition based on fMRI data (Kanwisher et al., 1997). Such a claim has been systematically shown to be unlikely following a number of behavioural and functional experiments that factor out the role of visual expertise in other domains (I.
Gauthier et al., 2014). Beginning with the famous ‘Greebles’ experiment, Gauthier and colleagues showed that when experts, trained in the visual recognition of figures, called
‘greebles’, were tested for changes in its features, they showed activation of the FFA while subjects with no greeble exposure did not (I. Gauthier & Tarr, 2002). This led to the ‘Undifferentiated-template Hypothesis’ which proposed that the ability to process an object as a whole, or holistically, increases with expertise so that the individual parts of that whole become less important. As such, the FFA may be more of an area responsible for holistic processing of object generally (I. Gauthier & Bukach, 2007; I. Gauthier et al., 2014; McGugin, Gatenby, et al., 2012; Slotnick & White, 2013).
27 Following evidence that the right hemisphere of the brain is dominant for face processing, regardless of the influence of left to right or right to left reading practices
(Megreya & Havard, 2011), it has been shown that the right FFA responds to left and central visual field presentations related to shape processing in addition to responding to faces (Slotnick & White, 2013). This has become known as Left Side Bias (LSB) and discussion continues to surround whether or not LSB is a feature of visual processing expertise or task-dependent holistic processing. It has been shown that both expert and novice Chinese writers showed LSB when fonts were familiar, but not when they are unfamiliar (Tso et al., 2014). This would suggest that holistic processing was task dependent rather than a result of visual expertise as this is not the case for faces – recognition of familiar and unfamiliar faces show LSB (Hsiao & Cottrell, 2009). Tso went on to show that experts in Chinese character recognition, who were also experts in
Chinese character writing, showed less evidence of holistic processing than novice writers. However, this conclusion may be questioned as the stimulus presentation may not have been short enough to force the experts to elicit a ‘chunking response’ response
(Farrington-Darby & Wilson, 2006). Even though the stimulus presentation time was reduced from 600ms to 500ms, this would still have been of sufficient duration to allow more than one fixation to occur, as average saccade latency is 200ms (Cameron, 1995).
Repeating the study using shorter presentations would be required to be sure that Tso’s claim is valid. Also, it may be that this second group, who had character recognition expertise in addition to writing expertise, may have confounded the findings by adding another variable that was not part of the initial experiment.
Recently, Gauthier has stated that even with the most advanced imaging techniques available at this time, it is difficult to detect a difference in the areas of the fusiform
28 gyrus that respond to faces or to an object by a task expert (I. Gauthier et al., 2014).
Therefore, object processing expertise might appear to be the result of a ‘nurture’ or object experience process that occurs in a particular part of the brain. However, the mechanism and role that WMC plays in the development of specific object processing expertise is unclear; particularly considering that face processing expertise is a human’s most basic expert ability and in present in infants of about 3-9 months of age (Kelly et al., 2009).
The relationship between nature and nurture was recently described as one where they are joint predictors of expertise (Kulasegaram et al., 2013 ). The authors outline two models: the Circumvention of Limits and the Independence Models: each proposing a different role of WMC in relation to DP. One describes WMC as being developed by
DP and becoming less important as expertise develops – the circumvention model as proposed Ericsson, 1993. The other, that WMC can predict expertise ‘independent’ of
DP and will still be a factor in expertise even after accounting for DP. According to these models, a familiar task would indicate that the Circumvention Model would be implicated. Conversely, should a task be unfamiliar, WMC would be employed as more processing power is required (Hambrick & Meinz, 2011; Kulasegaram et al., 2013 ), but the experts would still be expected to perform at a better level than novices. This latter model, utilising WMC in relation to unfamiliar tasks, relates to sight reading – an unfamiliar task.
Nevertheless, there are people who, despite the required hours of DP, never attain expertise in a domain (Hambrick et al., 2014). The connection between of WMC and
DP is beyond question, but many areas remain to be explored. Certainly, the cause of this relationship remains elusive, but the uncovering of additional correlations will
29 inform future research. Certain brain structures involved in object recognition expertise are implicated as being similar to those for face processing (I. Gauthier et al., 2014; I.
Gauthier & Tarr, 2002; Wong & Gauthier, 2009). This alludes to the nurture aspect of expertise development. The relationship between processing expertise and WMC is not obvious and it can be argued that neither superior WMC nor significant amounts of DP are required to achieve face processing expertise. However, past studies have shown that visual processing expertise is evident in the EM patterns of readers of text and readers of music score.
Eye Movements
Wade 2010, chronicled the nature of early EM research ‘from Aristotle to Yarbus’, noting that by the mid/late19th century it was found that EMs were not smooth but were in fact a series of jerks and stoppages - saccades and fixations respectively (Wade,
2010). Yarbus, a Russian scientist, (1914-1986), first documented that eye fixation patterns change when instructions are given. In other words, EMs could be modified by cognitive influences (Tatler, Wade, Kwan, Findlay, & Velichkovsky, 2010). This has become known as the ‘Eye-Mind Assumption’ and implies that the mind is processing whatever the eye is fixating (Dee-Lucas, Just, Carpenter, & Daneman, 1982).
Proficiency in reading text has been linked to EM patterns (Underwood, Hubbard, &
Wilkinson, 1990) with a number of EM models for text reading have been proposed (K.
Rayner & McConkie, 1976). This suggests that the regulation of fixation is controlled by the information gathered by the fixations themselves (K. Rayner & McConkie,
1976). In this way, it is the pattern of fixations that indicates whether or not a word has been processed and these patterns are modified by cognitive influences (Tatler et al.,
2010). Consequently, if words in the text are unfamiliar to the reader, longer and more
30 frequent fixations should occur with an increase in regressions (backward/regressive saccades) to gain semantic understanding (K. Rayner, Chace, Slattery, & Ashby, 2006).
Conversely, if details of the text are ‘clumped’ or ‘chunked’ into recognizable groups or patterns to facilitate more efficient processing (Gobet et al., 2001), expert readers fixate less frequently and for shorter periods of time (Underwood, Hubbard, et al., 1990).
Therefore, EM patterns can be used as a means of differentiating skilled from unskilled text readers because the unskilled tend to fixate each word individually (Juhasz &
Rayner, 2003). This supports the top-down view of EMs; that they are not automatic, but are purposeful and controlled according to cognitive processes (Chace, Rayner, &
Well, 2005). The Eye-Mind Assumption also proposes that a short-term working memory or ‘buffer’ is required when more information is necessary from the text for understanding, thereby necessitating a longer fixation (Dee-Lucas et al., 1982).
Experimentation in EMs has yielded a well-established body of knowledge regarding fixation and saccadic patterns during the reading of text and its relationship to cognitive processing (Ashby, Rayner, & Clifton, 2005; Balota, Pollatsek, & Rayner, 1985;
Binder, Pollatsek, & Rayner, 1999; Dee-Lucas et al., 1982; Ehrlich & Rayner, 1981;
Fleisher, 1986; Gobet et al., 2001; Juhasz & Rayner, 2003; Just & Carpenter, 1976;
Kennison & Clifton, 1995; Meseguer, Carreiras, & Clifton, 2002; Miellet & Sparrow,
2004; Underwood, Clews, & Everatt, 1990; Underwood, Hubbard, et al., 1990). Skilled and unskilled readers can activate neural information regarding words equally well, (in fact, unskilled readers are slightly better), but unskilled readers have difficulty suppressing irrelevant or inappropriate meanings of words and will experience interference in processing (Gernsbacher & Robertson, 1995). Dyslexic readers are a specific subgroup of unskilled readers who demonstrate further evidence that EM
31 patterns are indicative of processing inefficiencies – though the root cause remains unclear. Dyslexic readers generally exhibit an excessive number of EMs, particularly regressions, and demonstrate difficulty in holding fixation (Pavlidis, 1981).
Training the EMs of those identified as poor readers has a long history. It is based on the premise that the automaticity of oculomotor function can be improved with training by assuming that EMs are under ‘bottom up’ control. In the 1930s (Sisson, 1938), individuals were trained to read according to a set amount of fixations across the text, another group was instructed to read in order to extract meaning from the text and a third was given no instruction. After training, it was found that the first two groups had increased reading rate, though not significantly, compared with the group that had had no specific instruction. The authors concluded that while the reading rate showed an overall increase with training, the conscious effort involved in trying to control the eye movements in the first group may have unduly affected their comprehension of the text as their comprehension had reduced by 12%. Other researchers found that while EMs provided an indication of reading difficulty, they were not the cause but rather the symptom of a word recognition problem (Ekwall & Shanker, 1988 ). In more recent times, a ‘guided reading’ programme has been tested, where a frame is used to guide the reader’s progress (Reichle, Pollatsek, Fisher, & Rayner, 1998). Improvements in eye movements were used as a measure of advancement. That is, fewer fixations of shorter duration and fewer regressive saccades. Another study used a guided reading programme to improve oral reading fluency and reading comprehension abilities in 6th,
7th and 8th grade school children (Marrs & Patrick, 2002). The text was presented at a controlled speed across a computer screen and the speed increased in the hope of improving reading skills. Pre- and post-training measures were found to vary
32 unpredictably across age groups, but those already identified as being reading disabled improved more than their age-matched normal readers. Nevertheless, EM anomalies are found in poor readers (Juhasz & Rayner, 2003) but they are the result, not the cause, of poor reading abilities. EMs have been shown to be largely under top-down control
(Chace et al., 2005). Therefore, functional or bottom-up training methods alone, such as those described above, might be expected to have limited success in improving reading efficiency as was indeed found to be the case.
Other EM findings in relation to text reading include the following: that increased reading speed is related to a decrease in the number of fixations (K. Rayner et al., 2006), but it does not necessarily imply better comprehension (Underwood, Hubbard, et al.,
1990); reading comprehension is related to fixation duration - skilled readers are thought to extract information more efficiently by utilizing fixations of shorter duration and with fewer regressive saccades (Underwood, Hubbard, et al., 1990) and less skilled readers need to back-track and re-fixate more frequently to aid lexical and semantic processing (K. Rayner et al., 2006). In addition, it has also been shown that when visual parameters, such as blur, are changed in the text presentation, there is an effect on reading speed and EM patterns (G. Legge, 2007). Specifically, functional characteristics such as blur and contrast will affect saccade characteristics: direction, number, speed or latency, while cognitive features such as size and spacing will affect the fixation characteristics: number and duration (G. E. Legge, 2007).
While EM patterns for text reading are well established, there are gaps in the knowledge of EM patterns when sight reading music (Madell & Hebert, 2008). For music to be read one has to not only see, but also make sense of, the symbols on the page for them to be converted into a meaningful output by a vocalist or played on a musical
33 instrument; similar to reading words aloud. The use of the word ‘read’ in the context of music tends to imply a similar processing strategy to that of reading text. If music sight reading was simply a matter of pattern recognition, then as a musician becomes proficient at recognizing the patterns, the more adept he/she should become at sight reading and this should be evident in the EM patterns (Kinsler & Carpenter, 1995). In fact, features such as the ability to ‘chunk’ groups of notes into a unit rather than reading individual notes (Furneaux & Land, 1999; Sloboda, 1974; Truitt, Clifton,
Pollatsek, & Rayner, 1997; Wolf, 1976) and the effect of unexpected harmonic structures and increased complexity on fixation patterns (Sloboda, 1977), have been demonstrated in the EMs of musicians. These findings are akin to the patterns of expertise shown in text reading, but there is little research regarding the impact of the fundamental visual features of the score – blur, contrast and size. Nor are there many investigations into how these may be moderated by expertise - even though such parameters are known to have a marked effect on EM patterns when reading text and vary with expertise (G. E. Legge, 2007).
Research that is available for scrutiny regarding sight music expertise is also contradictory in parts. For example, when concluding whether expert sight readers fixate for longer or shorter times, one researcher claims longer (Van Nuys, 1943), one shorter (Young, 1971) and another that they are the same (Schmidt, 1981). Past investigations have not only been limited by the precision of the available technology, but have questionable methodology, incomplete statistical analyses, direct comparison of widely dissimilar skill levels of participants and utilizing only one or two observers
(including the researcher), leading to low statistical power. Most importantly, there is no control over the angular subtense of the note heads (Jacobsen, 1941; Lang, 1961; Van
34 Nuys, 1943; Weaver, 1943; York, 1952). This has long been known to fundamentally affect the performance of EMs (Tinker, 1946) and should be strictly controlled in order to draw any meaningful conclusions. Without this, results from an individual study or comparisons across studies will be confusing and, therefore, frequently contradictory.
Table 1 lists several major EM studies up to 1995. It highlights the dissimilarities in methodology leading to difficulties comparing one with another. (‘X’ denotes that data was not reported.) Systematic control of score parameters is required to clarify the relationship between EMs and sight reading expertise in order to confirm similarities between text and music reading. The level at which expertise is assigned must also demonstrate significant levels of DP (Ericsson et al., 1993) and superior WMC/VSTM
(Curby & Gauthier, 2010; I. Gauthier et al., 2014; D. Hambrick et al., 2014). Perhaps of greater importance in advancing the study of visual processing expertise, is whether music sight reading experience rather than expertise confers any visual processing advantage relative to non-musicians in other domains. That is, can it be shown that non- expert sight readers have better visual processing skills than non-musicians when viewing non-musical targets? If this is the case, there may be benefit in learning to read music so that visual span or speed of processing targets in peripheral vision might be enhanced and transfer into text reading performance.
In order to thoroughly investigate the relationship between EMs and expertise in music sight reading a number of experiments were undertaken to explore differences in EM response to changes in the physical nature of the score and stresses placed on the musicians in terms of speed requirements and disruption to the expected layout of the score. In addition, other tasks a that explored the cognitive nature of sight reading experts - the span of vision, the accuracy in visual search for or processing of letters,
35 words or numbers, the presence of left side bias in processing and even auditory tone discrimination - were compared to the standard of expertise assigned. The results were also compared with those of non-musicians when applicable. The following section outlines the rationale behind these tasks.
36 Table 1: Eye movement research methodologies from Jacobsen, 1941 to Kinsler and Carpenter, 1995.
Name of Preview Viewing Notehead Expert vs # Musical Metronome # Output Researcher(s) Allowed Distance Snellen Size Novice Trials Elements Used Subjects Mode Jacobsen 1941 Varied X X X X Various NO 3 Keyboard
Weaver 1943 YES 30.5 cm X NO X 3 excerpts NO 15 Keyboard
Van Nuys & YES 30.5 cm X X X X NO 12 Keyboard Weaver 1943 York 1952 YES X X YES 11 12 notes NO 120 Vocal
Lang 1961 YES X X YES X 12 short ‘normal’ or 20 Keyboard excerpts as fast as possible Young 1971 X 54.6 cm 6/12 YES 8 X YES 17 Keyboard
Halverson YES 43 cm 6/34 YES 12 2.5 – 4 YES 8 Vocal 1974 bars
Schmidt 1981 X X X YES X X X 12 Wind
Goolsby 1987 YES 48 cm X YES 24 4 bars YES 24 Vocal
Larsen 1990 X 34 cm 6/24 NO X 9 lines NO 9 Silent Reading Goolsby 1994 X 48 cm 6/18 YES X X X 2 Vocal or brass Kinsler & X 57 cm 6/18 NO 32 4 bars YES (mostly) 4 Tapping Carpenter 1995
37
Cognitive Processing
With expertise clearly defined, musicians could then be confidently divided into groups of expert and non-expert sight readers. As such, visual processing of non-musical targets could be investigated for domain transfer of expertise and comparisons made with non- musicians. The following experiments were conducted to tease out aspects of cognitive and visual processing abilities in order to further characterise the music sight reading expert.
Perceptual and Visual Spans
The perceptual span is the amount of visual information that can be processed in a single fixation. This differs from the visual span (which is said to be subject to the physical limitation of the visual system), being able to identify discrete targets beyond central fixation (G. Legge et al., 2007). The perceptual span is contextual (O’Regan, Levy-
Schoen, & Jacobs, 1983), is normally asymmetric and extends 15 characters to the right and 4 characters to the left for languages such as English that are read from left to right
(McConkie & Rayner, 1975; K. Rayner, Well, & Pollatsek, 1980). Learner or non-expert readers have been shown to have a smaller perceptual spans than expert readers: 11 and
14 characters to the right of fixation respectively (K. Rayner, 1986). This difference in size of span is thought to be due to the inability of less skilled readers to utilise parafoveal information as efficiently as a skilled reader (K. Rayner, 1986). Parafoveal information is subject to crowding effects (G. Legge et al., 2007; Levi, 2008; Pelli et al.,
2007) and mirror confusion (S. T. Chung, 2010; S. T. Chung & Legge, 2009).
38
It has been shown that reading speed and visual span are strongly correlated only in the sense that when you use peripheral vision the reading rate drops (G. Legge, 2007; G.
Legge et al., 2007; Yu, Cheung, Legge, & Chung, 2007), rather than the visual span being responsible for differences in reading rates between experts and non-expert readers as it is a physiological or ‘bottom up’ restriction, (G. Legge et al., 2007). The finding that the reading rate is not affected by the perceptual span size, and that this is not likely to be the cause of non-expert readers reading slower than skilled readers (K. Rayner, 1986), seems counter intuitive if the perceptual span is, in fact, larger in expert readers and the visual span is relatively constant between the two.
Traditionally, the visual span has been measured using a moving window method that consequently does not factor EMs into account. A recent study, looking at reading speed in relation to fixation duration when reading a sentence presented as a whole and read conventionally, found that a smaller visual span was related to increased number of fixations of longer duration (Risse, 2014). These results confirmed previous findings (G.
E. Legge, Ahn, Klitz, & Luebker, 1997) and suggest that the visual span can impact the reading speed and that the resultant EM patterns reflect the characteristics of expertise in text reading (Underwood, Hubbard, et al., 1990); the above example being a feature of novice EM patterns. Therefore, an expert would be expected to have a larger visual span.
Musicians and non-musicians, as well as expert and non-expert sight readers, should not show differences in visual span as both groups would be considered experts in the domain of text reading. In the same way, no differences would be expected in perceptual span sizes between the groups.
39
In addition, it was found that faster readers were more accurate at letter identification around central fixation (Risse, 2014, p7). Testing visual and perceptual spans using tachistoscopic presentation of targets around fixation and comparing the results between non-musician and musician groups may reveal other facets of expertise in visual processing.
Tone Frequency and Modulation Discrimination
Frequency Discrimination (FD) is the ability to judge a difference in pitch between two static tones. The lower the threshold of difference between the two tones, the better the
FD. Musicians have long been known to have superior FD skills compared to those of non-musicians (Banai & Yifat, 2011; Deguchia et al., 2012; Micheyl, Delhommeau,
Perrot, & Oxenham, 2006; Oganian & Ahissar, 2012; Spiegel & Watson, 1984).
Frequency Modulation Discrimination (FM) is the ability to detect a change in pitch from a carrier tone over time. This aspect of pitch discrimination has obvious implications in musicianship – from tuning an instrument to being able to play along with others. It's connection with text reading difficulties is less obvious.
However, phonological skills were found to be closely related to reading ability and dyslexia (Talcott et al., 1999). While there is evidence that relationships exist between
FD and FM discrimination capabilities and specific language impairment (SLI), neither poor FD nor poor FM discrimination have yet to be proved predictive of SLI (Kidd,
Shum, Wong, Ho, & Au, 2015). Nevertheless, it is suggested that readers need to be able match the visual input of a word with a pre-learned or remembered sound - a skill that is known as ‘perceptual anchoring’. A deficiency in perceptual anchoring is thought to play
40
a part in the variety of disorders that manifest as reading disability, but is particularly relevant in dyslexia (Ahissar, Lubin, Putter-Katz, & Banai, 2006; Banai & Yifat, 2011,
2012). Recognition of frequency modulations in pitch is considered to be most akin to spoken language and is strongly correlated with dyslexia and reading difficulties (Boets et al., 2011; Talcott et al., 1999). This is because the shape of the sound envelope created by speech sounds more closely resembles a tonal modulation (FM) than the static tone change of FD (Horst, 1989; Talcott et al., 1999). Dyslexics also have deficiencies in the ability to rapidly process sensory information over time – temporal processing. Such deficiencies in processing are linked to FM because it involves change in pitch over time
(Amitay, Ahissar, & Nelken, 2002; Anvari, Trainor, Woodside, & Levy, 2002; Atterbury,
1983). The temporal aspect of FM might imply that the dorsal processing stream is involved. This has led to the proposal that these two forms of auditory perception can be likened to vision - where FD corresponds to central visual acuity/ventral stream processing and FM to peripheral vision/dorsal stream processing (Demany & Semal,
1989).
Conversely, it has been suggested that FM may simply be a less accurate measure of FD because low frequency modulations may be processed as a FD judgement by pitch sampling at the outer limits of the range (Demany & Semal, 1989; Horst, 1989). If FD and FM are basically the same process, then the performance of musicians should be significantly better than non-musicians in FM, as is the case with FD. However, a ceiling effect may be observed when comparing expert and non-expert sight readers because both groups already have significantly better FD. If FM does not involve FD at low
41
frequency modulations, then a difference in FM between musicians and non-musicians might not be expected.
It is of considerable interest that FD can be improved with training and that improvement transfers to tones other than the training tone (Grimault, Micheyl, Carlyon, Bacon, &
Collet, 2003; Micheyl et al., 2006). Should a significant relationship between expertise in music sight-reading and either FD or FM skills be found it is correlational and not causal, but may bear upon aspects of future research into text reading remediation.
Rapid Automatized Naming
Rapid Automatized Naming, RAN, is measured by naming objects, colours, letters or numbers out loud as fast as possible (Lervag & Hulme, 2009) and is strongly linked with reading difficulties, particularly dyslexia (Stainthorp, Stuart, Powell, Quinlan, &
Garwood, 2010). The basic mechanism is thought to be the extended pause time required to process visual difference in or features of targets in order to identify them and it does not appear to be generalised slowness of processing as it does not occur in the auditory domain (Stainthorp et al., 2010).
RAN has been shown to correspond with other perceptual processing abilities that are known to impact upon holistic processing. In dyslexic children and adults, RAN has been related to deficits in Working Memory Capacity, Visual Working Memory and attention
(Beidas, Khateb, & Breznitz, 2013; Nergard-Nilssen & Hulme, 2014; Poll et al., 2013).
These also feature prominently as skills that are advanced in those with visual expertise in the musical domain. As such, correlations might be expected to exist between RAN and expert music sight readers.
42
Testing RAN in adults with text reading skills that are normal or above would expect to yield ceiling effects. This is because RAN is traditionally tested using individual numbers, shapes or colours. A list of basic words, the order of which is randomised repeatedly and then read out loud as fast as possible, might also be expected to demonstrate a ceiling effect. However, if expertise is characterised by superior WMC and the ability to ‘chunk’, as discussed earlier, and lower WMC is characteristic of readers with slower RAN, then expert music sight readers may perform faster in a rapid word identification task.
Visual Search
Simple visual search is the identification of a target among distractor targets and is thought to be the result of a low-level, feature ‘pop out’ effect (Woods et al., 2013). If the target features are very specific or recognisable, the search speed will increase (Redford,
Green, Geer, Humphrey, & Thiede, 2011). Conversely, search speed will decline with increased distractors. However, performance of a serial search – searching for a specific target among other visually similar targets for example - is thought to involve higher level functioning that involves spatial attention skills as well as working memory and reaches a ceiling effect in adults (Woods et al., 2013).
Recent research has compared music and language processing in terms of Syntactic
Working Memory and whether or not audition can interfere with language processing
(Fiveash & Pammer, 2014). It was found that musicians were distracted by music while performing a language task, particularly for expert musicians when the music comprised unexpected harmonic structures. The effect was less for non-experts and negligible for
43
non-musicians. It has also been shown that expertise in music results in a reduced crowding effect for musical targets (Wong & Gauthier, 2012). This would suggest that expertise in music sight reading might lead to a more efficient serial search of musical targets relative to non-experts. Furthermore, this effect show may respond to the introduction of a musical stimulus that may be correspond to the search target or be dissimilar to it – a group of congruent or incongruent tones respectively.
Comparing musicians with non-musicians in a visual search task involving musical targets presents an obvious challenge, but these difficulties would be largely circumvented if the targets employed are simplified and can be processed by non- musicians as a pattern matching task. A music-neutral pattern matching task would need to be performed in order to establish baseline search capabilities of the groups. This would yield valuable information about the differences that musicianship and/or musical expertise may bring to bear on visual search processing, and therefore, on music sight reading and its possible impact on visual processing in general.
Thesis Aims
The aim of this thesis is to investigate unknown characteristics and unresolved traits of piano music sight reading expertise. It is generally accepted that sight reading expertise is reflected in EM patterns that are similar to those of expert text readers. It is also linked to certain innate abilities as well as long periods of deliberate practice. However, little is known about how EM patterns are affected by changes in the visual parameters of the score (other than size), or if expertise might influence such responses. In addition, little is known about the extent to which sight reading expertise might transfer into related or
44
non-related perceptual domains and how non-expert sight readers then compare with non- musicians - a new area of investigation. Therefore, the thesis has three primary aims:
1. To clarify an objective method of assigning sight reading.
2. Extend the current knowledge of expert piano sight reading performance in
relation to EM patterns while manipulating visual features of the score: blur,
contrast, size, the removal of bar lines, visual alteration of music score and
imposing temporal restriction using metronome.
3. Explore the transfer of music sight reading expertise to other perceptual domains,
both musical and non-musical, including comparison with non-musicians.
Thesis Structure
This thesis comprises a series of studies that initially develops a picture of the EM patterns of the expert and non-expert piano music sight reader. This is followed by a series of studies designed to examine specific features of perceptual processing and how sight reading expertise might confer cognitive benefit when visually processing non- musical information.
The ensuing chapter, Chapter 2, outlines the general methodology for the EM studies throughout this thesis. Any deviations from these procedures will be specifically addressed as required.
Chapter 3 addresses the first thesis aim by providing a detailed exploration of the validity of the sight-reading expertise definition used in this thesis - the ability to perform at 6th
Grade AMEB sight reading standard; thereby addressing the first thesis aim.
45
Chapter 4 explores the second thesis aim by investigating expertise and EM patterns by examining the effect of blur, contrast and size changes; removing bar lines, changing the spacing and note beam directions and the imposition of time keeping using the metronome on performance.
Chapter 5 explores cognitive processing traits of expert piano sight readers relative to non-experts and non-musicians: their visual and perceptual spans, word reading speed, visual search of a non-musical target and visual search using music-like targets with congruent or non-congruent simultaneous auditory interference. This chapter addresses the third aim of the thesis.
Chapter 6 discusses the features of music sight reading expertise uncovered by the EM and cognitive processing experiments reported in this thesis, and concludes with recommendations for future research.
46
CHAPTER 2 Methods
Eye movement studies
Participants
Following the granting of ethics approval from the UNSW Human Research Ethics
Advisory Committee to perform the study, participants were recruited from within the
UNSW student body and reimbursed for their time. Consideration for inclusion in the study was based upon an individual self-selecting their ability to perform, on a keyboard, a sample sight reading melody as it appeared on the recruitment poster (See Appendix 1).
Each observer wore full visual correction and was capable of N5 resolution at a viewing distance of 60cm.
On occasion, sample sizes were as small as 8 subject in a group. This was, nevertheless, considered adequate as previous EM studies have had even fewer subjects. Specifically, the frequently cited seminal work of Kinsler and Carpenter had 4 study participants
(Kinsler & Carpenter, 1995). This is because such studies as Kinsler and Carpenter, 1995 and the present study, have methodology designed to demonstrate either the presence or absence of an effect (Anderson & Vingrys, 2001). According to Anderson and Vingrys,
2001, a sample size of even 5 participants is acceptable under these conditions. Even so, large variations in the abilities of the non-expert group were to be expected and this might play a part in weakening some of the findings. Nevertheless, any insignificant trends found would no doubt have yielded more conclusive results with more study participants. Due to time constraints, this was unfortunately not possible.
47
Stimulus
One hundred and twenty, 4-bar individual melodies were composed (see Figure 1a). Each melody was written in the treble clef, to be played by the right hand and limited to white notes only. The range was from Middle C to upper F: C4 to F5. Identical rhythmic components were used for each, in largely non-identical combinations and differing melodic content. Care was taken to include melodic patterns that did not always have an obvious tonal centre or with easily recognised melodic patterns. This was done in order to encourage playing according to what was seen rather than be guided by ear. A single line of music on one stave only was employed. This has been found to be sufficient to elicit an expertise response as using more complicated music confounds the results by introducing performance differences rather than measuring the fundamental characteristics of a musician’s EMs (Wong & Gauthier, 2009).
The stimulus viewing distance was calculated by questioning four pianists as to the approximate viewing distance of music when playing a standard upright piano. 60cm was chosen: the range was from 30 to 60cms, with 3 values between 50 and 65cm.
Figure 1a.
Figure 1b.
48
Figure 1. A sample musical excerpt illustrating 0.0 Gaussian Blur (1a) and 3.2 Gaussian Blur (1b).
Procedure
EM data was collected using the Arrington Research ‘ViewPoint’ USB220 eye tracker, the sampling rate being 220 times/second. The images were generated using a custom written programme for MATLAB and presented on a linearized 27-inch Mitsubishi
Diamond Pro monitor driven at a frame rate of 80Hz. The tracker was driven by a
Hewlett Packard ‘Elitebook 8470p’ PC (Intel Core i5 2.60GHz processor/8.00GB
RAM/16-bit Operating System).
The apparatus consisted of a single infrared camera mounted on a chin and headrest assembly that was mounted on an instrument table. The table was set so that the viewing distance to the screen was 60cm (see Figure 2). The participant’s height was carefully aligned using a canthus mark that was level with the centre of the computer screen. The camera was then calibrated according to the manufacturer’s instructions. Once calibration was successfully performed, a practice session was performed in order for the participant to become familiar and comfortable with the testing process: 4 seconds after a tone sounded, the music stimulus would appear on the computer screen. Participants were instructed to start playing the piece as soon as it appeared on the screen, as quickly and as accurately as possible, without looking down at the hand, without pre-reading and without stopping regardless of errors. After the participant finished playing, a visual noise patch was presented on the screen. The participant was instructed to fixate on it to eliminate any afterimages that may have been generated by the test stimulus. Sufficient
49
time was given to re-orientate the hands into position by touch between presentations.
After 6 trials, the full procedure was undertaken, following the same procedure.
Images were blurred using MATLAB (Version 2014b, image processing toolboxTM) by filtering the original image with a Gaussian filter. 4 levels of blur were created by setting the standard deviation of the filter to 0.0, 0.8, 3.2 and 12.8 degrees. Contrast levels of the filtered images were normalised before presentation. Utilizing digitally induced blur as opposed to refractive blur was chosen as it should not affect the interactions that we are examining (Akutsu, Bedell, & Patel, 2000). This approach to blurring had been used in previous psychophysical investigations (McAnany, Shahidi, Applegate, Zelkha, &
Alexander, 2011) and the abovementioned blur levels were duplicated for this project.
A pilot study revealed that the responses to blurred stimuli at the 10% contrast level were mostly indecipherable and therefore omitted from the analysis. Text reading studies have found that contrast is only a factor text below the 10% level (G. E. Legge, 2007), thus confirming a similarity between text and music reading. The remaining levels of reduced contrast, 20% and 40%, were individually tested in addition to 100% contrast ensure that reduced contrast music stimuli, even when blurred, induce similar EM patterns to those related to text reading under the same conditions. For the final analysis, 2 levels of blur were tested by setting the standard deviation of the filter to 0.0 (Figure 1a) and 3.2 degrees (Figure 1b). The 0.8 level blur was omitted as it was barely different from 0.0 level as was the 12.8 level as it was too difficult to see when combined with reduced contrast.
50
Figure 2: Experimental set-up showing piano keyboard, computer display and eye tracker camera.
The melodies were presented in a size equivalent to N5 and N10 print viewed at 60 cm.
N5 was chosen to approximate the size of print deemed to require ‘normal’ vision in
51
Optometric terms with N10 being ‘double’ N5 to test effects due to stimulus size. Targets smaller than N5 or larger than N10 were not examined. In practical terms, the N5 size operates towards the lower limit of comfortable, sustainable vision and larger than N10 beyond the sizes normally used in printed music in adult scores. Therefore, only these two sizes were used.
The final testing battery consisted of 27 different melodies in N10 and the same 27 melodies in the same order in the N5 size. Each combination of blur and contrast was presented 3 times at each size to each participant. The order was assigned as follows: melodies 1, 2 and 3 became the ‘N10: No Blur: 100% Contrast’ cell, 4, 5 and 6 the ‘N10:
No Blur: 40% Contrast’ cell and 7, 8 and 9 the ‘N10: No Blur: 20% Contrast’ cell. This pattern was repeated at N10 for the 0.8 level blur with 100%, 40% and 20% contrast and similarly for 3.2 level blur with 100%, 40% and 20% contrast. The same blur and contrast treatments were then allocated to the same 27 melodies in the same order for the N5 size so that melodies 28, 29 and 30 became the ‘N5:No Blur: 100% Contrast’ cell - melodies
1, 2 and 3 at half the size. Rather than randomize the presentation order for each participant, the list of 54 melodies was randomized and checked to ensure that there were no identical parameters in consecutive trials.
This list became the standard order of stimuli so that identical parameters were presented for a given melody to each participant in exactly the same order. This not only simplified analysis but ensured that outcomes were not confounded by differences in melodic and rhythmic content. The 3 trials in each condition were then averaged to provide the measure of observer performance for that condition. Preliminary analysis of the data
52
found that the results for the 20% and 40% contrast levels were not significantly different from the 100% level, as is the case for text reading (G. E. Legge, 2007). These were then grouped together for analysis – each blur and size level thereby having 9 trials for each participant.
In order to ensure that only score reading EMs were being measured, a specific portion of the data set was selected for analysis. For example, variations in the time taken to start playing after stimulus onset or differences in the cessation of reading EMs towards the end of the piece all needed to be eliminated. Therefore, the time that playing commenced,
T1, through to the time that playing ceased at the end of bar 3, T2 was used as the sound window for analysis.
The location of T1 and T2 was determined using FleximusicTM Audio Editor. The sound files were imported and the points on the wave file for T1 and T2 were determined by first filtering for noise and then manually marking the location of T1 and T2. This process was found to be repeatable to within 0.05 second. Once T1 and T2 were known in relation to the length of the sound file, it was then possible to calculate the number of samples between points T1 and T2. Therefore, EM parameters calculated between T1 and
T2 pertain only to the time period of interest, that is, when the music was being read.
Total Time, Fixation Number and Duration plus Saccade Number, Speed, Direction and
Latency were measured and analyzed using a purpose-written MatlabTM code.
Details of alteration to this basic method for the Spacing and Metronome EM studies are outlined in the relevant study sections.
53
CHAPTER 3 Sight Reading Expertise
Introduction
Examining the characteristics of visual expertise warrants that those categorised as experts are, in fact, experts in that domain. Research into sight reading expertise has often been based on a self-reporting paradigm and, as such, can be inherently subject to bias.
For example, of the major EM studies outlined in Table 1, expertise was assigned according to a teacher’s estimation (Halverson, 1974) and another if ‘considered’ skilled or otherwise (Goolsby, 1987); or detailed classification was undertaken according to a sight reading assessment, but the musical examples used were inadequately defined as
‘medium’ or ‘difficult’ (Young, 1971). The administration of a test that could reliably differentiate expertise would not only be more efficient, but would strengthen the validity of any subsequent experimentation.
Previous research has estimated sight reading expertise to be present at Grade 6 (Waters,
Townsend, & Underwood, 1998). In this study, the speed of detecting same/different features in two, 4-note presentations of 800ms duration was the criteria for assigning sight reading expertise. Another study allocated expertise on the basis of being 90% correct when playing a ‘difficult’ passage as assessed by six music faculty professors at
Indiana University (Young, 1971). However, the Waters et al. study might not be considered an actual measure of sight reading skill but rather a measure of domain relevant pattern matching as no physical output or playing of the music was involved.
Young’s study did not define ‘difficult’ by an independent measure and, as such, cannot be accurately replicated for the purposes of the current study.
54
A sight reading task capable of defining expertise would need to be of sufficient difficultly to elicit the ‘chunking’ characteristics from experts, yet remain achievable for the non-expert (Farrington-Darby & Wilson, 2006). Such a test could be developed using the key measure of expertise - the speed at which a task is performed (Bilalic, Langner,
Ulrich, & Grodd, 2011; K.A. Ericsson, Krampe, & Tesch-Romer, 1993; K. A. Ericsson,
Roring, & Nandagopal, 2007; Farrington-Darby & Wilson, 2006; I. Gauthier & Bukach,
2007). Therefore, the time taken to complete that task should delineate expertise. As 6th grade was the only independent, replicable level of musical attainment that the researcher was aware of from the literature, in consultation with a university academic of considerable expertise in piano pedagogy, 6th grade was selected as a realistic standard from which to examine expertise. Piano was chosen specifically for investigation in this thesis as the investigator has extensive experience with this instrument (8th Grade AMEB and an undergraduate degree in music performance) and capable of making judgements in regard to assessing 6th Grade sight reading.
Taking 6th Grade AMEB sight reading as the standard for expertise, known features of music sight reading expertise were explored to test whether this standard is significantly related to aspects of expertise reported in the literature, specifically, expended DP and speed of performance. Study 1 analysed the results of a survey taken by pianists that investigated aspects of their musical education and experiences. The main purpose of the survey was to probe the nature of expertise and deliberate practice. Study 2 tested the premise that speed is the hallmark of the expert by timing the performance of pieces as outlined in Chapter 2. Study 3 examined the feature of WMC as another hallmark of processing expertise. In each of the three studies, the results of the expert sight reading
55
group - as defined by the 6th Grade standard – were compared with those of the non- expert group. The hypothesis was that a pianist capable of performing a 6th Grade AMEB piano sight reading assessment task at, or near to perfection, can be confidently classified as having visual expertise. As such, this level could then be used as the criteria for the grouping pianists for the assessing of visual processing expertise in this thesis.
Study 1: The Survey
Surveys have been used to study various aspects of musicians’ general musical engagement and emotional responses to music (Chin & Rickard, 2010; Werner, Swope,
& Heide, 2006). The purpose of the current survey was to draw out specific aspects of musical engagement from the study participants. The topic areas, described in detail below, specifically probe the nature and extent of a musicians’ DP. That is, can it be shown that those participants described as expert music sight readers according to the 6th
Grade criteria, conform to Ericsson’s assertions of 10 years/10,000 hours of DP to become experts in a field (Ericsson et al., 1993). Are there other aspects of the sight reading experts’ musical experience that might also be significant to their development?
The survey consisted of 11 questions of original content and specific to the goal of eliciting information about the participants’ musical experiences within the framework of the thesis. Questions 1-5: aspects of the individuals’ formal music instruction - 5 response options: Questions 6 and 8: instrumental variety and playing experiences - Yes or No response required: Questions 9: second language - Yes or No response required and
Questions 10-11: current playing and sight reading frequency - 4 response options.
56
Each of these areas was then evaluated according to the level of expertise as defined in this thesis. The following section details the rationale behind the survey content.
Age at which formal training commenced and years of formal training
The age of commencement of formal music training has been found to positively correlate to sight reading expertise in past studies (K.A. Ericsson & Lehmann, 1996;
Gudmundsdottir, 2010; Kopiez & Lee, 2008). It may well be that the brighter, more highly motivated children with the opportunity to do so will start music lessons early
(Corrigall & Trainor, 2011). Therefore, any correlations found are by no means causatory and as expertise may be found in some who started later.
Grade of Practical and Theory Achieved
This is not necessarily the same as the total time of involvement in formal music education as some students are capable of attaining two grade levels in a year while others, only one. The inclusion of these categories was to tease out the influence of the effect of the depth of musical understanding independent of the time of engagement in these activities and whether there is a difference between the theoretical and practical/performance aspect of musical engagement. There is evidence that a higher level of theoretical musical understanding is associated with better sight reading in wind players (Elliott, 1982).
Training beyond Grades
This category was included to further clarify if the level of exposure to the form at a more complex level is a factor that determines expertise in sight reading.
57
Playing other Instruments, ensemble experience and improvisation
These categories have shown positive correlations in other studies, but the comparisons occur over different instruments, including voice, and once again suffer from lack of consistency in defining the expert (Daniels, 1986; Gudmundsdottir, 2010; Kopiez & Lee,
2008; Mishra, 2013; J. Mishra, 2014; A. J. Waters, Townsend, E., & Underwood, G. ,
1998; Woody, 2012).
Second language
It has long been known that speech and language share similar auditory processing structures (McMullen & Saffran, 2004; Moreno & Besson, 2006; Rogalsky, Rong,
Saberi, & Hickok, 2011) and of interest in this category is the notion of ‘far transfer’ - involving the effect of one skill on the performance or acquisition of another unrelated skill (F. H. Rauscher & Hinton, 2011).
Current frequency of playing and of sight reading
Of interest here is the ‘use it or lose it’ premise. Is it the case that, once acquired, sight reading skills are there for life at a particular level or does one get rusty without playing generally and/or without sight reading specifically?
Hypothesis
That expert music sight readers will show that DP is a significant feature of their musical experience.
58
Method
Subjects were recruited from an Australian university student body for EM studies in relation to musicians and non-musicians. Each musician participant completed a survey as part of the data collection for the main EM study (see Appendix 1) after completing the EM tasks. Categorization into the expert or non-expert group was based on whether he/she could perform a 6th Grade AMEB sight reading examination task at or near perfection on the piano as judged by the researcher. This criterion was found to delineate expertise on the basis of EM patterns as determined in the previous chapter. The two expertise groups were then compared across each of the survey criterion. 16 participants were designated expert and 18 as non-expert music sight readers.
Results
All findings in this thesis more than two standard deviations from the group’s mean were considered to be outliers and not included in the analyses.
The results were regrouped for ease of analysis. The answers to questions 8 and 9 were reduced to Yes = 1 and No = 0 for improvisation and 2nd language skills respectively.
Questions 10 and 11 were treated in a similar fashion - more than weekly = 1 and less than monthly = 0, for playing and sight reading frequency respectively. Question 4 had option 1 changed to represent no formal training above grades for detailed comparison with questions 1, 2, 3 and 5 and to Yes = 1 and No = 0 for comparing with questions 6 –
11. The categories in Question 1 were reversed to make correlations more intuitive.
59
Category p Value Experts Non-Experts
(n=16) (n=25)
Years of Formal Training p< 0.0001*** >10 Years 5-8 Years
Practical Grade Achieved p = 0.0001*** >AMUS Grades 4-6
Theory Grade Achieved p = 0.0009*** Grades 4-6 Grades 2-4
Current Frequency of Sight Reading p = 0.02* 36% 4% (> Weekly/< Monthly) >Weekly >Weekly
Training Beyond Grades p = 0.03* 43 % 12 %
Age Commenced Formal Training p = 0.04* 5-7 Years 8-10 Years
Ensemble Experience (Yes/No) p = 0.06 93 % Yes 64 % Yes
Play other Musical Instruments p = 0.19 79 % Yes 56 % Yes (Yes/No)
Improvisation (Yes/No) p = 0.51 71 % Yes 60% Yes
Read/Write another Language p = 0.73 57 % Yes 68 % Yes (Yes/No)
Current Frequency of Playing p > 0.9999 64% 64% (> Weekly/< Monthly) >Weekly >Weekly
Table 2: p values by category according to expertise with median, percentage and descriptors.
60
A non-parametric Mann-Whitney test was performed in each survey category in relation to sight reading expertise. The results appear below in Table 2. Correlations between survey categories using a 2-tailed Spearman’s Rho test was performed separately for the expert and non-expert sight reader groups and the results appear below in Table 3.
Age at which formal music training started was significantly earlier for experts (Mdn = 5-
7 years) was significantly greater than non-experts (Mdn = 8-10 years), U = 109, p =
0.04.
Years of formal music training for experts (Mdn = >10years) was significantly greater than non-experts (Mdn = 5-8 years), U = 46, p < 0.0001.
Practical Grade attained was significantly higher for experts (Mdn = >AMUS) than for non-experts (Mdn = 6-8 Grade), U = 56, p = 0.0001.
Theory Grade attained was significantly higher for experts (Mdn = 4-6 Grade) than for non-experts (Mdn = 2-4 Grade), U = 70, p = 0.0009.
Training beyond AMEB grade was significantly higher for experts (Mdn = Performance
Cert.) than non-experts (Mdn = Performance Cert.), U = 121, p = 0.03.
Playing a second instrument was not significantly different between expert and non- expert sight readers. For both experts and non-experts, the majority can play a second instrument, U = 135.5, p = 0.19.
Playing in an ensemble was also not significantly different between expert and non- expert sight readers: both experts and non-experts have ensemble experience, U = 124.5, p = 0.06.
61
Improvisation experience was also not significant between experts and non-experts: both group improvise, U = 155, p = 0.51.
Speaking a second language was not significant: the majority of experts and non-experts both having a second language, U = 156, p = 0.73.
Both experts and non-experts play more than once every week: U = 174.5, p > 0.9999.
Significantly more experts sight read more than once every week than non-experts: U =
119.5, p = 0.02.
From Table 2, expert music sight readers, when described as those who achieved successful 6th Grade sight reading performance, were likely to have had formal music training for over 10 years and commenced before the age of 7. While having achieved moderate to high grades in theory, they were accomplished performers to AMUS level with 42% of this group having a further advanced performance qualification. The expert sight reader did not play more often than the non-expert, but does sight read more often.
He/she was slightly less likely to be bi-lingual, but more likely to improvise and be a multi-instrumentalist who has played in an ensemble when compared with his/her non- expert counterpart.
Spearman’s Rho analysis of within group correlations (Tables 3a and 3b) revealed that the non-experts showed no strong correlations between any of the survey variables other than the age of commencement of training which correlated with the years of training and the grades of practical and theory correlated significantly with each other. The expert
62
sight readers’ group, on the other hand, were not significant for within group correlations for these variables.
The most outstanding finding was the significance of ensemble experience within the expert sight reader group. Ensemble experience was significantly less likely for those with training beyond regular AMEB grades. However, those with ensemble experience had a significant, positive correlation with improvisation and 2nd language ability, but a significant, negative correlation with current playing and sight reading frequency. In addition, a strong correlation was found between those expert sight readers who currently had a playing frequency of more than once per week and those who also had a current sight reading frequency of more than once each week.
63
Table 3a: Spearman’s Rho values for survey variables - experts sight readers. Significant values are highlighted.
Spearman's Age Years of Prac Theory Beyond Other Ensemble 2nd Playing SR Improv. Rho Started Training Grade Grade Grades Instrument Experience Language Frequency Frequency
Age Started Years of 0.550 Training Prac 0.117 0.471 Grade Theory 0.196 0.479 0.042 Grade Beyond -0.072 0.028 0.281 0.003 Grades Other 0.791 0.671 0.894 0.103 0.101 Instrument Ensemble 0.667 0.671 0.894 0.103 -0.320* 0.531 Experience
Improv. 0.667 0.671 0.894 0.103 0.228 0.440 0.439*
2nd 0.633 0.671 -0.866 0.103 -0.125 0.251 0.320* 0.091 Language Playing 0.949 0.707 0.894 0.224 -0.043 0.026 -0.372* 0.141 -0.258 Frequency SR 0.335 0.707 0.984 0.224 -0.043 0.026 -0.372* 0.141 -0.258 1.000* Frequency
64
Table 3b: Spearman’s Rho values for survey variables – non-experts sight readers. Significant values are highlighted.
Sperman's Age Years of Prac Theory Tertiary Other Ensemble 2nd Playing SR Improv. Rho Started Training Grade Grade Training Instrument Experience Language Frequency Frequency
Age Started Years of Training 0.475* Prac Grade 0.069 0.617* Theory Grade 0.151 0.412* 0.505* Tertiary Training 0.130 0.370 0.368 0.213 Other Instrument -0.100 0.000 -0.100 -0.821 0.079 Ensemble Experience -0.051 0.051 0.200 -0.900 0.021 0.175 Improv. 0.205 0.300 0.205 -0.667 0.302 -0.066 -0.102 2nd Language 0.400 0.205 0.400 -0.700 -0.274 -0.263 -0.157 -0.210 Playing Frequency 0.205 0.103 -0.205 -0.791 0.236 0.161 0.042 -0.238 -0.021 SR Frequency 0.000 0.354 0.258 0.000 -0.075 -0.230 -0.272 -0.250 0.140 0.272
65
Discussion
When interpreting these data, one must keep in mind that the results pertain to pianists only. Therefore, care needs to be taken when drawing conclusions. For example, it cannot be concluded from this data that a tertiary music student is an expert sight reader. Only one of the cohort of 34 was enrolled in an undergraduate music programme and this participant was categorised as an expert sight reader. However, another had recently completed a university music degree and was a non-expert sight reader. Consequently, conclusions drawn from studies that use enrolment in a tertiary music programme as the definition of sight reading expertise may be misleading. Those that had training beyond
AMEB grades in this study, apart from these two university musicians, had performance- based certificates and diplomas. Therefore, a more appropriate description of the expert sight reader in this context is that they were more likely to have advanced performance qualifications rather than be a tertiary music graduate student.
Also, the solo nature of the piano would likely account for the significant negative correlation between training beyond the AMEB grades and ensemble experience.
Another instrument, for example the violin, might be expected to yield a different relationship between these two variables.
Starting formal training early in life is significant for expertise in sight reading: p = 0.001 and agrees with several other studies (K.A. Ericsson & Lehmann, 1996;
Gudmundsdottir, 2010; Kopiez & Lee, 2008). It still needs to be kept in mind that these individuals may be simply more intelligent and highly motivated individuals with the opportunity to do so (Corrigall & Trainor, 2011) or it may be that they showed musical
63
talent at an early age and this was nurtured from this early age (Howe et al., 1998;
Lombardo & Deaner, 2014; Wai, 2014). Table 3a shows that early commencement of formal training is more positively correlated with improvisation in the expert group than in the non-expert sight reading group: r = 0.667 and r = 0.205 respectively. It may be that the natural musical ‘talent’ manifests at a young age as the ability to create music or to improvise. The child is then directed into formal music training – the nature of which is to faithfully reproduce someone else’s work – where the ability to sight read is a valuable tool for success. If innate features are involved in the development of sight reading expertise, recent research has shown that it is not due to a difference in brain structure per se (A. Norton et al., 2005). This study found that the brains of children with the drive to learn music were not different the structure of brains of children with the drive to learn music when compared with those who do not - despite findings of structural differences in adult musicians’ brains. Therefore, the argument that some have a predisposition to excel in certain aspects of musical endeavour due to a genetically inherited brain structure seems unlikely. This does not exclude functional traits that might be inherited and predispose some to visual processing expertise, such as working memory capacity.
Improvisation was not found to be significant in relation to sight reading expertise: p =
0.51. This result is contrary to other findings (Daniels, 1986; Gudmundsdottir, 2010;
Kopiez & Lee, 2008; Mishra, 2013; J. Mishra, 2014; A. J. Waters, Townsend, E., &
Underwood, G. , 1998; Woody, 2012). The most likely explanation is that the participants were all piano players. In the context of western art musical tradition, the piano is mainly a solo instrument. True reproduction of established works is the goal of performance, not improvisation. However, it is interesting to ponder the connection between early
64
improvisational talents as being the impetus for the child to be nurtured in formal music training in the first place. The positive correlation of early commencement of training and improvisation: r = 0.667 might suggest this. Expert sight readers in this study were highly correlated with great achievement in performance in this tradition and this level of achievement involves exhaustive amounts of commitment and practice, probably to the exclusion of other forms of musical expression.
Current sight reading frequency was not significant for expertise, but expert sight readers do engage in sight reading activities more frequently: p = 0.02. This does not necessarily imply that continuing to sight read maintains these skills as it could be a manifestation of the corollary - those who are good at the skill will continue to apply it in their musical practise. However, it seems more likely that when the study participants do play, they were likely to sight read. Table 3a shows that playing frequency and sight reading frequency are perfectly correlated for expert sight readers: r = 1.
A more detailed examination of the relationships between ensemble playing and improvisation needs to be undertaken before definitive claims can be made about the nature of expert music sight readers. For example, to claim that expertise in music sight reading is simply expert pattern recognition (Wolf, 1976) denies the strong positive relationships that were found between the age that training commenced, the years of formal training and the practical grade achieved for expert sight readers with improvisation. Of interest is also the meaning of a stronger relationship between 2nd language ability and the younger commencement and years of training in the experts.
Teasing out what prompted the commencement of music lessons in the first place,
65
whether from observed ‘natural talent’ or perceived social or educational benefit, needs to be identified.
Does training beyond grades to AMus and above for expert sight readers stifle the love of ensemble playing and improvising or playing another instrument or is it just that there is no time for such pursuits? Are piano players playing the piano in their ensembles or playing other instruments and are they improvising on these instruments perhaps because their formal training in piano might have stifled the creativity required to do so? This trend is suggested from Tables 3a and 3b, where playing other instruments is much more strongly correlated with improvisation and ensemble experience in the expert group than in the non-expert group.
Fluency in another language does not appear to be significantly related to sight reading expertise: p=0.73. Considering approximately 2/3 of both groups were bilingual this aspect of visual processing can be said to be irrelevant in this context. However, 2nd language ability has a significant positive correlation with ensemble experience for experts while it is a negative, non-significant correlation for the non-expert sight readers.
The question remains whether this might be the manifestation of a more cultural relationship and beyond the scope of this thesis.
The high significance of the level of theoretical grade attained: p=0.0009 could be surprising when considering that sight reading ultimately involves a performance output.
However, an ‘expert’ is characterised by the ability to ‘chunk’ elements of their skills into smaller units for efficient processing (Ashby et al., 2005; Gobet et al., 2001; Heller,
1982; E. Kowler, 2011; G. E. Legge, 2007; Meseguer et al., 2002; K. Rayner, 1998; K.
66
Rayner et al., 2006; Truitt et al., 1997; Underwood, Hubbard, et al., 1990). Expert music sight readers have been shown to exhibit these characteristics (Furneaux & Land, 1999;
Goolsby, 1987; V. Kinsler & R. Carpenter, 1995; Schmidt, 1981; Sloboda, 1974, 1977;
Truitt et al., 1997; Wolf, 1976; Wurtz et al., 2009). The consequence of this is that musicians’ EM patterns are known to be disrupted when unusual or unexpected rhythmic or harmonic structures are encountered (Sloboda, 1977; Wurtz et al., 2009). The EM results from Chapter 4 of this thesis bear witness to the pattern recognition and chunking skills of the expert music sight readers as defined in this thesis. Therefore, extensive knowledge of the ‘rules’ of western art music would be a logical prerequisite for efficient visual processing.
The survey results are consistent with the claim that expertise is the result of DP
(Ericsson et al., 1993). Certainly the majority of the significant findings relate to DP rather than innate abilities – age of commencement of training, performance and theory grades achieved and years of formal training. This lends weight to the summary from
Mishra that “sight-reading is a teachable activity rather than a stable characteristic and that sight-reading is a skill that improves with the musicality of the performer” (J.
Mishra, 2014). However, the types of questions asked were to explore aspects of DP and no information regarding early giftedness or further evidence of innate abilities was requested. Nevertheless, the aforementioned positive correlation between improvisation and early age of training is worthy of further investigation.
Limitations to the purely DP approach to expertise is evidenced by the knowledge that, although many try, many fail to achieve sight reading expertise (Campitelli & Gobet,
67
2011; Hambrick et al., 2014). This is thought to be limited by the individual’s innate cognitive abilities, particularly WMC (Meinz & Hambrick, 2010). Nevertheless, the exact mechanisms by which such expertise is acquired are yet to be discovered.
The Survey showed that high levels of DP correlate with the expert music sight readers, as defined by this thesis. The following study, Study 2, investigates speed of performance and its relationship with expertise.
Conclusions
The survey results agree with other studies regarding the place of DP in the acquisition of expertise while recognising that the means of such skill acquisition remains elusive
(Elliott, 1982; Meinz & Hambrick, 2010). As such, this first feature of expertise from the sight reading literature was correlated the 6th Grade level of assessment. However, questions arise as to the exact nature of the impetus to start early music training and nature of the piano as a primarily solo instrument in a formal training context may impact upon other conclusions that can be drawn from these results. The follow study investigates speed of task performance as a feature of visual processing expertise.
Study 2: Speed of Performance
Hypothesis
That expert music sight readers will perform at significantly faster speeds than non– expert music sight readers.
68
Methods
19 participants were seated at a piano keyboard facing a computer screen at a distance of
60 cm. 54 x 4 bar musical excerpts were individually displayed on the screen. The participants were instructed to play each piece as quickly and as accurately as possible – see Chapter 2 ‘methods’ for details. At the completion of the study, each participant was asked to play a 6th Grade piano sight reading assessment task as a first-pass performance and was ranked by the researcher as an expert (n=8) or non-expert (n=10) sight reader.
Results
Total time was calculated from the EM recordings for each condition and the results analyzed using Graphpad ‘Prism’TM. Two-way ANOVA were performed to determine if specific effects existed regarding performance time for blur and size between expert
(within subjects’ factors) and non-expert music sight readers (between subjects’ factors) and the results were summarized in Figure 3. The Average Total Time when performing from T1 to T2 was plotted as function of blur grouped for size. T1 was located at the beginning of Bar 1 and T2 at the end of Bar 3. Expert sight readers and Non-Expert sight readers are represented in each figure by the circle and square respectively. The error bars signify one standard error of the mean (SEM).
The expert sight readers performed significantly faster overall: F (1, 31) = 12.46, p =
0.001. No significant interaction was found between expertise and size: F (1, 31) =
0.2247, p = 0.64. A larger within-group variation was found for the non-experts than for the expert sight readers. Mean total time for experts at N10 and N5 being 6.363 seconds
69
SD = 0.71 and 6.158 second SD = 0.70 respectively. Mean total time for non-experts at
N10 and N5 being 9.625 seconds SD = 3.56 and 8.649 second SD = 2.684 respectively.
T o ta l T im e
1 2 N o n -E x p e r t S ig h t R e a d e r s E x p e r t S ig h t R e a d e r s
) 1 0
s
c
e
s
(
e 8
m
i T
6
N 1 0 N 5 S iz e
F i g u r e 3 : S iz e (N 1 0 a n d N 5 ) p lo tte d a g a in s t T o ta l T im e w h e n p e rfo rm in g fo rm T 1 to T 2 fo r e x p e rt a n d n o n -e x p e rt m u s c i s ig h t re a d e rs . (E rro r b a rs = S E M ).
Discussion
Accurately assigning expertise was necessary at the outset of this thesis in order to ensure that subsequent studies have meaningful outcomes. Given that personal reporting of expertise is subject to bias (I. Gauthier et al., 2014; Lombardo & Deaner, 2014), a quantitative measure would warrant that conclusions made as a result of further testing would be valid. Regardless of the size of the targets or whether they were clear or blurred, there is a marked expertise effect for Total Time in all conditions.
70
Using time as the decisive variable for processing expertise is appropriate as it has been shown to be the common thread in the literature and is evidence of the superior processing abilities of the expert (K.A. Ericsson et al., 1993; K. A. Ericsson et al., 2007;
Farrington-Darby & Wilson, 2006). Admittedly pianists generally read music written for two hands and not a single line of notes with no additional dynamic or harmonic markings. However, in order to examine this type of processing, simplification of the task has been shown to be valid (Wong & Gauthier, 2009) and sufficient to exhibit expertise.
As such, the stimuli used in this study were considered to have been sufficient to demonstrate the desired effects. Requesting that such basic pieces be played ‘as fast and as accurately as possible’ might meant that if ‘chunking’ strategies were being employed by the experts, then they would perform the piece much faster than the non-experts. The results show that this is significantly the case.
While the finding was convincing, testing at a single grade level may have resulted in some participants being included in the non-expert category that should have been classified as experts. The criteria for inclusion as an expert was an accurate or near accurate performance at 6th grade level, but it is possible that Grade 5, for example, may be a more exact delineator. As such, it may be a factor affecting the findings between expert and no-experts in other cognitive processing areas and account for the large within group variation of non-experts and the failure for some findings to reach significance.
However, this variation it is more likely to be the result of a broader skill range in the non-expert group. Future studies with finer gradations of sight reading grades compared with time would be required to pinpoint more closely when the characteristics of expertise begin to appear. Even so, the level of difference based on the single criteria of
71
6th grade was highly significant for time between the two groups. This was considered sufficient to demonstrate evidence of expertise.
Conclusion
The second study found that a marked expertise effect existed for speed of performance when subjects were grouped for expertise according to their ability to accurately sight read at 6th grade AMEB level. This further strengthens the use of 6th Grade as the benchmark of sight reading expertise. The final study in this chapter examines the working memory capacity (WMC) of music sight readers. Study 3 tested if WMC results were significant for expertise when the criteria of 6th Grade sight reading success was used as the level of assigning sight reading expertise.
Study 3: Working Memory Capacity (WMC)
Background
Working Memory Capacity (WMC) has been previously discussed in relation to the innate abilities that visual experts demonstrate and as the main factor that prevents DP from eliciting expertise from anybody who undertakes it. The seminal work by Alan
Baddeley and Graham Hitch (Baddeley & Hitch, 1974) provided the springboard from which much subsequent research in the area has come. They proposed that there is a short-term visual and short-term auditory holding facility, called the ‘phonological loop’ and the ‘visuo-spatial sketch-pad’ respectively. It is from this store that items can be processed by a the ‘central executive’ (see Figure 4 below). This differed from the theories of the time which held that there was a Short Term Memory facility that fed
72
directly into Long Term Memory (Atkinson & Shiffrin, 1968). Baddeley and Hitch’s model differed by the addition of a working memory that could process using information from these short-term stores without necessarily involving long term memory. Their key findings related to the capacity and interactions between these two storage systems and the central executive. Of particular interest in the current context is the involvement of the phonological loop in the recall of visually presented stimuli.
Figure 4: The Working Memory Capacity Model (Baddeley, 2007)
The phonological loop involves deliberate, subvocalized rehearsal of items to be recalled.
It could vary from individual to individual and be affected by such things as stroke or trauma (Baddeley, 2007). It has been shown that the phonological loop can be disrupted by articulatory suppression (Murray, 1967). This occurs when a subject is required to repeat aloud a word like ‘the’ while attempting to memorise a string of items such as numbers or letters. Interestingly, Baddeley and Hitch’s initial experiments did not find a significant effect for ‘articulatory suppression’ (Baddeley & Hitch, 1974). Many
73
succeeding researchers, including themselves, later showed that articulatory suppression dramatically reduced the number of items that could be recalled from visually presented lists. This is because the visual material is unable to be recoded into verbal storage for rehearsal and recall as this capacity is occupied.
A measure of visual memory can be obtained using the Corsi Bock test (Milner, 1971). In this test, the examiner taps out an order of squares within a matrix of 9 and the subject is required to tap in the same sequence. The number per trial is increased until errors are made. It has been shown that articulatory suppression can slow down the rate of response in such a test, but not significantly affect the accuracy of recall (Vandierendonck, Kemps,
Fastame, & Szmalec, 2004). These authors also discuss the limitations of the Corsi test as a pure measure of visual memory. This is because, by its very nature, it is utilising visuo- spatial skills (tapping on blocks or on a computer screen presenting blocks) and is subject to significant interference when another spatial skill is imposed on the task. It has been found to be limited to 4 objects but can be extended if some measure of integration of individual parts into a single part can be achieved (Luck & Vogel, 1997). This is also known as ‘chunking’ and is a factor in facilitating more efficient processing and is the role of the ‘central executive’ and a feature of processing expertise (Gobet et al., 2001).
While superior WMC and VSTM have both been linked with expertise (Curby &
Gauthier, 2010; I. Gauthier et al., 2014; D. Hambrick et al., 2014), the utilisation of the phonological loop is integral to the attainment of a superior measure of WMC. Therefore,
WMC has been measured to assess innate processing abilities. If superior WMC can be found to be present in musicians generally, that is for category experience, it would
74
suggest that WMC may not be an innate feature required for expertise. In other words, do expert music sight readers have significantly superior WMC compared with non-experts?
The present study found such a difference. This finding supports the expertise literature from other domains (Hambrick et al., 2014) and the rationale of using 6th Grade AMEB sight reading assessment as an indication of sight reading expertise.
Hypothesis
That WMC is superior in music sight reading experts as defined in this thesis.
Methods
Participants
Following the granting of ethics approval from the UNSW Human Research Ethics
Advisory Committee to perform the study, 22 musicians were recruited from within the
UNSW student body and were reimbursed for their time. The musicians were then subdivided into 9 expert and 12 non-expert music sight readers according to this thesis protocol. One musician was omitted from the expertise classification due to not being a pianist. The participants were drawn from a different cohort to those in Chapter 4 studies
1 and 3 (Appendix 3).
Stimulus
The WMC was assessed using an online assessment from the following website: http://www.gocognitive.net/sites/default/files/stm.v1.0.a_1_0.swf. ‘GoCognitive’ began in 2008 and continues to be managed by cognitive science researchers at the University
75
of Idaho, USA. The site features numerous educational and testing resources freely available for researchers in the field.
Procedure
The subjects were seated 60 cm in front of a computer screen. An initial sequence of 4 numbers was presented on the screen with 2 second intervals between each number presentation. At the end of the sequence, the participant in then required to recall the numbers in order and type them into the computer and press ‘’ (see Figure 5). The number of items is increased by one until a mistake is made, at which point the test is repeated at the previous level until 3 correct responses are given and the process repeated.
The WMC score - number of items recalled - was reported after 20 trials.
Figure 5: Screen shot of ‘GoCognitive’ WMC task
76
Results
A t-test was performed to determine if expertise effects existed between expert and non- expert music sight readers and non-musicians for WMC and significance was assigned at the 0.05 level. The results were summarized in Figure 6. Non-expert sight readers and expert sight readers were plotted against the WMC Score. The error bars signify one standard error of the mean (SEM).
It was found that expert sight readers had significantly larger WMC than non-expert sight readers: M = 7.04, SD = 0.75 for experts and M = 6.76, SD = 0.89, t (18) = 2.643, p =
0.02.
W o r k in g M e m o r y C a p a c ity
d 8 .0
e
l
l
a c
e 7 .5
R
s
t
i g
i 7 .0
D
f
o
r
e 6 .5
b
m u
6 .0 N E x p e r t N o n -E x p e r t S ig h t R e a d e r s S ig h t R e a d e r s
Figure 6: WMC for expert and non-expert sight readers.
77
Discussion
WMC is the feature of innate ability linked to music sight reading expertise. It is the explanation for individual differences in performance despite equivalent DP (Hambrick et al., 2014). As Hambrick et al. also note, “some normally functioning people may never acquire expert performance in certain domains, regardless of the amount of deliberate practice they accumulate” (p 43). Consequently, the superior WMC of the expert music sight readers over non-experts in this study corroborates such assertions. The results further strengthen the use of 6th Grade AMEB sight reading as a valid level at which to demarcate sight reading expertise and is an important step in examining the cognitive abilities of the music sight readers.
Hambrick’s claim receives additional legitimacy when the survey results from the previous section are more closely examined. The average length of formal training for non-expert sight readers was 5-8 years. However, exactly half of this group had 8-10 years of DP – the average DP of the expert group.
Collectively, the results support the theory that innate abilities, like WMC, are required for expertise. DP is also a factor, but the survey results indicate that extensive DP did not always result in sight reading expertise.
Conclusions
WMC was found to be a feature of sight reading. The allocation of expertise according to the Grade 6 criteria has, therefore, been further supported. The key indicators of expertise
- evidence of DP, speed of performance and superior WMC - have been confirmed as
78
being significantly better for musicians that could successfully perform a 6th grade
AMEB sight reading assessment.
As a result, the following EM studies in Chapter 4 and cognitive studies in Chapter 5 have used Grade 6 AMEB sight reading ability as the criteria for expertise. Those subjects who could successfully perform at this level were placed in the ‘expert’ group and those who could not were labelled ‘non-expert’.
79
CHAPTER 4 Eye Movements and Music Sight Reading Expertise
Chapter 4 describes 3 experiments designed to investigate the EM patterns of piano music sight readers. Piano was selected for these studies as keyboard instruments were investigated in many of the foundational studies investigating the relationships between
EMs and sight reading (Jacobsen, 1941; Lang, 1961; Van Nuys, 1943; Weaver, 1943;
Young, 1971) and from which conflicting deductions have been made (see p.34 of this thesis). In addition, the researcher has extensive knowledge of the instrument and its pedagogy and formally trained pianists were relatively common amongst the cohort of university students from which this study drew participants.
Study 1explored how the EM patterns of expert and non-expert sight readers were affected when visual parameters of the score - blur and size – were systematically altered.
Study 2 looked at the effect that altering the predictable visual characteristics of the music score – spacing, note beam position and removal of bar lines - had on EM patterns and Study 3, the effect on EM patterns of imposing a speed constraint using a metronome.
Study 1: The Effect of Blur and Size on EMs when Sight Reading Music
Having categorised musicians according to their sight-reading ability, this study examined whether the physical attributes of the music score – blur, contrast and size – affected EMs in a manner similar to the patterns demonstrated for expertise in text reading literature.
80
Introduction
Studies in text reading have shown that expert readers have different EM patterns when compared with those of non-experts. Specifically, experts have fewer fixations that are of shorter duration (Underwood, Hubbard, et al., 1990). This is because they are able to
‘chunk’ groups of words into recognisable units for more efficient processing (Gobet et al., 2001). Non-expert text readers are characterised by fixating words individually
(Juhasz & Rayner, 2003) and backtracking more frequently, thereby increasing processing time (K. Rayner et al., 2006). In addition, it has also been shown that when visual parameters, such as blur, are changed in the text presentation, there is an effect on reading speed and EM patterns (G. Legge, 2007). Specifically, functional characteristics such as blur and contrast will affect saccade characteristics - direction, speed or latency - while cognitive features such as size and spacing will affect the fixation characteristics: number and duration (G. E. Legge, 2007).
The EM patterns of musicians have long been known to exhibit the chunking processing characteristics of expertise (Furneaux & Land, 1999; V. Kinsler & R. Carpenter, 1995;
Sloboda, 1974; Truitt et al., 1997; Wolf, 1976). However, little is known about the effects of changing basic visual parameters of the score - such as blur and size – on EM patterns and specifically how that relates to expertise. Altering features of the score may not have significant effects on expert music sight readers as other visual processing experts been shown to have a larger working memory capacity (WMC) that can be utilised when and if the short term processing reserves are taxed (K.A. Ericsson et al., 1993; Grabner, 2014).
The non-experts, on the other hand, may be significantly affected for this very reason.
81
In this study, EM patterns of musicians, as they sight read music score, were measured under conditions in which the size and blur of the music score was systematically changed. As a measure of EM performance, the duration and number of fixations, the number of forward and regressive saccades and their speeds and latencies were objectively measured. This was in order to determine whether the features that distinguish an expert from a non-expert text reader - fewer fixations of shorter duration and fewer regressive saccades in a shorter overall time - are similar to those of music reading; thereby further strengthening the delineation of expertise defined according to this thesis. In addition, control of the fundamental features of the visual stimulus - blur and size - were manipulated to further expand our understanding of how EMs are affected when sight reading music and how expertise might impact these responses. Conclusive evidence was found for perceptual chunking strategies employed by expert sight readers, and that blur and size affected EM performance under certain stimulus conditions.
Hypothesis
That the EM patterns of both expert and non-expert music sight readers will be affected by changes in blur and size when the read music score. These changes will be consistent with those found for expert and non-expert text readers.
Methods
19 participants’ EMs were measured as per the procedures outlined in the general methods section. At the completion of the study, each participant was asked to play a 6th
Grade piano sight reading assessment task as a first-pass performance and was ranked by
82
the researcher. 8 subjects were placed in the expert sight readers’ category and 10 in the non-expert. 1 non-expert was omitted due to an incomplete data set.
Results
Fixation number and duration, saccade speed, direction, number and latency were calculated from the EM recordings for each condition using purpose-written MatlabTM code and the results analyzed using Graphpad ‘Prism’TM. EM parameters for expert and non-expert music sight readers were subjected to a two-way analysis of variance having two levels of blur (0.00 and 3.2) and two size levels (N10 and N5). All effects were statistically significant at the .05 significance level and the results were summarized in
Figure 7. The Average Number of Fixations (Fig. 7a), Average Fixation Duration Fig.
7b), Average Number of Forward Saccades (Fig. 7c), Average Number of Regressive
Saccades (Fig. 7d), Average Forward Saccade Speed (Fig. 7e), Average Regressive
Saccade Speed (Fig. 7f) and Average Saccade Latency (Fig 7g) when performing from
T1 to T2 were plotted as function of blur grouped for size. Expert sight readers and non- expert sight readers were represented in each figure by the circle and square respectively.
The error bars signify one standard error of the mean (SEM).
83
F ix a tio n D u r a tio n # F ix a tio n s
2 0 3 7 0
1 8
3 5 0
)
s
s n
1 6 c
o
e
i
s
t
(
a 3 3 0
e
x
i
m
F i
1 4
# T
3 1 0 1 2
B lu r 0 .0 3 .2 0 .0 3 .2 B lu r 0 .0 3 .2 0 .0 3 .2 S iz e N 1 0 N 5 S iz e N 1 0 N 5
F ig u r e 7 a . F ig u r e 7 b .
# F o r w a r d S a c c a d e s # R e g r e s s iv e S a c c a d e s
1 1 0 9 0
1 0 0 8 0
9 0
s
s e
e 7 0
d
d
a
a c
8 0 c
c
c
a
a
S
S
6 0 # 7 0 #
5 0 6 0
5 0 4 0 B lu r 0 .0 3 .2 0 .0 3 .2 B lu r 0 .0 3 .2 0 .0 3 .2 S iz e S iz e N 1 0 N 5 N 1 0 N 5
F ig u r e 7 c . F ig u r e 7 d .
F o r w a r d S a c c a d e S p e e d R e g r e s s iv e S a c c a d e S p e e d
0 .0 7 0 .0 7
d
d
e
e
e
e
p
p
)
S
S
0
0 .0 5 e 0 .0 5
0
e
d
)
0
d
a
c
1
a
c
e
c
c
s
x
/
c
a
c
g
a
S
e
e
S
s
e
d
/
(
d
v
g
i
r
e s
a 0 .0 3 0 .0 3
s
d
(
e
w
r
r
g
o
e
F R
0 .0 1 0 .0 1 B lu r 0 .0 3 .2 0 .0 3 .2 B lu r 0 .0 3 .2 0 .0 3 .2
S iz e N 1 0 N 5 S iz e N 1 0 N 5
F ig u r e 7 e . F ig u r e 7 f .
S a c c a d e L a te n c y
5 0 0
L e g e n d :
4 0 0
y
c
n e
t E x p e rt S ig h t R e a d e rs
)
a
s
c
L
e s c 3 0 0
i . N o n -E x p e rt S ig h t R e a d e rs
m
d
(
a
c
c a S 2 0 0
B lu r 0 .0 3 .2 0 .0 3 .2
S iz e N 1 0 N 5
F ig u r e 7 g .
F ig u re 7 . B lu r, (d eg rees o f S D ) an d S ize, (N 1 0 an d N 5 ), p lo tted ag ain st 7 (a) # F ix atio n s, 7 (b ) F ix atio n D u ra tio n , 7 (c ) # F o rw a rd S ac c ad e s, 7 (d ) # re g re ssiv e S ac c ad e s, 7 (e ) F o rw a rd S ac c ad e S p ee d , 7 (f) R eg ressiv e S ac ca d e S p ee d an d 7 (g ) S a cca d e L aten c y w h e n p e rfo rm in g fro m T 1 to T 2 fo r e x p e rt an d n o n -ex p ert m u sic sig h t rea d ers. E rro r b ars = S E M .
84
Number of Fixations
N10 and Blur: The results revealed a significant effect of expertise; F (1,31) = 4.842, p =
0.04. Expert sight readers performed significantly fewer fixations than non-experts, particularly when the large score became blurred. There was no significant interaction between blur and expertise for N10; F (1,31) = 0.8134, p = 0.37.
N5 and Blur: There was no significant effect of expertise; F (1,30) = 0.3639, p = 0.55 and no significant interaction between blur and expertise for N5; F (1,30) = 0.0046, p =
0.95.
No Blur and Size: There was no significant effect of expertise; F (1,30) = 1.120, p = 0.30 and no significant interaction between size and expertise for no blur; F (1,30) = 0.2126, p
= 0.65.
3.2 Blur and Size: There was no significant effect of expertise; F (1,31) = 3.595, p = 0.07 and no significant interaction between size and expertise for 3.2 blur; F (1,31) = 2.120, p
= 0.16.
Large (N10), blurred (3.2) score was the greatest differentiator of expertise for the number of fixations. In this condition, expert sight readers decreased fixations (M = 121,
SD = 35.42), while the non-experts increased fixations (M = 177, SD = 70.10) were significant: p = 0.01.
Fixation Duration
Saccadic latency was taken into account when calculating Fixation Duration. That is, latency values were subtracted from the fixation duration values generated by the eye
85
tracker because this was a measure of the total time that the eye was not performing a saccade and therefore included the latency.
N10 and Blur: There was no significant effect of expertise; F (1,32) = 0.0038, p = 0.95 and no significant interaction between blur and expertise for N10; F (1,32) = 0.0005, p =
0.98.
N5 and Blur: The results did not quite reveal a significant effect of expertise; F (1,33) =
3.908, p = 0.057. Expert sight readers’ fixation durations were significantly shorter than non-experts when the smaller score became blurred. There was no significant interaction blur and expertise for N5; F (1,33) = 0.0792, p = 0.78.
No Blur and Size: There was no significant effect of expertise; F (1,33) = 1.663, p = 0.21 and no significant interaction between size and expertise for no blur; F (1,33) = 1.588, p
= 0.22.
3.2 Blur and Size: There was no significant effect of expertise; F (1,32) = 1.077, p = 0.31 and no significant interaction between size and expertise for 3.2 blur; F (1,32) = 0.9386, p
= 0.34.
The main finding regarding fixation duration was for small (N5) clear (0.00 blur) score.
The expert sight readers had shorter durations (M = 0.315, SD = 0.053), while the non- experts were longer durations (M = 0.339, SD = 0.028), close to significance: p = 0.06.
Number of Forward Saccades
N10 and Blur: The results revealed a significant effect of expertise; F (1,30) = 5.709, p =
0.02. Expert sight readers performed significantly fewer forward saccades than non-
86
experts particularly when the large score became blurred. There was no significant interaction blur and expertise for N10; F (1,30) = 1.238, p = 0.27.
N5 and Blur: There was no significant effect of expertise; F (1,31) = 0.6602, p = 0.63 and no significant interaction between blur and expertise for N5; F (1,31) = 0.2178, p =
0.64.
No Blur and Size: There was no significant effect of expertise; F (1,31) = 1.626, p = 0.21 and no significant interaction between size and expertise for no blur; F (1,31) = 0.0228, p
= 0.88.
3.2 Blur vs Size: The results revealed a significant effect of expertise; F (1,30) = 4.428, p
= 0.04. Expert sight readers performed significantly fewer forward saccades than non- experts. The interaction between size and expertise for 3.2 blur were not sigificant; F
(1,30) = 3.243, p = 0.08.
Large (N10) score, particularly when blurred (3.2 blur) has the greatest effect upon the number of forward saccades depending on expertise. The expert sight readers have fewer
(M = 61, SD = 8.62), while the non-experts perform more (M = 933, SD = 36.14) and were significant: p = 0.009.
Number of Regressive Saccades
N10 and Blur: The results revealed a significant effect of expertise; F (1,30) = 4.189, p =
0.05. Expert sight readers performed significantly fewer regressive saccades than non- experts; only slightly reducing when the large score became blurred as the non-experts
87
increased. There was no significant interaction blur and expertise for N10; F (1,30) =
0.1870, p = 0.67.
N5 and Blur: There was no significant effect of expertise; F (1,30) = 0.4117, p = 0.51 and no significant interaction between blur and expertise for N5; F (1,30) = 0.0355, p =
0.85.
No Blur and Size: There was no significant effect of expertise; F (1,30) = 1.484, p = 0.23 and no significant interaction between size and expertise for no blur; F (1,30) = 0.2489, p
= 0.62.
3.2 Blur and Size: There was no significant effect of expertise; F (1,30) = 2.600, p = 0.18 and no significant interaction between size and expertise for 3.2 blur; F (1,30) = 1.427, p
= 0.24.
Similar to forward saccades, large (N10) score blurred (3.2 blur) has the greatest effect upon the number of regressive saccades. The expert sight readers have fewer (M = 59,
SD = 18.85), while the non-experts perform more (M = 77, SD = 30.19) approaching significance: p = 0.059.
Forward Saccade Speed
No significant expertise or interaction effects were found for any condition of blur or size: p > 0.18 for expertise and interaction measures.
Regressive Saccade Speed
88
No significant expertise or interaction effects were found for any condition of blur or size: p > 0.13 and p > 0.14 for expertise and interaction measures respectively.
Saccadic Latency
N10 and Blur: There was no significant effect of expertise; F (1,31) = 1.636, p = 0.21 and no significant interaction between blur and expertise for N10; F (1,31) = 0.0505, p =
0.82.
N5 and Blur: The results revealed a significant effect of expertise; F (1,30) = 9.736, p =
0.004. Expert sight readers performed significantly shorter saccadic latencies than non- experts for both conditions of blur. There was no significant interaction blur and expertise for N5; F (1,30) = 0.0126, p = 0.91.
No Blur and Size: There was no significant effect of expertise; F (1,31) = 1.620, p = 0.21 and no significant interaction between size and expertise for no blur; F (1,31) = 0.5505, p
= 0.46.
3.2 Blur and Size: There was no significant effect of expertise; F (1,30) = 1.052, p = 0.31 and no significant interaction between size and expertise for 3.2 blur; F (1,30) = 0.2443, p
= 0.62.
Discussion
The results show that the main condition to effect saccadic latency is the larger (N10) size and it is not affected by blur. No significant interactions were found to exist between blur and size in relation to expertise. This may be due to the levels chosen for the study being of insufficient difficulty to place strain on processing resources to produce such a
89
response. This was particularly evident for experts in their Total Time as it barely changed from N10 to N5 or 0.00 to 3.2 blur (see Figure 3). Conducting additional studies using more blur and size levels may demonstrate more precisely how blur and size interacts between the groups. Nevertheless, the levels used in this study were of sufficient difficulty to address the issue of similarities between text reading and music sight reading expertise. In particular, the relationship between size and fixation characteristics, and blur and saccades.
An expert reader of text is expected to show an EM pattern that is characterized by fewer fixations of shorter duration with fewer regressive saccades than a non-expert (K. Rayner et al., 2006). This reflects the experts’ ability to 'chunk' pieces of visual information into units for more efficient processing (Gobet et al., 2001). The results suggest that this to be case for sight reading music score in this study.
Previous research into text reading found that blurring a stimulus effected saccade characteristics while changing the font size effected fixation features (G. E. Legge, 2007).
Size
No Blur: As mentioned, changes to the font size of print alters the fixation characteristics of EMs when reading text (G. Legge, 2007). The results for music sight reading show this trend for fixation durations when musicians sight read (see Figure 7b). In particular, the larger score size does not appear to affect the fixation durations, but the smaller size does.
The non-expert sight readers in the N5 condition came close to having significantly longer fixation durations than the experts: p = 0.06. This indicates that the non-experts were encountering processing difficulty with the smaller score - a known feature of text
90
reading (Engbert, Longtin, & Kliegl, 2002; E. Kowler, 2011; K. Rayner, 1998; K. Rayner et al., 2006).
When interpreting these results in light of previous research, it would appear that the size of the stimulus needs to be taken into account. Table 1 showed that size of the target was not always reported. It is, therefore, not surprising that one study claimed that fixation durations were longer for experts (Van Nuys, 1943), one shorter (Young, 1971) and another that fixations durations were the same between experts and non-experts (Schmidt,
1981). The only study that reported both the size of the target and the testing distance is
Young, 1971. The reported target equates to a stimulus slightly smaller (N4.6) than the
N5 stimulus used in this study. As the music score can be assumed to have been clear, the current study confirms the findings of Young, 1971 – experts’ fixation durations are shorter than non-experts with a clear target of equivalent size to N5 @ 60cm. On the other hand, Schmidt, 1981, may have had a target and testing distance more like N10 @
60cms. The current study also confirmed those findings – that fixation durations are no different between experts and non-experts.
Ultimately, the results agreed with Legge’s findings for text reading - that the size of the stimulus will affect fixation durations. This suggests that similar processing strategies may be being used for text. In this study, the smaller stimulus appears to pose a processing problem for non-expert sight readers.
One explanation for this finding may be that the N5 size is beyond the lower limit of
Critical Print Size (CPS) - the range of letter sizes over which a person can read at maximum speed. This means that small stimuli will tend to slow down the reading rate.
91
CPS is defined using whole letter size as opposed to the Minimum Angle of Resolution
(MAR) of those letters. Whole letter sizes having a resolution between 0.2 and 2 degrees at 40cm constitute the CPS (G. E. Legge, 2007). The N5 note head at 60 cm viewing distance subtended 0.132 degrees, which is below the CPS threshold. N10 sat comfortably within the range and would have afforded little challenge to reading speed.
This may account for the similar fixation durations for N10 score, as both groups are within a comfortable size for efficient processing. It would be expected, therefore, that
N5 score might produce an effect according to the CPS criterion. But it does not account for the opposite response found between the two groups when reading N5 score.
Recalling that saccadic latency was taken into account when determining fixation duration, its relationship with size can be seen from Figure 7g. While no expertise effects were found for size, a significant difference was found for saccadic latency in the N5:No
Blur condition: p = 0.03. The experts’ latency was significantly shorter than the non- experts. Despite a relative difference in the N10 condition being found, it was not significant. A longer saccadic latency is indicative of processing uncertainty (Cameron,
1995). Therefore, this finding along with the increased fixation duration previously discussed, suggests that the non-experts experienced processing difficulty relative to the experts in the N5 condition.
It can be seen that the expertise groups to had the opposite response when reading N10 when compared with reading N5, but it was not significant. The non-experts generally decreased speed for smaller score while the experts increased speed for smaller score.
92
To summarise the effect of size on EMs: the N5:No Blur condition appears to pose a processing challenge to non-expert sight readers. This was shown in their increased fixation duration and longer saccadic latency relative to expert sight readers in this condition.
Blur (3.2): The only expertise effect for size when the stimulus was blurred was found for the number of forward saccades: p = 0.04. The large target caused the non-experts to have significantly more forward saccades than the experts: p = 0.009, but almost no difference when small and blurred: p = 0.83: with the experts increasing and non-experts decreasing in number. This will be discussed in more detail below with reference to the effect that blur has on saccade characteristics.
Blur
N10: When the music score was large and blurred, expertise effect occurred for the number of forward saccades: p = 00.02, the number of regressive saccades: p = 0.049 and saccadic latency: p = 0.004. This paralleled the number of fixations, (with an expertise effect for N10: p = 0.03), as a fixation follows every EM. This suggests that blur could have had a major impact on the non-expert even when the size of the score was within the
CPS. It may be causing them to make many more EMs. It is interesting to note that the non-experts performed more forward and regressive saccades. This increase in forward saccades may be somewhat surprising as being taught to ‘look ahead’ is a common teaching strategy for music sight reading (J. Mishra, 2014). In this situation, the increase in forward EMs is prompted by physical features of the score (blur) and can be seen as an unnecessary inefficiency imposed upon the novice reader and should be avoided.
93
Expertise in music sight reading has been described as involving more than the mechanics of 'looking ahead' (Sloboda, 1985). Indeed, the mechanism of being able to
'chunk' information as an expert relates to 'top down' rather than 'bottom up' control
(Gobet et al., 2001). This would suggest that mechanisms not under conscious control are involved in the evolution of the expert music sight reader (J. Mishra, 2014).
Nevertheless, blur has been shown to effect the accuracy of performance (G. E. Legge,
Rubin, & Luebker, 1987) and should, therefore, be avoided for those of all levels of sight reading expertise.
In summary, a blurred score caused an increase in the number of forward and regressive saccades and saccade latencies for N10 score. Blur had negligible impact upon the saccade characteristics of either expertise group. In light of Legge’s claim that blur causing saccade features to be altered, it was found to be only for the larger size.
While EM patterns can be explained by the relationships between the number and duration of fixations and the speed and latency of saccades, the nature of the pattern is unexpected in light of the text reading literature. For example, if smaller score is causing processing difficulties for the non-experts – as indicated by an increase in the fixation duration - it would be expected that there be an accompanying increase in the number of fixations (K. Rayner et al., 2006). However, the number of fixations decreases from N10 to N5 for the non-expert. This would suggest that there is also something about the size of the N10 score that is difficult for the non-experts to process.
The number of fixations and duration findings are evidence of the ‘chunking’ abilities of the experts, assumed to be based on their skill in recognising note patterns. That is, they
94
can perform fewer fixations that are generally of shorter duration when playing the musical excerpts in this study. However, the musical pieces were deliberately composed to avoid such harmonic and melodic expectations that might favour the expert in pattern recognition and, in doing so, disrupt these skills. Despite this attempt at disruption, these results suggest that expertise may be influenced by a more generally enhanced processing ability than a domain specific recognition of musical patterns.
If the N10 condition is easier to process based on the CPS and if both expertise groups have fewer recognisable harmonic and melodic patterns in the score, similar EM patterns might be expected (Sloboda, 1977; Wurtz et al., 2009). This was obviously not the case.
A possible explanation is that the non-experts have a smaller visual and/or perceptual span. The perceptual span would be akin to the eye hand span and has been shown to be larger in experts (Sloboda, 1974; Truitt et al., 1997) and takes into account contextual information (O’Regan et al., 1983). Conversely, the perceptual span in text reading is smaller in less skilled readers (K. Rayner, 1986). The visual span, on the other hand, is said to be constrained by the structure of the visual system and not related to expertise
(G. Legge et al., 2007; Yu et al., 2007). Therefore, the reduced perceptual span of the non-expert might be responsible, forcing more fixations relative to the expert at N10.
However, if contextual information is less relevant by having minimised recognisable patterns in the music, the results might point to a difference in visual span. According to the literature, this is a functional bottle-neck and would not be possible (G. Legge et al.,
2007). This assertion was tested and reported in a later part of this thesis.
95
The EM results for the non-expert sight readers could also be accounted for by considering peripheral crowding. As previously discussed, the Total Time is significantly different between the two groups and could be accounted for by the compensatory mechanisms of trading off the number of fixations with their duration. This, may be so for non-experts, but it is not really the case for the experts. As a group, their EMs are not significantly affected by the smaller score across any condition. An explanation is that the non-experts experience more of a crowding affect for musical notes and, as such, require longer fixations. Recent research has found this to be the case (Wong & Gauthier, 2012) and the results of this present study correspond with those findings.
The expert may be experiencing some difficulty in processing N10:No Blur as they also show more fixations of shorter duration relative to their N5 condition. This is reflected in their longer saccade latency for N10 relative to N5. The fact that both groups’ saccade latency reduced for N5 suggests that the smaller score might fall within both spans. It is then subject to a measure of peripheral crowding effects for the non-experts – hence their increased fixation duration. Whether or not the score size can be reduced to a point where experts are forced to fixate as long as the non-experts, remains unanswered. Nevertheless, the current findings suggest that CPS alone cannot explain differences between expertise groups and may be expertise-dependent. That is, rather than be an absolute target size at a given distance as is the CPS for print, the CPS for print or music score may be dependent upon the reading expertise of the individual at a given size and distance. Further investigation into the relationship between peripheral crowding and sight reading expertise is to follow in subsequent studies in this thesis.
96
Conclusion
The results of this study suggest that many of the changes to EM features that occur when blur or size changes are introduced to text reading stimuli have a musical counterpart in sight reading. Size changes affected fixation durations and blur changes affected the number of forward and regressive saccades. Expert sight readers were shown to ‘chunk’ as they were able to complete the performances of the pieces with fewer fixations of shorter duration. However, when processing a stimulus of a size below the CPS, the fixation durations for non-experts were opposite to those of the experts. The non-expert sight readers in the N5 condition had increased fixation durations indicating that they were encountering processing difficulty with the smaller score. The functional effect of print size alone cannot be responsible for this variation of response between the groups.
The concept of reduced peripheral crowding for music notation in experts was introduced as a possible explanation for the difference (Wong & Gauthier, 2012). That is, when the score size is reduced, more notes will fill the peripheral field. If expert sight-readers have less peripheral crowding for musical targets, it might be that the non-experts experience increased crowding and respond with increased fixation duration. This idea is explored in
Chapter 5 of this thesis.
Study 2: The Effect of a Visually Disrupted Score on EMs when Sight Reading
In the previous section, expert music sight-readers were shown to be able to 'chunk' groups of notes for efficient processing as evidenced by fewer, shorter fixations. Size was found to be a hindrance for non-experts in that when the score was small, the fixation duration increased indicating that more time is required to process music presented below
97
the CPS. Also, the number of fixations was greater when the score was large. Crowding effects and a smaller visual span were proposed as the basis of these EM differences.
Having examined EMs with the score presented in a normal visual format, a study of expertise in relation to the visual disruption of music score was undertaken. Disruption of the expected visual format of music score may impact upon a musician’s ability to
‘chunk’ musical information and manifest as disrupted EM patterns. As such, this might indicate that learned spatial relationships in music notation are important for acquisition of sight reading expertise.
Background
Changing spacing in text affects word identification and manifests as saccade programming and longer programming (latency) results in shorter saccades and longer fixations (Perea & Acha, 2009). Decreased reading rate is also associated with spacing manipulations as is an increase in regressive saccades (K. Rayner, Fischer, & Pollatsek,
1998). Not only is spacing important, but also phrasing has been shown to aid comprehension in speech and syntax (Restle, 1972).
In relation to music sight reading, studies show that expertise is linked to speed and accuracy of pattern recognition, (Waters et al., 1998; A. Waters, Underwood, & Findlay,
1997) and that unexpected harmonic structures increased complexity of fixation patterns when reading music (Sloboda, 1977; Wurtz et al., 2009). However, what has yet to be determined is how expert and non-expert music sight readers respond when visual features of the music score are disrupted.
98
Notated western art music consists of ellipsoid symbols on a 5-note stave that are subject to established guidelines when presented visually. Traditional 4-part harmony involves lining up soprano, alto, tenor and bass parts vertically on the staff. In a particular chord, if one part requires its note to be held for 2 beats, (minim), another part may have two notes of half that duration (crotchets) or four notes of a quarter of that duration (quavers) to be sounded across the timeframe equal to the 2 beats of the minim. Conventional notation allows for these groups of smaller duration notes to be spaced so that notes to be sounded at the same time are roughly aligned vertically. In addition, should a chord have all 4 parts requiring a 2 beat duration, the next chord will not be placed directly against this chord as if in the position of a quaver or crotchet. It will be spaced as if the quavers or crotchets are actually there. This means that the spacing of the notes contains temporal information and interference with the expected duration/space relationship expected to affect processing (See Figure 8 below).
The direction of the note beam, or line that is attached vertically above or below a particular side of the note is also subject to rules, particularly for a single line of music. If the note head falls in a space or on a line above the centre line of the stave, the beam must go down from the left side. If the note head is below, the beam must go up from the right. If the note head is on the middle line it can go either way. Alteration of the direction and location of the note beam may introduce uncertainty in the score by disrupting the melody/beam expectation.
99
Figure 8: An example of music score illustrating the temporal aspects of music notation.
The bar line is the delineator of temporal grouping. For example, if there are four beats per bar, a bar line will appear after each note grouping of four beats. This is not unlike full stops or commas in the printed word in that it provides semantic cues to the reader. If semantic understanding is unclear, a text reader will perform longer and more frequent fixations with an increase in regressive saccades (K. Rayner et al., 2006). Uncertainty also causes an increase in saccadic latency (Cameron, 1995).
What has yet to be determined is how the alteration of semantic cues in music, that is, removal of bar lines and alteration note spacing and beam characteristics, might affect
100
expert and non-expert music sight readers. The present study measured EM patterns of these two groups as they read and played musical excerpts while these parameters were manipulated. Subtle fixation differences were found between the expert and non-experts sight readers.
Hypothesis
That both groups will show some disruption to their EM patterns when sight reading visually disrupted score, but such changes will be specific to their level of processing expertise.
Methods
Participants
A total of 20 people participated in the study. 9 were assigned to the expert sight reader group and 11 to the non-expert sight reader group. 1 member of each expertise group had participated in the previous EM study. All other participants were new subjects. Expertise was assigned according to the 6th Grade standard.
Stimulus
The music stimuli were presented at the larger size, N10 and with no blur. This is because it was found in the previous study, that this size of stimulus resulted in both expertise groups having fixation durations. In this way, it is more likely that any variations in EM that might be found will be a result of the disrupted stimuli only.
Procedure
101
Identical protocols were employed for stimulus presentation and recording as per the previous EM studies. Fixation and saccade characteristics were measured and compared between the two performances. The participants in this study were not the same subjects that were recruited for the previous EM study.
Each participant was required to sight read and play 10 specifically composed musical excerpts of 4 bars duration as quickly and as accurately as possible. This process was repeated for the same pieces with the bar-lines removed, altered inter-note spacing and unpredictable beaming directions (see Figure 9 below).
Figure 9a.
Figure 9b.
Figure 9: A sample musical excerpt illustrating normal spacing (4a) and disrupted spacing with omission of bar lines and incorrect note beaming (4b).
Results
Separate 2-way ANOVA and paired t-tests were performed to determine if specific effects existed between expert and non-expert music sight readers when music spacing was disrupted and significance was assigned at the 0.05 level. The results were summarized in Figure 10. The Average Total Time (Figure 10a), Average # Fixations
(Figure 10b), Average Saccade Latency (Figure 10c), Fixation Duration – less saccadic latency (Figure 10d), # Forward Saccades (Figure 10e), # Regressive Saccades (Figure
102
10f), Average Forward Saccade Speed (Figure 10g) and Average Regressive Saccade
Speed (Figure 10h) were measured when performing from T1 to T2 were plotted. The error bars signify one standard error of the mean (SEM).
Total Time
The results revealed a significant effect of expertise; F (1,40) = 28.16, p < 0.0001. Expert sight readers performed significantly faster than non-experts over both conditions. There was no significant interaction between music spacing and expertise for time; F (1,40) =
0.0248, p = 0.88.
# Fixations
No significant expertise effect: F (1,38) = 1.870, p = 0.18 or interaction between spacing and expertise were found: F (1,38) = 0.1688, p = 0.68.
Saccadic latency
No overall expertise effect was found for spacing: F (1,39) = 0.4818, p = 0.49. However, the disruption in spacing caused the expert group to have a significant increase in saccadic latency: F (1, 7) = 2.817, p = 0.03, while the non-experts showed little change.
Fixation Duration
As per the previous study, the saccadic latencies were subtracted from the duration measure form the eye tracker. No general expertise effects were found: F (1,39) = 0.2672, p = 0.61 and no interaction between expertise and spacing was found for FD: F (1,39) =
0.0641, p = 0.80.
103
T o ta l T im e T 1 :T 2 T o ta l # F ix a tio n s
1 2 6 5
5 5
) s
s 1 0
n
c
o
e
i
s
t
(
a 4 5
e
x
i
m
F i
8
# T 3 5
6 2 5
g g g n n g n i i in i c c c c a a a a p p p p S S S S l d l d a e a e m t t r p m p u r u o r o r N is N is D F ig u re 1 0 a D F ig u re 1 0 b
S a c c a d ic L a te n c y F ix a tio n D u r a tio n
3 0 0 0 .0 6 5
)
)
s
s
c
c
e
e s
2 5 0 s
(
m
(
n 0 .0 6 0
y
o
i c
2 0 0 t
n
a
e
r
t
u
a D
L 0 .0 5 5
1 5 0
c
n
i
o
i
d
t a
1 0 0 a
c
x c
i 0 .0 5 0
a
F S
g g g g n in in in i c c c c a a a a p p p p S S S S l d l d a e a e t m t m p r p r u u o r o r s N s N i F ig u re 1 0 d I F ig u re 1 0 c D D
# F o r w a r d S a c c a d e s # R e g r e s s iv e S a c c a d e s 3 5
3 5
s
s e
3 0 e
d
d a
a 3 0
c
c
c
c a
2 5 a
S S
2 5
e
d
r
v i a 2 0
s 2 0
s
w
e
r
r
o
g F
1 5 e
1 5
#
R
# 1 0 1 0
g g g g in in in in c c c c a a a a p p p p S S S S l d l d a e a e m t m t r p r p o u u r o r N s N s I I D F ig u re 1 0 e D F ig u re 1 0 f
L eg en d : E x p e r t S ig h t R e a d e r s . N o n -E x p e r t S ig h t R e a d e r s
F o r w a r d S a c c a d e S p e e d R e g r e s s iv e S a c c a d e S p e e d
0 .0 6 0 .0 7
)
)
d
d 0
0 0 .0 6
e
e
0
0
e
e 0
0 0 .0 5
p
p
1
1
S
S
x
x
e
e 0 .0 5
c
c
d
d
e
e
a
a
s
s
/
/
c
c
g
g c
c 0 .0 4
e
e a
a 0 .0 4
d
d
S
S
( (
0 .0 3 0 .0 3
g g g g in in in in c c c c a a a a p p p p S S S S l d l d a e a e m t m t r p r p u o u o r r N s N s I I F ig u re 1 0 h D F ig u re 1 0 g D F ig u r e 1 0 : N o r m a l a n d D is r u p te d S p a c in g p lo tte d a g a in s t T o ta l T im e ( F ig u r e 1 0 a ) , N u m b e r o f F ix a tio n s ( F ig u r e 1 0 b ) , S a c c a d ic L a te n c y ( F ig u r e 1 0 c ) , F ix a tio n D u r a tio n le s s S a c c a d ic L a te n c y ( F ig u r e 1 0 d ) , N u m b e r o f F o r w a r d S a c c a d e s ( F ig u r e 1 0 e ) , N u m b e r o f R e g r e s s iv e S a c c a d e s ( F i g u r e 1 0 f ) F o r w a r d S a c c a d e S p e e d ( F ig u r e 1 0 g ) , a n d R e g r e s s iv e S a c c a d e S p e e d ( F ig u r e 1 0 h ) w h e n p e r f o r m in g m u s ic a l e x c e r p ts f r o m T 1 to T 2 f o r e x p e r t a n d n o n - e x p e r t m u s ic s ig h t r e a d e r s . E r r o r b a r s = S E M .
104
# Forward Saccades
No general expertise effects were found: F (1,38) = 1.493, p = 0.23 and no interaction between expertise and spacing was found for the number of forward saccades: F (1,38) =
0.1492, p = 0.70.
# Regressive Saccades
Regressive saccades behaved in a similar fashion to forward saccades showing no significant expertise effects: F (1,38) = 2.278, p = 0.14 and no interaction between expertise and spacing was found: F (1,38) = 0.2732, p = 0.60.
Forward Saccade Speed
No general expertise effects were found: F (1,38) = 1.130, p = 0.29 and no interaction between expertise and spacing was found for the number of forward saccade speed: F
(1,38) = 0.0109, p = 0.92.
Regressive Saccade Speed
In a similar fashion, no general expertise effects were found: F (1,38) = 2.238, p = 0.14 and no interaction between expertise and spacing was found for the number of regressive saccade speed: F (1,38) = 9.469e-066, p = 0.99.
Expert sight readers performed significantly faster than non-experts: p=<0.0001. Spacing disruption had no significant effect on the Total Time within either group. Saccadic latency was the only other measure to reach significance and this was for experts only when encountering disrupted spacing – the latency increased significantly: p = 0.03.
105
Discussion
The musicians in this study were different to those taking part in the blur and size study as well as the following metronome study. Their expertise group was assigned according to their ability to perform he 6th Grade sight reading assessment task. It can be seen from
Figure 10a that the Total Time to perform the standard N10:No Blur condition was significant: p <0.0001: as it was in the blur and size study. Therefore, placing them in their expertise groups for this and the subsequent cognitive studies would appear to be justified.
The Total Time taken to perform from T1 to T2 was negligibly affected by disruption to the music score for either group. The experts, yet again, performed significantly faster and demonstrated ‘chunking’ characteristics in both conditions by performing faster with fewer fixations of shorter duration than the non-experts. However, the experts revealed that the visual disruption of the score interfered with their normal cognitive processes as evidenced by an increased saccadic latency in the disrupted condition (Cameron, 1995).
Opinion is divided on the relationship between fixations and saccades. Some researchers have suggested that uncertainty causes saccade cancellation and increased fixation duration (Perea & Acha, 2009; Yang & McConkie, 2001). Others advocate that the response to uncertainty is for longer latencies with fewer and shorter saccades (Cameron,
1995; Kowler & Anton, 1987). The results from the current study appear to agree with the latter model as a significant increase in latency was found. However, no firm conclusions can be drawn due to the large errors of the mean within each group. This may be due to having too few subjects in the study, insufficient disruption of the score to
106
cause a large enough response in most observers or have something to do with the role of spacing and punctuation differences between text and music.
For example, punctuation in text is vital for the reader to understanding meaning.
Removing key elements of punctuation completely alters the meaning of a sentence even though the same words are used in exactly the same order. For instance, the sentence,
“Go and eat, Grandma!” does not mean the same thing as ‘Go and eat Grandma!” The absence of the comma between 'eat' and 'grandma' changes the sentence from an appeal for Grandma to have a meal to a request for someone to have Grandma as the meal! In the context of this study, the bar line was the phrase delineator. Unlike removing the comma in a sentence, the removal of the bar line in music does not alter meaning - a minim followed by a crotchet followed by two quavers is a group of notes of the same pitch and duration regardless of whether or not they are separated by a bar line. Altering the space between the notes or the direction of the beam also does not alter a note’s pitch or duration, but it does change the way it should look. As such, it has the potential to create an element of uncertainty and, therefore, it has the potential to disrupt the automatized cognitive processing habits of the reader. Such a result was evident in the responses of expert music sight readers.
The previous study examined the effects of changing the size and blur features of the score and explored differences between the responses of the expertise groups. It was found that smaller score caused a processing problem for the non-experts that resulted in their increase in fixation duration relative to the experts. Blurred score caused an increase in the numbers of saccades performed by the non-experts without significantly altering
107
their fixation durations. The physical parameters of the score, when presented at those specific levels of blur and size at 60cms, appeared to have little effect on the EM patterns expert sight readers. However, by altering the expected locations - spacing patterns and other visual cues that enhance musical interpretation according to the rules of western art music notation – the experts appeared to encountered a processing problem. This manifested as a significant increase in latency compared with the normal spacing condition. The non-experts, on the other hand, did not show the significant increase in
EMs or fixation durations that occurred with the changing of the more functional features of the score. This suggests that the experts were less able to benefit from their extensive knowledge of the music form in an automatic fashion. Perhaps their less crowded peripheral vision for musical notes was somehow sabotaged by the unexpected spaces and inappropriate structures in the field of view. Regardless of the exact aetiology of the problem, the chunking mechanisms were interrupted for the experts. Future investigations introducing greater visual complexity and disruption to the normality of the score – such as the inclusion of unexpected, non-musical symbols - might shine further light on the nature of interference effects in relation to expertise.
Another consideration is the role of the micro-fixation. Previous studies have shown that object identification can be attained following a fixation of as little as 80-100ms duration
(Salthouse & Ellis, 1980). As technology has improved and/or the ability to measure and account for the noise in the system, the figure has diminish – 50ms (K. Rayner, Pollatsek,
& Binder, 1998) and 40ms (Nystrom & Holmqvist, 2010). The current study was not sufficiently sensitive to detect such small fixation durations. Also, the role of these micro fixations has yet to be determined in relation to visual processing expertise in music sight
108
reading and may yield valuable insights into differing processing strategies under different conditions.
Conclusion
The present study showed that expert music sight readers were more susceptible to disruption in the visual appearance of music score than their non-expert counterparts.
This manifested as the increased saccadic latency found in the disrupted condition.
These results further develop the picture of expertise in music sight reading. The expert responded minimally to structural interference of the score, Study 1, but appeared to react with greater uncertainty than the non-expert when encountering interference that disrupts cognitive processing.
Study 3: The Effect of Imposed Time-Keeping on EMs when Sight Reading
Study 1 reported how certain blur and size changes affected the EM patterns of non- expert sight readers as they read music score, while Study 2 showed that expert sight readers were more affected by disruption to the expected visual presentation of the score.
In both studies, the participants were instructed to play the pieces ‘as quickly and as accurately as possible’ so that the resultant performance speed was at the individual’s discretion to ensure accuracy. Imposing a specific timing regulation would mean that some musicians would be playing slower while others might be stretched to the limit of their ability or even beyond, unable to conform to the time pressure required. The measurement of EM patterns of musicians such conditions is the subject of Study 3.
109
Introduction
The metronome in music pedagogy is a time-keeping device that emits clicks or tones at a prescribed rate. It is valuable for teaching the rudiments of rhythm and can also be used as an adjunct to mastering notation (Miller, 2012). There is a general consensus that the metronome is an indispensable tool for helping novice percussionists keep in time, but that there is little advice or suggestion on how best it is to be used other than to simply use it (Falle, 2011).
Recent research has shown that as the level of musicians’ expertise increases, so too does the use of the metronome in their personal practice time. Also, the use of the metronome is, in fact, a predictor of expertise when measured by the grade of performance attained and of a higher score in that grade and its use significantly increases at Grade 6 level and beyond (Hallam, Rinta, Varvarigou, & Creech, 2012). Conversely, this research shows that there is minimal use of the metronome up to that grade level.
It is, therefore, surprising that the metronome’s importance is stressed in early percussion instruction, albeit with little specific direction, and in the practice patterns of more experienced musicians, but appears to be somewhat overlooked in other early instrumental teaching when all musicians – not just percussionists - need to keep accurate time. Differences exist between the EM patterns of expert and non-expert music sight readers (V. Kinsler & R. H. S. Carpenter, 1995; Sloboda, 1977; Wurtz et al., 2009). The preceding studies confirm this; as well as describing differences in EMs when visual expectations are disrupted. However, little is known about how the imposition of time
110
restriction using an audible time-keeper might affect EM patterns of musicians as they read a music score.
The present study investigated the EM patterns of expert and non-expert music sight- readers of keyboard as they read and played musical excerpts initially without and then with the aid of the metronome. Fixation duration and number as well as forward and regressive saccade speed and number were measured. Based on these findings, speed boundaries were found within which performance accuracy was maintained for expert sight-readers and beyond which accuracy was lost for non-experts.
Background
Clear vision involves the ability to detect and identify an object relative to its background
- the smaller the difference that can be detected, the better the visual acuity. A person with good visual acuity has a lower threshold of difference detection than a person with poor vision – they require smaller differences in size, brightness and/or contrast in order to be able to detect and identify an object. Similarly, in the auditory domain, there exists an asynchrony threshold whereby a person with a low threshold can tap more accurately to a beat. That is, there is a smaller time difference between the onset of the beat and the actual tap than for a person with a higher threshold. Differences in the asynchrony threshold have been found, not only between musicians and non-musicians, but between the types of instrument played and the players’ level of expertise (Krause, Pollok, &
Schnitzler, 2010). It was found that professional pianists and percussionists had equivalent asynchrony thresholds while amateur pianists and non-musicians had increasing higher thresholds and were less sensitive to timing changes (Ehrle & Samson,
111
2005; Krause et al., 2010; Repp, 2010). Accuracy in synchronising was also found to be more accurate in those musicians who had commenced training before the age of 7 and that this earlier trained group was also more accurate in synchronising with a visual stimulus (D. Watanabe, Savion-Lemieux, & Penhune, 2007).
When comparing the ability to synchronise a tap with an auditory stimulus and a visual stimulus, it was found that participants were able to synchronise more accurately with the auditory over the visual (Repp & Penel, 2002). The study also concluded that motor control is better guided by auditory information and that this was the case even when both auditory and visual stimuli were presented at the same time.
In summary, the evidence suggests that the utilisation of a metronome in early instrument education enhances rhythmic skills and accuracy in time-keeping. However, there is little science behind how best to utilise the metronome. The present study examined how the imposition of a metronome impacted upon the EM patterns of expert and non-expert music sight-readers and whether this information could be used to prescribe how best to use it in music instrumental practice. Recommendations for its judicious use, while maintaining accuracy in performance, were made.
Hypothesis
That the EM patterns of non-expert music sight readers will show evidence of cognitive load when excessive timing demands are imposed by a metronome.
Methods
Participants
112
The same 19 subjects participated in this study as in the earlier blur and size study: 8 experts and 10 non-expert sight readers.
Stimulus
The music stimuli were presented at the larger size, N10 and with no blur – as in the previous study. This is because these parameters posed the least processing challenge to either group.
Procedure
Identical protocols were employed for stimulus presentation and recording as per the previous EM study. Fixation and saccade characteristics were measured and compared between the two performances.
Each participant was required to sight read and play 9 specifically composed musical excerpts of 4 bars duration as quickly and as accurately as possible. This process was repeated for the same pieces with the participants instructed to play in time with the metronome. The setting was 120MM (120 beats per minute).
Results
Separate two-way ANOVA and paired t-tests were performed to determine if specific effects existed between expert and non-expert music sight readers when a metronome set at 120MM was imposed and significance was assigned at the 0.05 level. The results were summarized in Figure 11. The No Metronome and Metronome Conditions were plotted against Total Time (Figure 11a), Number of Fixations (Figure 11b), Total Fixation
Duration, (Figure 11c), Fixation Duration less the Saccadic Latency (Figure 11d),
113
Saccadic Latency (Figure 11e), Number of Forward Saccades (Figure 11f), Forward
Saccade Speed (Figure 11g), Number of Regressive Saccades (Figure 11h) and
Regressive Saccade Speed (Figure 11i) when performing musical excerpts from T1 to T2 for expert and non-expert music sight readers. Error bars = SEM.
An overall expertise effect for Total Time was found: F (1,31) = 10.86, p = 0.0025.
Experts performed significantly faster in both conditions. There was no interaction between expertise and metronome for time: F (1,31) = 1.595, p = 0.22.
There was an overall expertise effect for the number of fixations: F (1,30) = 10.48, p=0.003: non-experts executing significantly more fixations. However, the introduction of the metronome caused the opposite response from each group: the experts decreased the number of fixations when the metronome was introduced and the non-experts increased. The number of fixations being significantly different between the groups in the metronome condition: p = 0.03. The introduction of the metronome saw the expert sight readers decrease the number of fixations, but it was not significant: p = 0.08.
The groups responded in the opposite direction for Fixation Duration. However, the data revealed no significant expertise or metronome effects: F (1,32) = 0.0199, p = 0.89 and F
(1,31) = 0.6621, p = 0.42 respectively. No interaction was found between expertise and metronome for fixation duration: F (1,32) = 1.615, p=0.21. However, the fixation duration drops for the non-experts when the metronome was introduced, but it was not significant: p = 0.09.
114
Both groups increased saccadic latency, but no significant effects were found. Expertise:
F (1, 30) = 0.0028, p = 0.96, Metronome: F (1,30) = 0.69, p = 0.41 and Interaction: F
(1,30) = 0.1470, p = 0.70 respectively.
The non-experts performing significantly more EMs - forward and regressive - than the experts. Overall expertise effects were found for the number of forward saccades: F
(1,31) = 9.779, p = 0.004: the non-experts performed significantly more forward saccades than non-experts. There was no metronome effect: F (1,31) = 0.4381, p = 0.51: and no interaction between expertise and metronome for forward saccades: F (1,31) = 1.465, p =
0.24. Similarly, for regressive saccades a significant expertise effect: F (1,31) = 7.801, p
= 0.009 with no metronome or interaction effects: F (1,31) = 0.0478, p = 0.83 and F
(1,31) = 0.8272, p = 0.37 respectively.
Forward and regressive saccade speeds were not significantly affected by the metronome for either expertise group and demonstrated large with group variations.
115
T o ta l T im e T 1 :T 2 # F ix a tio n s
1 1 2 0 0
) s
s 9
n
c
o e
i 1 5 0
s
t
(
a
e
x
i
m
F i
7
# T
1 0 0
5
t t t t e e e e M M M M o o N N
F igure 11a F ig ure 1 1b
F ix a tio n D u r a tio n S a c c a d ic L a te n c y
0 .0 3 4 0 .0 3
)
s
c
e
y
s
c
(
n n
0 .0 3 2 e o
t 0 .0 2
)
i
a
s
t
c
a
L
e
r
s
c
u
i
m
d
D
(
a n
0 .0 3 0 c 0 .0 1
o
c
i
a
t
S
a
x
i F 0 .0 2 8 0 .0 0 t t t t e e e e M M M M o N o N F ig u re 1 1c F ig ure 1 1d
L eg en d : E x p e r t S ig h t R e a d e r s . N o n -E x p e r t S ig h t R e a d e r s
# F o r w a r d S a c c a d e s # R e g r e s s iv e S a c c a d e s
1 0 0 1 0 0
s
e
s e
9 0 d 9 0
d
a
c
a
c
c c
8 0 a 8 0
a
S
S
e
v
d 7 0 7 0
i
r
s
a
s
e w
6 0 r 6 0
r
g
o
e
F
R 5 0
5 0
# #
4 0 4 0 t t t t e e e e M M M M o o N N F igure 11e F igure 11f
F o r w a r d S a c c a d e S p e e d R e g r e s s iv e S a c c a d e S p e e d
0 .0 8 0 .0 8
d
d
e
e
e
e
p
p
S
S
e
e
d )
0 .0 6 ) 0 .0 6
d
a
c
c
a
c
e
e
c
c
s
s
/
/
c
a
g
g
a
S
e
e
S
e
d
d
(
(
d v
0 .0 4 i 0 .0 4
r
s
a
s
e
w
r
r
g
o
e F 0 .0 2 R 0 .0 2 t t t t e e e e M M M M o o N N F igure 11g F igure 11h
F ig u r e 1 1 : N o M etro n o m e an d M etro n o m e p lo tted ag ain st 1 1 a) T o tal T im e, 1 1 b ) # F ix atio n s, 1 1 c) F ix atio n D u ratio n , 1 1 d ) S acca d e L aten cy , 1 1 e) # F o rw ard S accad es, 1 1 f) # R e g re ssiv e S a cc a d e s, 1 1 g ) F o rw a rd S a cc a d e S p e e d a n d 1 1 h ) R e g re ssiv e S a cc a d e S p e ed w h en p erfo rm in g fro m T 1 to T 2 fo r ex p e rt an d n o n -ex p ert m u sic sig h t read ers. E rro r b ars = S E M .
116
Discussion
The introduction of the metronome added a level of interference to the reading and performing of music score. Each participant’s original playing speed was their fastest and most accurate possible speed; the metronome setting of 120MM forced both groups to play faster.
The average time for the non-experts to play from T1 to T2 at their fastest correct speed was 9.46 seconds. When the metronome was set at MM=120, T1 to T2 was then required to be played at 6 seconds. This represented an incremental increase in speed of approximately 58% over their average fastest correct speed to perform from T1 to T2. As a result, they were pushed beyond their ability to preserve accuracy; as evidenced from the audio files of the experiment. Different strategies were employed to cope with this performance stress: some played in time for one bar at a time, pausing to process the next bar which was then played in time, followed by another pause; some played at 60MM
(half the speed); while others ignored the metronome completely. It is, therefore, not surprising that the EM patterns are quite different, if not opposite, to those of the experts at the same speed.
The non-expert group showed an increase in speed. This indicated an overall attempt to keep up and was achieved by increasing the number of fixations while decreasing their duration. The saccades were consequently of shorter duration and therefore, shorter length and with longer latencies. This is characteristic of a ‘speed/accuracy trade-off’
(Cameron, 1995); where increased errors are allowed in favour of speed (p. 76). This is similar to findings for text reading where it has been shown that increased processing
117
pressure leads to shorter saccades of less duration that have taken longer to programme
(Kowler & Anton, 1987). The results of this study also confirm other researchers’ findings of short duration fixations under time pressure for text reading (Gobet et al.,
2001; Underwood, Hubbard, et al., 1990).
The average time for the experts to play from T1 to T2 at their fastest correct speed was
6.34 seconds. When the metronome was set at MM=120, T1 to T2 was then required to be played at 6 seconds duration. This represented an increase in speed of approximately
6% from their average fastest correct speed to perform from T1 to T2. At this pace, the experts were able to maintain accuracy of performance. Their EMs showed the opposite response to those of the non-experts; a decrease in the number fixations with an increased duration. This is indicative of saccade cancellation in response to processing stress (Yang
& McConkie, 2001) rather than increasing the number of fixations.
The results indicate that the expertise groups have responded differently to the pressure of keeping up with an imposed timing framework. Because the amount of stress differed compared with their initial speed, the conclusion cannot be drawn that the non-expert group would always adopt the same reading strategy regardless of the relative change in speed. That is, the non-expert group may adopt the strategy of saccade cancellation with increased fixation duration if the increase in stress was of an order of magnitude similar to that imposed on the expert group in this study. In the same way, if the expert group was exposed to a more extreme stress they may adopt the speed/accuracy trade-off strategy and perform more EMs that have taken longer to programme with shorter fixation durations.
118
Nevertheless, when considering the research supporting the superior timing skills of percussionists (Ehrle & Samson, 2005; Krause et al., 2010; Repp, 2010), the benefit of early training (Falle, 2011; D. Watanabe et al., 2007) and that motor control is better guided by auditory information (Repp & Penel, 2002), it seems that the metronome is a grossly underutilized tool in general music instrumental instruction. However, the results of the present study suggest that it would be unreasonable to simply set the metronome at a prescribed speed and hope that a student is able to manage to play the piece successfully - especially if that speed is excessively greater than what might be their personally selected speed to ensure accuracy of performance. A 58% increase in speed was imposed on the non-expert group and their performance failed. Despite their slower average time taken to play from T1 to T2, the performances of the non-expert sight- reading group were, nevertheless, accurate at their original, self-selected speed to ensure performance accuracy. Studies of typists indicated that their self-selected speed was somewhat conservative and approximately 10-20% below potential (K. A. Ericsson et al., 2007). It is reasonable to suggest that an individual’s fastest, most accurate speed be initially assessed and used as a basis from which to increase the metronome rate.
While sight-reading expertise can be demonstrated in the EM patterns of pianists who can correctly perform a 6th Grade sight reading examination passage, there is also a significantly greater use of the metronome by musicians at and above 6th Grade level
(Hallam et al., 2012). However, it does not follow that using the metronome ensures music sight-reading expertise and such a recommendation is not a recipe for ensuring such success. Music sight-reading expertise is a multi-factorial endpoint attained after many years of training with extensive theoretical understanding of western art music
119
forms (K.A. Ericsson et al., 1993; Hambrick et al., 2014; Meinz & Hambrick, 2010;
Repp, 2010). This study does, however, highlight the apparent underutilisation of the metronome in many areas of music education and that more extensive research needs to be done into the earlier introduction and systematic use of the metronome in early instrumental instruction.
In order to better understand how incremental speed changes might affect music sight- readers, future studies should concentrate on smaller increases in speed to determine at what percentage increase performance accuracy is lost and examine how the EM patterns may change at these points. It is not clear from the present study whether the resultant
EM patterns are related to expertise or extreme processing stress regardless of expertise.
That is, would the experts show an EM response to that of the non-experts at a metronome setting of MM = 190? Conversely, would the non-experts’ response resemble the experts with the metronome set at MM=127?
Therefore, based on the results of this study, it is realistic to suggest that the metronome be set well within the conservative 10-20% buffer of the student’s fastest speed in order to maintain accuracy (K. A. Ericsson et al., 2007). From this point, speed can be incrementally set until the desired level is obtained; thereby maintaining accuracy throughout the process.
Conclusion
EM patterns indicated that music sight readers experienced cognitive load when pushed to perform faster than their fastest accurate speed. The experts, at a 6% increase in speed, were situated well within their potential to perform accurately. The non-experts, at a 58%
120
increase in speed, were well beyond the 10-20% buffer and were unable to maintain an accurate performance. The resultant EM patterns were opposite and may be more related to the level of processing stress encountered and not necessarily associated to the level of sight reading expertise.
Formulating a definitive set of recommendations for the use of metronome was not possible, but some suggestions were made. Nevertheless, the results do consolidate some intuitive practices (involving slowing a piece down to a manageable speed and then slowly increasing it to the required level) for using the metronome in instrumental teaching, encourage consideration for its earlier and more habitual use in an individual’s private practice and highlight areas of further investigation.
Expertise, Music Sight Reading and Eye Movements – a Summary
Table 4 below summarises the findings from the EM studies. Total Time was significantly faster for the expert music sight readers in every condition and has not been included. Forward and regressive saccade speeds did not reach significance in any condition and have also not been included.
It seems that experts perform longer, shorter or similar fixation durations relative to the non-experts depending on the target size. This suggests that the broad range of the CPS is due to expertise rather than purely the size and, therefore, the angular subtense of the stimuli. The nature of visual and perceptual span will follow in subsequent sections of this thesis in order for this idea to be more fully explored.
121
The effect of blur was to alter saccade characteristics and this is also similar to the findings for text reading. However, these were found only in the large (N10) condition, not in the smaller (N5). When the N10 target became blurred, the non-experts performed significantly more EMs in both direction and saccadic latency reduced. Of particular interest is the only significant size effect found. At N10:3.2 blur condition, the non- experts were executing significantly more forward saccades than the experts. At N5:3.2 blur, they became almost identical. This again might point to the influence of the periphery and how the span and/or the crowding effects might differ between the expertise groups. If the span is smaller for the non-experts, it might be expected that they will require more EMs as they can fix fewer notes into that span. Also, the blur of the target would render the more peripheral notes in that span even less likely to be deciphered thereby increasing the number of EMs required. When the size is smaller, more notes can then be within the span and the blur becomes less relevant.
The imposition of the metronome caused different responses that were perhaps linked more with processing load than expertise. A reasonable hypothesis that could be drawn from the study is that when a moderate speed stress is placed on a performer, the result appeared to be EM cancellation with increased fixation duration. Extreme speed stress – as was the experience of the non-expert group – appears to cause an increase in both saccadic latency, the number of EMs performed with a decrease in fixation duration. This could be further investigated by stressing the expert group at an appropriate level.
Another more general processing hypothesis to be tested, based on these results, is whether expertise dictates the type of response that occurs when a processing challenge is
122
encountered. That is, do experts generally adopt a saccade cancellation strategy with increased fixations durations while non-experts perform more EMs of shorter duration – a more speed/accuracy trade off approach?
Significant results in Table 4 are marked with an asterisk and the colours highlight opposite responses that were found between the expertise groups. It can be seen that the non-expert group does appear to have responded with a speed/accuracy trade off pattern
(Cameron, 1995; Kowler & Anton, 1987) for both the score disruption and metronome conditions while the experts showed a saccade cancellation pattern (Perea & Acha, 2009;
Yang & McConkie, 2001). As such, the response might not be an either/or response but related more to the expertise level of the individual for the specific task being undertaken.
Looking to other visual processing expertise groups and designing experiments to confound their strategies would explore these ideas as well as looking at deeper disruption and stressors in music sight reading to confirm the current findings.
Non-Expert Sight Readers Expert Sight Readers
Blur @ SL ↓ EM ↑ ● FD ↑ SL ↓ EM ↓● FD ↑ N10 Size
Size @ 3.2 SL ↓ EM ↓ FD ↑ SL ↓ EM ↓ FD ↓ Blur
Disruption SL ↑ EM ↑ FD ↓ SL ↑ ● EM ↓ FD ↑
Metronome SL ↑ EM ↑ ● FD ↓ SL ↑ EM ↓ ● FD ↑
123
Table 4: Summary of significant effects reported in the present study. Legend: SL – Saccadic latency, EM – Eye Movements, FD – Fixation Duration. Denotes a significant finding.
In conclusion, these studies suggest that music sight reading demonstrates many similar
EM patterns to those of text reading. Changing the blur and size characteristics of score show similar effects to those of text reading, but the expertise groups were affected differently depending on the size of the target. This suggests a difference in the ability to analyse targets in the periphery and raised the question of whether expert sight readers might be generally less susceptible to the effects of peripheral crowding. They also appear to have a different EM response to cognitive load, which raises the question of whether it is the way a processing expert responds to challenge or if it is relative to the amount of challenge imposed at a particular level of expertise.
Nevertheless, these studies have shown a number of significant differences in EM patterns between the levels of expertise as defined previously. That is, the results of the
EM studies further support the relationship between sight reading expertise and the ability to successfully perform a 6th Grade AMEB sight reading assessment. Those achieving this level were also found to have significantly superior features of expertise - hours of DP and superior WMC. As such, the following cognitive studies were undertaken using the 6th Grade criterion as the basis of assigning expertise.
It is to be noted that the majority of the participants in the following studies were not from the same cohort as the original EM study. The survey data included both groups of participants.
124
CHAPTER 5 Cognitive Processing and Music Sight Reading Expertise
Background
Chapter 5 explores the visual processing characteristic of the two expertise groups - using the 6th Grade AMEB sight reading as the bench mark of expertise.
Study 1 examined the Working Memory Capacity (WMC) capabilities of non-musicians when compared with the two expertise groups of music sight readers. WMC was found to be an innate ability said to be responsible for the development of expertise along with
Deliberate Practice in Chapter 3.
Study 2 explored the visual spans and developed the questions raised in relation to EM patterns and different sizes of target. That is, whether there were differences to be found in the perceptual and visual spans of musicians and non-musicians when measured using letters – non-musical targets.
Study 3 further assessed the notion of speed as the hallmark of expertise by utilising a modified version of the Rapid Automatized Naming (RAN) using simple words rather than individual shapes – again exploring differences in target type and non-musicians.
Study 4 examined the auditory abilities of the subject by testing their ability to distinguish between subtle changes in pure sound tones.
Study 5 measured the accuracy of visual search using musical and non-musical targets. In addition, different sounds patterns were introduced that were designed to correlate or not correlate with the target being searched.
125
These cognitive studies have been selected based on their known relevance in text reading expertise. For example, the perceptual span has been found to be broader in expert text readers (K. Rayner, 1986) and poor discrimination of tone frequency modulation has been linked with reading dyslexia (Ahissar et al., 2006; Amitay et al.,
2002; Anvari et al., 2002). To date such testing has either not been directly applied to music sight reading expertise. The following studies were the first step towards examining the question of whether processing expertise transfers across domains and whether expertise or exposure to musical training can confer a visual processing benefit in non-musical domains.
Study 1: Working memory Capacity (WMC)
Background
WMC is known to be a predictor of expertise in chess players (Hambrick et al., 2014;
Meinz & Hambrick, 2010) and in music sight reading (Hambrick et al., 2014). In Chapter
3, WMC for numbers, (a non-musical target), was shown to be significantly superior in sight reading experts compared with non-expert music sight readers. In order to examine the role of WMC in music sight reading more generally, the difference between musicians and non-musicians remained to be explored. The present study investigated the relationship between musical experience rather than expertise and WMC by dividing the musicians into expert and no-expert groups according to their performance of a 6th grade
AMEB sight reading assessment task.
126
Hypothesis
That WMC is a general measure of visual processing expertise.
Methods
Participants
Following the granting of ethics approval from the UNSW Human Research Ethics
Advisory Committee to perform the study, 39 participants were recruited from within the
UNSW student body and were reimbursed for their time. 22 were musicians and 17 had no music training and were classified as non-musicians. The participants were drawn from a different cohort to those in Chapter 4 studies 1 and 3 (Appendix 3).
Stimulus
WMC was assessed using an online assessment from the following website: http://www.gocognitive.net/sites/default/files/stm.v1.0.a_1_0.swf. A more detailed description is on pp 75-76 of this thesis.
Procedure
The subjects were seated 60 cm in front of a computer screen. An initial sequence of 4 numbers was presented on the screen with 2 second intervals between each number presentation. At the end of the sequence, the participant was then required to recall the numbers in order and type them into the computer. The number of items was increased by one until a mistake is made at which point the test is repeated at the previous level until 3 correct responses are given and the process repeated. WMC is reported after 20 trials.
127
Results
Non-parametric Mann-Whitney U tests were performed to determine if specific effects existed between musicians and non-musicians for WMC and significance was assigned at the 0.05 level. The results were summarized in Figure 12 where Musicians and Non-
Musicians were plotted against the WMC Score - the number of digits correctly recalled.
The error bars signify one standard error of the mean (SEM).
Musicians and Non-Musicians showed no significant difference in WMC. Musicians:
(Mdn = 6.96) and non-musicians (Mdn = 6.59), U = 126, p = 0.29.
W o rk in g M e m o ry C a p a c ity
7 .5
d
e
l
l
a
c
e R
7 .0
s
t
i
g
i
D
f o
6 .5
r
e
b
m u
N 6 .0
s s n n ia ia ic ic s s u u M -M n o N
Figure 12: Musicians and non-musicians plotted against number of digits recalled.
128
Expert and non-expert sight readers were then separated and compared with the non- musicians. A 1-way ANOVA was performed to determine if expertise effects existed between expert and non-expert music sight readers and non-musicians for WMC and significance was assigned at the 0.05 level. The results were summarized in Figure 13:
Non-Musicians, non-expert sight readers and expert sight readers were plotted against the
WMC Score. The error bars signify one standard error of the mean (SEM).
W o rk in g M e m o ry C a p a c ity
8
d
e
l
l
a
c
e
R
s
t
i g
i 7
D
f
o
r
e
b
m u
N 6 N o n -M u s ic ia n s N o n -E x p e r t E x p e r t S ig h t S ig h t R e a d e rs R e a d e rs
Figure 13: Non-musicians, non-expert sight readers and expert sight readers plotted against number of digits recalled.
An expertise effect was not significant for WMC: F (2,33) = 2.945, p=0.07. However, the comparison (Mann-Whitney test) between expert and non-expert sight readers was
129
significance: WMC was significantly superior for experts: (Mdn = 7.45) than non-experts
(Mdn = 6.36), U = 20, p = 0.02.
Discussion
The results show WMC being significantly better for expert music sight readers over non-experts, but not for musicians as a group over non-musicians. It is possible that there were individuals in the non-musician group who had visual expertise in another area – chess or gaming for example (Hambrick et al., 2014; Meinz & Hambrick, 2010) and this may account for non-musicians having better WMC than non-expert sight readers. No information was gathered on non-musicians’ possible areas of expertise in this study.
Chess experts are known to have superior WMC (Hambrick et al., 2014; Meinz &
Hambrick, 2010). Many hours of DP in chess do not guarantee expert status being achieved (Campitelli & Gobet, 2011) and, as such, have similar traits to sight reading expertise. Campitelli quotes a study of WMC in chess players compared with non-players and found no difference in their results (A. J. Waters, Gobet, & Leyden, 2002). This is similar to the current musician/non-musician findings. Water’s study did not compare
WMC in expert chess players with non-expert chess players. Given the results between expert and non-expert sight readers, superior WMC would logically exist for them and other visual processing experts as well. Consequently, the non-musician group could demonstrate better WMC than non-experts sight readers if such individuals were part of the non-musician group.
Table 5, below, outlines a summary of expertise features found thus far in the thesis. It will be expanded throughout subsequent sections to facilitate coherent comparison of
130
results across studies. All of the study participants, having undertaken literacy training, would have achieved text reading expertise. Music training of a formal nature, similar in content to the AMEB syllabus, can produce musicians capable of sight reading. They do not have visual processing expertise in this domain, but are differentiated from non- musicians by having undertaken deliberate practice of a musical instrument. Expertise has been shown to occur when DP co-exists with superior WMC. Both the expert and the non-experts have DP as a feature different from non-musicians, but only the experts have the innate ability, WMC, that can allow expertise to develop in conjunction with substantial DP.
CATEGORY Non-Musician Non-Expert Sight Expert Sight Reader Reader
SKILL Literacy Literacy + Music Literacy + Music Sight Sight Reading Reading
OUTCOME Text Reading Expert Text Reading Expert Text Reading Expert
FEATURES DP DP
WMC
Table 5: Summary of Processing Features (Version 1) showing the skills of the expertise groups as a result of both nature and nurture influences.
The remaining studies in this chapter report the findings of experiments that further probe the extent to which formal music training impacts visual expertise. Do the experts get all
131
of the processing benefit or does some cognitive advantage be shown to be conferred to the non-expert sight readers over non-musicians. The findings will be summarised by additions to the table above.
Conclusions
The WMC was found to be a feature of sight reading expertise and not of musicians generally. The significance of this finding will be further explored in the general discussion in light of the findings of the remaining studies.
Study 2: The Perceptual and Visual Spans
The EM findings from Study 1 in Chapter 4 suggested that non-expert sight readers encountered processing difficulties at N10 – the larger target size. It was suggested that they might have a smaller visual span and, as such, were forced to deploy more short saccades/more fixations. A narrower visual span for music targets might be expected given that expert sight readers have less peripheral confusion for music targets (Wong &
Gauthier, 2012). Study 3 from Chapter 3 showed that the expert sight readers had significantly enhanced WMC compared with non-experts and this was measured in a non-musical domain. Assessing visual spans using non-musical targets allows comparison, not only between music expertise groups, but also with non-musicians. In this way, Table 5 will be expanded to illustrate what visual processing advantages might be present in sight reading musicians.
132
Introduction
The human visual system utilises two types of span: the perceptual and the visual span
(O’Regan et al., 1983). The visual span is the region within which the eye can correctly identify letters around fixation and differs from the perceptual span in that the latter takes into account contextual information and is generally larger than the visual span (O’Regan et al., 1983). A smaller perceptual span has been found in novice text readers and thought to be due to the inability of less skilled readers to utilise parafoveal information (detail to the side of fixation) as efficiently as a skilled reader (K. Rayner, 1986). Parafoveal information is subject to crowding effects (when objects in side vision become indistinct due to the overlapping receptive fields of retinal ganglion cells), (G. Legge et al., 2007;
Levi, 2008; Pelli et al., 2007) and mirror confusion of peripheral objects/letters (S. T.
Chung, 2010; S. T. Chung & Legge, 2009).
The visual span is thought to be smaller than perceptual span and limited to the structure of the visual system and therefore a ‘bottom-up’ control (G. Legge, 2007). Smaller visual spans are related to a decrease in reading rate (G. Legge, 2007; G. Legge et al., 2007; Yu et al., 2007). Such spans were measured using a moving window method where EMs are not factored into account. However, a recent study, looking at reading speed in relation to fixation duration, when reading a sentence conventionally, found that a smaller visual span was related to increased number of fixations also of longer duration (Risse, 2014).
This EM pattern reflects the characteristics of a novice text reader (Underwood, Hubbard, et al., 1990). It is not known if an expert text reader might have a larger visual span if it is measured in a manner that more closely resembles normal reading. It has also been found
133
that faster readers were more accurate at letter identification around central fixation,
(Risse, 2014, p7). Testing visual and perceptual spans using tachistoscopic presentation of targets around fixation may reveal other facets of expertise in visual processing. For example, whether crowding effects exist, the nature of which might vary with processing expertise.
In addition, the larger perceptual spans in expert text reader may be reflecting the minimised crowding and mirror confusion effects of a larger visual span. This is not unreasonable given that recent research in music sight reading has shown that experts in that field have reduced crowding effects for musical notes when compared with non- expert sight readers (Wong & Gauthier, 2012). This was also substantiated in the previous chapter of this thesis, where non-expert music sight readers increased fixation duration when reading small score compared with experts.
Therefore, the relationship between the visual and perceptual spans may not be as independent as originally indicated. The visual span for text may increase as expertise develops. This occurs concurrently with increased domain knowledge, which leads to the contextual understanding of words and increased perceptual span. What is currently not known is if the visual span has flexibility in this way, as it is thought to be fixed according to the physical constraints of the visual system.
A feature of visual processing is Left Side Bias (LSB) where details to the left side of fixation are more accurately identified than the right side. It is thought to be feature of expert processing of objects rather than holistic processing by some researchers (Hsiao &
Cottrell, 2009; Tso et al., 2014). In this study, experts in Chinese font identification did
134
not show LSB when fonts were unfamiliar, but only when they were familiar (Tso et al.,
2014). However, familiar or unfamiliar faces show LSB (Hsiao & Cottrell, 2009). That is, holistic processing of faces is involved even if the face is unfamiliar. If facial recognition expertise is a more foundational level of visual processing, then this might not be the expected outcome – why would an unfamiliar font not involve holistic processing for an expert reader as it is still a Chinese character? By way of explanation, perhaps Tso’s target presentation was too long (500ms) and the experts were able to execute more than one fixation - only 250ms are required to initiate a saccadic EM (Cameron, 1995). Details to the right of fixation could then be perceived and dampen the feature of LSB. Keeping the target presentation at less than 250ms might yield different results, as might testing for LSB in non-Chinese characters among expert and non-expert Chinese readers.
The current study investigated the visual and perceptual spans using letters and words presented tachistoscopically around central fixation. The results were compared between musicians and non-musicians and expert and non-expert music sight readers. The perceptual span was found to be similar across all groups. The visual span was found to be significantly different between musicians and non-musicians.
Hypothesis
That the visual span for letters is not a fixed entity but is influenced by visual processing expertise in other domains.
Methods
Participants
135
Following the granting of ethics approval from the UNSW Human Research Ethics
Advisory Committee to perform the study, 36 participants were recruited from within the
UNSW student body and were reimbursed for their time: 22 were musicians and 15 had no music training and were classified as non-musicians. The musicians were then subdivided into 9 expert and 13 non-expert music sight readers according to this thesis protocol.
Stimulus
Perceptual span was tested using words of 3, 5, 7, 9, 11, 13, 15 and 17 letters. Each word size category comprised 3 different words – 40 trials in all. The words were generated using a custom written programme for MATLAB, presented on a linearized 27-inch
Mitsubishi Diamond Pro monitor driven at a frame rate of 80Hz. and presented in random order for a duration of 250ms to ensure a single fixation. Each letter was in upper case and measured 18x22 pixels, corresponding to a 6/36 Snellen letter at the viewing distance of 60cm. Each letter was separated by half of the width of the letter - 9 pixels - to maximize acuity/minimize crowding (T. Norton, Corliess, D. and Bailey, J., 2002). The visual span was tested using strings of 3, 5, 7 and 9 random letters. Each string size category comprised 5 strings that had been pre-randomized for presentation – 20 trials in all. The stimuli were generated and presented as above. The word and random letter lists are presented in Appendix 2 and experimental design illustrated below in Figure 14.
136
A B C D
■ Tone B T W
Fixation target 100ms 200 ms Response
Figure 14: Visual and perceptual span experimental design. A: Fixation target presented on the computer screen until button is pressed. B: Warning tone sounds for 100ms. C: Letter target appears on screen for 200ms. D: Verbal response
Procedure
Each participant was seated 60cm from the computer screen and was instructed to call out the words (in the case of perceptual span), the results were recorded, transcribed and scored according to correct identification. For example, a 9-letter word scored 9 points if correctly identified and zero if incorrect. In the case of the visual span, as many letters as possible in the correct order were called out, recorded, transcribed and scored. Two scoring methods were employed. ‘Raw score’ was a measure of absolute correctness.
That is, if the letter string was PZYWNOF and all were identified in the correct order with no additions or omissions, the score was 7/7. If the response was PZYMHFO, the score would be 4/7 or 57%. The ‘Weighted Score’ allowed for the detection of a wider visual span by accepting correctly identified letters that had been reversed due to peripheral crowding effects (S. T. Chung, 2010; S. T. Chung & Legge, 2009). Each string was scored as follows: a point for each letter seen, a point for correct letter identification
137
and a point if identified in the correct position. The above example would be scored 7+
5+3 = 15/21 or 71%.
The relationship between the groups in terms of left side bias (LSB) for processing
(Hsiao & Cottrell, 2009) was investigated by calculating the LSB Ratio. This was found by scoring the 7 and 9 letters strings for absolute correctness to the left and right of fixation. These levels were chosen because the majority of participants correctly identified 3 and 5 letter strings. The example above would have a ratio of 3:0 – 3 correct to the left of fixation and none to the right.
Results
Unpaired t-tests were performed to determine if specific effects existed between musicians and non-musicians for perceptual and visual spans and significance was assigned at the 0.05 level. The results were summarized in Figure 15. Musicians and
Non-Musicians plotted against Span Score (% Correctly Identified) for Visual Span_Raw
(Figure 15a), Visual Span_Weighted (Figure 15b), Perceptual Span (Figure 15c) and Left
Side Bias (Figure 15d). The error bars signify one standard error of the mean (SEM).
Perceptual Span
No significant difference was found between musicians and non-musicians: t (33) =
0.2411, p = 0.81.
Visual Span
138
The visual span was significantly broader for musicians than for non-musicians for both the weighted and raw score analyses. Weighted Visual Span: t (33) = 3.682, p = 0.0008.
Raw Visual Span: t (32) = 2.865, p = 0.007.
LSB
Musicians and non-musicians were significant for LSB: F (1, 63) = 592.7, p < 0.0001.
The LSB ratio was calculated using the data from the visual span experiment. As such, the breadth of the LSB was significantly greater, but the ratio of left correct to right correct is not significant: Mean L/R difference non-musicians = 22.54 +/-1.4 and musicians = 22.2 +/-1.1.
139
V isu a l S p a n _ R a w V isu a l S p a n _ W e ig h ted
8 5 8 5
d
d
e
e
i
i
f
f
i
i t
t 8 0 8 0
n
n
e
e
d
d
I
I
y
y l
l 7 5 7 5
t
t
c
c
e
e
r
r
r
r o
o 7 0 7 0
C
C
% % 6 5 6 5 M u sic ia n s N o n -M u sic ia n s M u sic ia n s N o n -M u sic ia n s
F ig u re 1 5 a F ig u re 1 5 b
P e r c e p tu a l S p a n L e ft v s R ig h t L e tte rs C o rre c t
9 0 M u s ic ia n s 3 0
t N o n -M u s ic ia n s
t
c
c
e
r
e r
r 8 5
r
o o
C 2 0
C
s
r
s
r
e
t
e
t t
t 8 0 e
e 1 0
L
L #
7 5 0 M u sic ia n s N o n -M u sic ia n s L e ft C o rre c t R ig h t C o rre c t
F ig u re 1 5 c F ig u re 1 5 d .
F ig u r e 1 5 : M u sician s an d N o n -M u sician s p lo tted ag ain st S p an S co re (% C o rrectly Id en tified ) fo r V isu al S p an _ R aw (F ig u re 1 5 a), V isu al S p an _ W eig h ted (F ig u re 1 5 b ), P ercep tu al S p an (F ig u re 1 5 c) an d L eft S id e B ias (F ig u re 1 5 d ). T h e erro r b ars sig n ify o n e stan d ard e rro r o f th e m ea n (S E M ).
A 1-way ANOVA was then performed to determine if specific effects existed between between non-musicians, non-expert sight readers and expert sight readers for Visual
Span_Raw, Visual Span_Weighted, Perceptual Span and LSB Ratio with significance assigned at the 0.05 level. The results were summarised in Figure 16. Non-Musicians,
Non-Expert Sight readers and Expert Sight Readers plotted against Span Score (%
Correctly Identified) for Visual Span_Weighted (Figure 16a), Visual Span_Raw (Figure
16b), Perceptual Span (Figure 16c) and plotted against Score (L:R Correctly Identified)
140
for Left Side Bias Ratio (Figure 16d). The error bars signify one standard error of the mean (SEM).
Perceptual Span
No significant musicianship effect was found for perceptual span: F (2,31) = 0.7281, p =
0.49. Multiple comparisons also revealed no significant between groups differences.
Visual Span
RAW - A significant effect for musicianship was found for the visual span: F (2,31) =
3.810, p = 0.03. The raw visual span was equivalent between the expert and non-expert sight readers. Post hoc multiple comparison tests showed that both the expert and the non-experts approached significance with the non-musicians: t (31) = 2.261, p = 0.06 for experts and t (31) = 2.421, p = 0.06 for non-experts.
WEIGHTED - A significant effect for musicianship was found for the visual span: F
(2,33) = 6.449, p = 0.004. Post hoc multiple comparisons revealed that experts had a significantly higher span than non-musicians: t (33) = 1.916, p = 0.0035.
LSB
A significant LSB ratio was found: F (1, 59) = 549.0, p <0.0001. Multiple comparisons revealed no significant differences between groups.
141
)
V isu a l S p a n _ W e ig h ted ) V isu a l S p a n _ R a w
d
d
e
e
i
i
f
f i
8 5 i 8 0
t
t
n
n
e
e
d
d
I
I
y
8 0 y
l
l
t
t c
c 7 5
e
e
r
r
r
r o
7 5 o
C
C
%
% 7 0
(
(
e
7 0 e
r
r
o
o
c
c
S
S
n
6 5 n 6 5
a
a p
N o n - N o n - E x p e r t p N o n - N o n - E x p e r t
S S M u s ic ia n s E x p e r t S ig h t M u s ic ia n s E x p e r t S ig h t R e a d e r s S ig h t R e a d e r s R e a d e r s S ig h t R e a d e r s F ig u r e 1 6 a . F ig u r e 1 6 b .
L e ft S id e B ia s_ L S B P e r c e p tu a l S p a n
9 0 3 0
t
t
c
c e
e N o n -M u sic ia n s
r
r r
r 8 5 2 0 o
o N o n -E x p e r t S ig h t R e a d e r s
C
C
s
s E x p e r t S ig h t R e a d e r s
r
r
e
e
t
t
t t
e 8 0 1 0
e
L
L
# #
7 5 0 t t t N o n - N o n - E x p e r t c c c e e e r r r M u s ic ia n s E x p e r t S ig h t R e a d e r s r r r o o o S ig h t R e a d e r s C C C F ig u r e 1 6 c . ft ft ft e e e L L L
F ig u r e 1 6 d .
F i g u r e 1 6 : N o n -M u s ic ia n s , E x p e rt a n d N o n -E x p e rt S ig h t R e a d e rs p lo tte d a g a in s t S p a n S c o re (% c o rre c tly id e n tifie d ) fo r V is u a l S p a n _ R a w (F ig u re 1 6 a ), V is u a l S p a n _ W e ig h te d (F ig u re 1 6 b ), P e rc e p tu a l S p a n (F ig u re 1 6 c ) a n d L S B R a tio (F ig u re 1 6 d ). E rro r b a rs s h o w S E M .
Discussion
It was expected that the perceptual span might be similar between the groups as the task was simply word recognition. The tachistoscopic presentation would not allow for the contextual and semantic information available as when reading (K. Rayner, 1998), but may rather be testing knowledge of English vocabulary. As such, the results might be influenced by participants not having English as their first language. However, 4 of the study participants were known by the researcher to have English as their second language: 3 musicians and 1 non-musician. The mean perceptual span for musicians was
83.83, (SD 1.77). The musicians scored 86, 85 and 51, while the non-musician scored 84.
142
Therefore, it was considered not to be a weakness in the experimental design. Similar testing using an individual’s mother tongue would be required to totally validate this claim. Nevertheless, the perceptual spans were not significantly different between the three groups.
The finding of the visual span being significantly broader for musicians than for non- musicians was not expected given that it is considered to be somewhat fixed (G. Legge,
2007; G. Legge et al., 2007; Yu et al., 2007). However, the methods used to attain those results predominantly using a moving window testing in the peripheral field and could account for the variation in findings. Testing in the central field is more akin to natural conditions and has been shown to be more accurate for faster readers (Risse, 2014).
Recognition of letters would be expected to be enhanced for a fast/expert reader relative to a non-expert perhaps, but not necessarily between musicians and non-musicians. All study participants were either undergraduate, postgraduate or academic staff at UNSW and, therefore, considered to have attained expertise in reading English text. One explanation for a broader visual span in musicians is that formal music training somehow reduces peripheral crowding effects for non-musical objects.
Further evidence for this hypothesis is the weighted visual span score. The weighting allows for correctly identification of letters but with mirror-reversals (S. T. Chung, 2010;
S. T. Chung & Legge, 2009). The non-musicians’ span scores improve, but the difference between the non-musicians and the experts becomes more significant when the non- experts are removed from the analysis: from p = 0.007 to p = 0.004. This suggests that musicians might be subject to less peripheral confusion for the recognition of non-
143
musical objects. On the other hand, the expert and non-expert sight readers’ differences were significant in the weighted situation only, suggesting that the experts might have further enhanced peripheral processing abilities.
Another hypothesis is that this may also be a manifestation of their better working memory capabilities. For example, the Sperling ‘visual sensory memory’ (Sperling,
1960) tested subjects with a 50ms exposure of random letters. The correct position as well as the correctly identified letter in the wrong position were recorded an analysed. No difference was found between measuring the span in these different ways. However, a span of only 6 letters was used. Calculating the size according to the parameters given,
(0.45inch letter viewed at 22 inches), approximates to 6/9 (20/60) notation at that distance. No indication of the spacing between the letters was given. The present study used letters of 6/36 size separated by ½ the width the minimise crowding effects. These are 4 times the size of the Sperling targets and yet the span is not much different. This suggests that perhaps this measurement is one of memory span rather than visual span.
Nevertheless, the question remains whether the span is a physical processing bottleneck.
Sterling’s method could be expanded to vary size, spacing and numbers of objects to gain a better understanding. Expertise would need to be considered in this process as one of
Sperling’s subjects, NJ (see P5, Sperling 1960), clearly had a greater immediate recall of letters compared with the other participants. Might NJ have had visual processing expertise in some area to account for this?
The method used in this study for measuring visual span might call into play the phonological rehearsal strategies of WMC (Baddeley & Hitch, 1974). However, is was
144
not an expertise effect that was found, just musicianship. On the other hand, Legge’s method used only 3 targets per presentation, which is below the memory recall threshold.
By not increasing the target size in the periphery to account for the drop in identification acuity, Legge might not be measuring a structural bottleneck as much as quantifying the visual acuity in the periphery/overlapping receptive fields or even memory recall as
Sperling did. As Sperling stated, objects were seen in the periphery, but they were not able to be named. The naming of objects is not what the peripheral retina is for, but to direct attention and fixation towards the object for accurate identification.
However, the number of objects seen rather than the number correctly identified may be of more relevance in the context of reading either text or music. As stated, correct identification is not the purpose of the peripheral retina. The limit of correct target identification is of interest if the observer has reduced foveal vision capabilities as in certain disease states like macular degeneration. However, in the normal vision situation the ‘gist’ of what is occurring in the peripheral view is of interest. From this point, the
EMs can be directed to a point on the text or score where foveation and attention are the means of verification. Differences between identification, detection and crowding in relation to expertise would need to be studied by testing variations in size, location and separation of non-specific targets in order to make form conclusions about the physical nature of the visual span. Requiring that subjects attend only to the targets in the periphery while focusing centrally might eliminate the effect of the superior WMC of expertise.
145
Expert and novice Chinese writers show LBS when fonts were familiar but not when they are unfamiliar. This is not the case for faces – familiar and unfamiliar show LSB (Hsiao
& Cottrell, 2009; Tso et al., 2014). The current study used familiar targets – letters - and compared participants in groups of known visual processing expertise from a different domain. The equivalent differences between left and right letter identification scores for the three groups suggests that LSB is a feature of holistic processing generally as found by Hsiao, (Hsiao & Cottrell, 2009) rather than expertise. The higher scores of the musicians are reflective of their larger visual spans or WMC when compared with non- musicians - the ratios of letters correctly identified to the left and right of fixation was equivalent across the groups.
Testing the visual span of non-musicians using musical targets proved difficult. It was originally intended to test the visual span using musical targets as well as letters but was abandoned as non-musicians were unable to reliable detect more than one shape either side of the centre line on the music staff. Such a target was eventually used for the visual search task (see Figure 21). Non-musicians could interpret these 3 shapes in their correct position, when presented centrally, as a pattern matching exercise. Testing the visual spans using a target of equal familiarity to both groups was used. By testing with letters, an experience effect was found rather than an expertise effect.
As such, the Summary of Processing Features Table can be expanded as follows in Table
6 below. Music training enhances an individual’s peripheral processing abilities for a non-musical target as evidenced by the visual span results. LSB ratio is a feature of all groups, although the numerical values are influenced by the visual span differences and
146
evidence of holistic processing. Further evidence for a reduced peripheral crowding effect for expert music sight readers using non-musical targets will be examined in the following sections on Random Word Speed Reading and Visual Search and the summary of findings table expanded accordingly.
CATEGORY Non-Musician Non-Expert Sight Expert Sight Reader Reader
SKILL Literacy Literacy + Music Literacy + Music Sight Sight Reading Reading
OUTCOME Text Reading Expert Text Reading Expert Text Reading Expert
FEATURES LSB DP DP LSB WMC Visual Span LSB Visual Span
Table 6: Summary of Processing Features (Version 2) showing the addition of LSB and Visual Span.
Conclusions
The study results suggest that the visual span may be more indicative of WMC and/or peripheral crowding measures than a physical processing bottleneck, but this is not conclusive. Musicians showed significantly less peripheral confusion for non-musical targets by showing broader visual spans. This is the first measure that suggests an enhanced visual processing advantage in musicians who have learnt to read music that transfers to another domain. LSB ratio was found to be equivalent for non-musicians and
147
expert and non-expert sight readers – indicating that all groups are engaging in general holistic processing of familiar letter targets.
Study 3: Rapid Automatized Naming of Words - WordRAN
Speed of processing is the hallmark of expertise in a domain. This was demonstrated in
Chapter 3 of this thesis. What is not known is whether speed of visual processing in one domain (music notation) can transfer to speed of processing in another (word reading).
Having found LSB and visual span advantages to be present in musicians, this may confer an advantage in reading words aloud as quickly as possible. Study 4 investigated whether this could be possible.
Background
Rapid Automatized Naming, (RAN), is measured by naming objects, colours, letters or numbers out loud as fast as possible (Lervag & Hulme, 2009). A poor RAN ability is strongly linked with reading difficulties, particularly dyslexia (Stainthorp et al., 2010).
The basic mechanism is thought to be the extended pause time required to process visual difference in or features of targets in order to identify them. It does not appear to be a generalised slowness of processing as it does not occur in the auditory domain
(Stainthorp et al., 2010). Poor readers may not be able to process a word as a single unit, but rather rely on the sequential processing of the individual letters of a word. This results in the slower, albeit often correct, response. The ‘holistic’ nature of object processing was confirmed in a study which found that the fusiform gyrus was activated more for objects than for individual letters (Misra, Katzir, Wolf, & Poldrack, 2004). Word reading has also
148
been shown to utilise holistic processing as word reading and object identification to utilise nearby neural substrates (Lervag & Hulme, 2009) .
In dyslexic children and adults, RAN has been related to deficits in Working Memory
Capacity, Visual Working Memory and attention (Beidas et al., 2013; Nergard-Nilssen &
Hulme, 2014; Poll et al., 2013). Such skills are advanced in those with visual expertise in music sight reading (Hambrick et al., 2014; D. Hambrick et al., 2014) and WMC was seen to be superior for this group in Chapter 2 of this thesis. Conventional RAN testing in adults with text reading skills, that are normal or above, would be expected to show ceiling effects because it tests individual objects like a letter or number. A list of basic words, the order of which is randomised repeatedly and then read out loud as fast as possible should also be expected to demonstrate a ceiling effect. However, if superior visual processing skills are present - characterised by superior WMC or enhanced visual span - then expert music sight readers may perform faster in a rapid word identification task. In a similar fashion, correlations may even exist between a word speed reading task and music sight reading experience. Such a relationship has yet to be tested between musicians and non-musicians and expert and non-expert music sight readers.
The current study investigated the random word speed reading abilities of these groups and found random automatized word speed reading (WordRAN) to be a feature of expertise.
Hypothesis
That musicians will have enhanced WordRAN relative to non-musicians as will expert music sight readers because it is a measure of cognitive processing expertise.
149
Methods
Participants
Following the granting of ethics approval from the UNSW Human Research Ethics
Advisory Committee to perform the study, 34 participants were recruited from within the
UNSW student body and were reimbursed for their time: 20 were musicians and 14 had no music training and were classified as non-musicians. The musicians were then subdivided into 9 expert and 11 non-expert music sight readers according to this thesis protocol with 1 musician omitted (brass instrumentalist).
Stimulus
7 lines of 15, 2-4 letter words, each 16 point ‘Traditional Arabic’ font, were presented on a laminated card. The target was copied from a test page in the ‘Intuitive Overlays Test’
(Watkins, 2000), designed for testing reading speed using coloured overlays. The words used were: to, not, play, come, see, look, cat, you, and, for, up, the, my, is, dog. These appeared in random order across the 7 lines.
Procedure
The participants were instructed to read the words along each consecutive line aloud as fast as possible. The readings were recorded and the sound file opened using the
FleximusicTM Audio Editor. Time markings were placed at the beginning of the 2nd line and the end of the 6th line of words. These points were chosen as a reflection of ‘mid- stream’ reading without being affected by any anticipatory EMs at either end of the trial.
This time measure, in seconds, was then divided by the number of words read (75) giving
150
the Reading Rate (RR) in words per second. RRs were compared for musicians and non- musicians and expert and non-expert music sight readers.
Results
An unpaired t-tests was performed to determine if specific effects existed between musicians and non-musicians followed by a 1-way ANOVA comparing non-musicians, non-expert sight readers and expert sight readers Reading Rate (RR). Significance was assigned at the 0.05 level and the results were summarized in Figure 17. Musicians and
Non-Musicians (Figure 17a) and Non-Musicians, Non-Expert Sight Readers and Expert
Sight Readers (Figure 17b) were plotted against Reading Rate (words/second). The error bars signify one standard error of the mean (SEM).
Reading rate was significant between musicians (Mdn = 3.62) than for non-musicians
(Mdn = 3.09), U = 73, p=0.04. A significant effect for musicianship was also found: F
(2,29) = 9.933, p = 0.0005. It can be seen from Figure 17 that the expert music sight readers are responsible for this effect: t (29) = 4.071, p = 0.0005.
151
W o rd R A N W o rd R A N )
4 .5 ) 4 .5
c
c
e
e
s
s
/
/
s
s d
4 .0 d 4 .0
r
r
o
o
w
w
(
(
e
3 .5 e 3 .5
t
t
a
a
R
R
g
g n
3 .0 n 3 .0
i
i
d
d
a
a
e
e R 2 .5 R 2 .5 M u s ic ia n s N o n -M u s ic ia n s N o n - N o n -E x p e r t E x p e r t M u s ic ia n s S ig h t S ig h t F ig u re 1 7 a . F ig u re 1 7 b . R e a d e rs R e a d e rs
F i g u r e 1 7 : M u s ic ia n s a n d N o n -M u s ic ia n s (F ig u re 1 7 a ) a n d N o n -M u s ic ia n s , N o n -E x p e rt S ig h t R e a d e rs a n d E x p e rt S ig h t R e a d e rs (F ig u re 1 7 b ) p lo tte d a g a in s t R e a d in g R a te (w o rd s /s e c o n d ). E rro r b a rs s h o w S E M .
Discussion
The testing of the rapid automatized naming of words was designed to explore whether the speed associated with expert sight reading performance was specific to the musical domain only, or it is a more generalised processing expertise that transfers outside of the domain of processing expertise. On the basis that conventional RAN was correlated with
WMC (Beidas et al., 2013) and WMC with sight reading expertise (Hambrick et al.,
2014), the reading rate was expected to correlate with sight reading expertise. It was highly significant: p=0.0005. This indicated that it was a feature of general visual processing expertise as it had transferred to non-musical targets.
Of interest was the relationship between visual span and WordRAN. Reduced peripheral crowding for letters was evident in the non-expert sight reader group over the non- musicians as the weighted visual span was enhanced: p=0.001. But, WordRAN between non-musicians and non-expert sight readers was practically identical: p=0.96. This
152
suggests that, despite the enhanced visual spans of the non-expert group, speed of processing is still shown to be a unique feature of visual processing expertise and is likely linked with their superior WMC. The words were familiar to all, as were the letters in the visual span study, but expertise effects were found nonetheless.
The role of EMs was not investigated as the WordRAN test was administered. Under time pressure, there is a tendency for shorter duration fixations to occur (Gobet et al.,
2001; Underwood, Hubbard, et al., 1990) with shorter saccades and longer latencies
(Kowler & Anton, 1987) or cancellation of saccades leading to fewer fixations of longer duration (Yang & McConkie, 2001). It is not known how the EM patterns might differ between groups when performing the WordRAN task. Further testing, while measuring
EMs, may display different strategies employed under time pressure. This may expand the findings from the EM studies when the experts favoured saccade cancellation and the non-experts while the non-experts increased the number of saccades and showed shorted fixation durations.
153
CATEGORY Non-Musician Non-Expert Sight Expert Sight Reader Reader
SKILL Literacy Literacy + Music Literacy + Music Sight Sight Reading Reading
OUTCOME Text Reading Expert Text Reading Expert Text Reading Expert
FEATURES LSB DP DP LSB WMC Visual Span LSB Visual Span WordRAN
Table 7: Summary of Processing Features (Version 3) with the addition of WordRAN.
Conclusions
Surprisingly, WordRAN was found to be enhanced in musicians relative to non- musicians. Furthermore, WordRAN was statistically superior in expert music sight readers over both non-expert sight readers and non-musicians. As a feature of the enhanced WMC of experts, it has conferred a visual processing benefit to another domain.
Study 4: Frequency Discrimination and Modulation
Background
Frequency Discrimination (FD) is the ability to judge a difference in pitch between two static tones of different frequency. The lower the threshold of difference between the two
154
tones, the better the FD. Musicians have long been known to have superior FD skills compared to those of non-musicians (Banai & Yifat, 2011; Deguchia et al., 2012;
Micheyl et al., 2006; Oganian & Ahissar, 2012; Spiegel & Watson, 1984). Frequency
Modulation Discrimination (FM) is the ability to detect a change in pitch from a carrier tone over time. That is, does the pitch remain constant or does the pitch change or
‘modulate’ either up or down throughout the duration of the tone? The smaller the change in pitch that can be detected, the better the Frequency Modulation Discrimination.
Phonological skills are closely related to reading ability and dyslexia (Talcott et al.,
1999). In particular, FM (Boets et al., 2011; Talcott et al., 1999) as the shape of the sound envelope created by the speech sounds of consonants more closely resembles a tonal modulation (FM) with a slow attach (see Figure 18a), while the static tone change of FD, with a short ‘attack’ phase (see Figure 18b), resembles vowels (Horst, 1989; Talcott et al., 1999). This change in tone over time is linked to dyslexia because sufferers have deficiencies in the ability to rapidly process sensory information over time, that is, temporal processing (Amitay et al., 2002; Anvari et al., 2002; Atterbury, 1983).
Figure 18a.
155
Figure 18b.
Figure 18: Sample sound envelopes showing slow ‘attack’ phase of the sound – FM/vowels (Figure 18a) and showing fast attack – FD/consonants (Figure 18b).
Some researchers have suggested that FD and FM are representations of the same process. FM at lower frequencies may simply be FD judgements performed at the outer limits of the modulation range (Demany & Semal, 1989; Horst, 1989). If that is the case, it would be expected that musicians would outperform non-musicians in FM as they are known to do so in FD. In addition, a ceiling effect might be expected between expert and no-expert music sight readers unless a transfer of processing expertise occurs from the visual to the auditory domain. It is currently not known how expertise might modify FD and FM in this way.
Hypothesis
That FM and FD are features of expert visual processing expertise.
Methods
Participants
Following the granting of ethics approval from the UNSW Human Research Ethics
Advisory Committee to perform the study, 42 participants were recruited from within the
156
UNSW student body and were reimbursed for their time: 25 were musicians and 17 had no music training and were classified as non-musicians. The musicians were then subdivided into 10 expert and 14 non-expert music sight readers according to this thesis protocol with 1 musician omitted (brass instrumentalist) as in the previous studies.
Stimulus
The stimuli were pure sine tones generated by a custom written programme for
MATLABTM, powered by a Hewlett Packard ‘Elitebook 8470p’ PC (Intel Core i5
2.60GHz processor/8.00GB RAM/16-bit Operating System).
Procedure
Each participant was seated in an enclosed room without headphones. The FD trials were controlled by a MATLABTM programme, powered by a Hewlett Packard ‘Elitebook
8470p’ PC (Intel Core i5 2.60GHz processor/8.00GB RAM/16-bit Operating System). programme. The first of two tones was presented at a frequency of 1000Hz for 250 ms, followed 500 ms later by a tone of +/- 40Hz also sounded for 250 ms. Using a 2AFC procedure, each participant was to press ‘a’ on the computer keyboard if first tone was the higher of the two or ‘d’ if the second tone was the higher of the two. This order was randomly assigned across the trials. An adaptive 3-up/1-down staircase reversal procedure was utilised estimating to 79% correct as per Wetherill & Levitt (Wetherill,
1965). The comparison tone was then varied by an interval of 8Hz and halved after each reversal. The final value of a participant’s FD threshold was given as the average frequency difference of the final four reversals.
157
The FM task was conducted in a similar fashion to the FD task with two tones presented using the same 2AFC procedure and threshold calculation. Each of the two tones centred on a frequency of 1000Hz. One of the two tones was sinusoidally modulated - the participants to press ‘a’ if it was the modulated or ‘warped’ tone was presented first or ‘d’ if it was second. The initial modulation was +/- 20Hz, with an initially step of 4Hz that was then halved until the threshold reached and calculated in the same manner as the FD task.
Results
FD and FM thresholds were computed for each participant and the results analyzed using
Graphpad ‘Prism’TM. A Mann-Whitney test was used to determine the effects between musicians and non-musicians and a Kruskal-Wallis test to compare non-musicians and expert and non-expert music sight-readers. Such non-parametric tests were used because of the unequal variance exhibited by the groups. Frequency Modulation threshold (Hz) was plotted against Musicians vs Non-Musicians (Figure 19a), and Non-Musicians vs
Non-Expert Sight-Readers vs Sight-Readers (Figure 19b) for a 1000Hz tone.
Frequency Discrimination threshold (Hz) was plotted against Musicians vs Non-
Musicians (Figure 20a), and Non-Musicians vs Non-Expert Sight-Readers vs Sight-
Readers (Figure 20b) for a 1000Hz carrier tone. The error bars signify one standard error
158
of the mean (SEM).
F r e q u e n c y M o d u la tio n F r e q u e n c y M o d u la tio n 2 .0 2 .0
1 .5
1 .5 )
)
z
z
H
H
(
(
M
M F F 1 .0 1 .0
0 .5 0 .5 M u s ic ia n s N o n -M u s ic ia n s N o n - N o n -E x p e r t E x p e r t M u s ic ia n s S ig h t S ig h t F ig u re 1 9 a. R e a d e rs R e a d e rs F igu re 1 9 b.
F i g u r e 1 9 : F re q u e n c y M o d u la tio n (H z ) p lo tte d a g a in s t M u s ic ia n v s N o n -M u s ic ia n s (F ig u re 1 9 a ) a n d N o n -M u s ic ia n s v s N o n -E x p e rt S ig h t-R e a d e rs v s N o n -M u s ic ia n s (F ig u re 1 9 b ) fo r a 1 0 0 0 H z c a rrie r to n e . E rro r b a rs s h o w S E M .
F re q u e n c y D is c r im in a tio n F re q u e n c y D is c r im in a tio n
)
)
z
z H
H 4 0 4 0
(
(
n
n
o
o
i
i
t
t a
a 3 0 3 0
n
n
i
i
m
m
i
i
r
r c
c 2 0 2 0
s
s
i
i
D
D
y
y c
c 1 0 1 0
n
n
e
e
u
u
q
q e
e 0 0
r
r F F M u s ic ia n s N o n -M u s ic ia n s N o n - N o n - E x p e r t M u s ic ia n s E x p e r t S ig h t R e a d e rs F ig u r e 2 0 a . S ig h t R e a d e rs F ig u r e 2 0 b .
F i g u r e 2 0 : F re q u e n c y D is c rim in a tio n (H z ) p lo tte d a g a in s t M u s ic ia n s v s N o n -M u s ic ia n s (F ig u re 2 0 a ) a n d N o n -M u s ic ia n s v s N o n -E x p e rt S ig h t-R e a d e rs v s N o n -M u s ic ia n s (F ig u re 2 0 b ) fo r a 1 0 0 0 H z c a rrie r to n e . E rro r b a rs s h o w S E M .
Both the FD and the FM thresholds were significantly smaller for musicians than non- musicians. FM was significantly smaller for musicians (Mdn = 0.72) than non-musicians
(Mdn = 1.19), U = 93.50, p = 0.03. FD was significantly smaller for musicians (Mdn =
159
4.75) than non-musicians (Mdn = 8.50), U = 102, p = 0.03. That is, musicians were significantly more sensitive to static differences and modulations on pitch.
A Kruskal-Wallis test comparing experts, non-experts and non-musicians for FM was not significant for the effect of musicianship: 2 (35) = 3.265, p = 0.19 with a mean rank FM score of 21.57 for non-musicians, 18.38 for non-experts and 13.56 for experts. Multiple comparisons between each group also revealed no significant differences in FM ability.
However, a similar analysis for FD was significant for the effect of musicianship: 2 (37)
= 6.707, p = 0.03 with a mean rank FD score of 23.69 for non-musicians, 19.73 for non- experts and 11.72 for experts. Controlling for Type I error across tests by using the
Bonferroni correction, expert sight readers were found to have significantly finer FD ability compared with non-musicians: p = 0.03.
Discussion
This study confirmed the findings of other researchers that FD and FM capabilities are enhanced for musicians compared with non-musicians (Banai & Yifat, 2011; Deguchia et al., 2012; Micheyl et al., 2006; Oganian & Ahissar, 2012; Spiegel & Watson, 1984).
However, comparing the three groups revealed that the relationship between FD and expertise was not the same as FM and expertise at 1000Hz.
That is, expert and non-expert sight readers have similar levels of FD, but the experts are significantly better than the non-musicians. On the other hand, for FM, the three groups vary but no significant differences were found between one another. It might be that an expertise effect exists when using different carrier tones. This may be because FD and
160
FM are similar for normal hearers at 40 to 60Hz, but above this level, FD is more sensitive than FM (Formby, 1985). Another study found that the FM thresholds for non- musicians were approximately 200% greater than those of a musician for low and high frequencies (average values of 436 and 1743 Hz respectively). The threshold in the mid- range, (average of 897Hz) only increased by 40% (Spiegel & Watson, 1984). As such, testing within the mid-range, being close to the frequency used in this study, may demonstrate an expertise effect as FD but not high or low enough to reveal a significant effect for FM. This is suggested by the trends in Figure 19. Different carrier frequencies would need to be used to investigate this thoroughly.
It has been shown that FD can be improved with training and that improvement transfers to tones other than the training tone (Grimault et al., 2003; Micheyl et al., 2006). Despite differences in the frequency of tones tested and whether pure tones or harmonically complex tones were used, FM improvements were not measured in previous studies, only
FD. This is surprising considering that FM discrimination involves temporal processing and linked to reading difficulties (Ahissar et al., 2006; Amitay et al., 2002; Anvari et al.,
2002) and may have implications upon aspects of text reading remediation. For example, if FD can be improved in non-musicians to a level of musicians with just 14 hours of DP at 330Hz (Micheyl et al., 2006), and 1000Hz has been the carrier tone used to establish links with specific learned differences (Kidd et al., 2015), FD training at 1000Hz would be an ideal frequency to begin. Both FD and FM discrimination need to be tested over a wider range of carrier frequencies with comparisons made between expert and non-expert music sight readers and non-musicians with non-expert sight readers. Only then can it be shown if FM is a general feature of expertise or if it might be dependent upon the
161
frequency involved. Such a result might also be possible if FM is very sensitive to expertise and that the non-expert group contains experts – as discussed earlier in the thesis.
The Summary of Processing Features Table can be expanded to illustrate the position of
FD and FM threshold abilities in Table 8 below.
CATEGORY Non-Musician Non-Expert Sight Expert Sight Reader Reader
SKILL Literacy Literacy + Music Literacy + Music Sight Sight Reading Reading
OUTCOME Text Reading Expert Text Reading Expert Text Reading Expert
FEATURES LSB DP DP LSB WMC Visual Span LSB FD/FM @ 1000 Hz Visual Span WordRAN FD/FM @ 1000 Hz FM @ ? Hz
Table 8: Summary of Processing Features (Version 4) adding FD and FM characteristics.
Conclusion
At 1000Hz, FD and FM thresholds were significantly smaller for musicians than for non- musicians as expected and confirms existing findings. The results suggest that while FD is not influenced by expertise in music sight reading, the findings for FM are not as clear.
It is possible that FM is superior in expert sight readers, but that testing at 1000Hz may
162
not be sensitive enough to detect it or that some experts have ‘bled’ into the non-expert group.
The following study looks at whether expert sight readers have additional skills in visual processing in relation to non-musical stimuli.
Study 5: Visual Search of Musical and Non-Musical Targets
Thus far, Studies 1-4 of Chapter 4 have developed a picture of the cognitive processing capabilities of expert and non-expert music sight readers as they were defined in Chapter
3. Expert music sight readers not only have a number of superior processing skills relative to non-experts: WMC and WordRAN, but non-experts have superior Visual
Span, FD and FM thresholds @ 1000Hz relative to non-musicians. The final aspect of cognitive processing to be explored in this thesis is visual search.
Background
Simple visual search is the identification of a target among distractor targets and is thought to be the result of a low-level, feature ‘pop out’ effect (Woods et al., 2013). If the target features are very specific or recognisable, the search speed will increase (Redford et al., 2011). Conversely, search speed will decline with increased distractors. However, performance of a serial search – searching for a specific target among other visually similar targets for example - is thought to involve higher level functioning that involves spatial attention skills as well as working memory and reaches a ceiling effect in adults
(Woods et al., 2013).
163
The role of Working Memory has received consideration in the literature regarding general visual search characteristics (Anderson, Vogel, & Awh, 2013). It is considered to be the place where for short-term analysis of potential targets occurs. A connection between WMC and music sight reading expertise was found in an earlier part of this thesis. Therefore, WMC may correlate with skill in visual search of either music-like or non-musical targets.
It has also been shown that expertise in music results in a reduced crowding effect for musical targets (Wong & Gauthier, 2012). What is not yet known is whether having expertise in music sight-reading confers any advantage when performing a visual search task with musical or non-musical targets. Based in earlier Visual Span results, where musicianship had an advantage for a non-musical target, there may be an advantage in a non-musical target search over non-musicians but reach a ceiling between the two groups of sight readers. However, with EMs involved, expertise features such as those shown for
WordRAN may not only result in better performance for musicians generally, but for expert sight readers specifically.
An aspect of processing expertise that has been recently investigated involved exploring whether or not audition can interfere with language processing (Fiveash & Pammer,
2014). The researchers found that musicians were distracted by music in a language task, particularly the experts and specifically when the music comprised unexpected harmonic structures when compared with ‘normal’ harmonic structures. The music interference effect was less for non-experts and negligible for non-musicians. This represents interference across domains – words and music. What has yet to be explored is whether
164
interference in the same domain across different senses is moderated by visual processing expertise.
In the current study, musicians and non-musicians were recruited to perform a visual search task. The targets comprised non-musical targets (Landolt C) and musical-like targets pre-empted by either a beep tone, congruent tones or non-congruent tones. The results were analysed for speed and accuracy between musicians and non-musicians and between expert and non-expert music sight-readers.
Hypothesis
That musicians, generally, will have better visual search accuracy for a non-musical target and expertise will be present for musical targets. Congruent and non-congruent tones are expected to inhibit the performance of musicians but not non-musicians.
Methods
Participants
Following the granting of ethics approval from the UNSW Human Research Ethics
Advisory Committee to perform the study, participants were recruited from within the
UNSW student body and were reimbursed for their time. Each observer wore visual correction if appropriate and was capable of N5 resolution at a viewing distance of 60cm.
The level of sight reading expertise was assessed using the criteria established in this thesis. A total of 39 people participated in the study – 22 musicians and 17 non- musicians. The musicians were further subdivided for the sight-reading assessment into 9 expert and 12 non-expert sight-readers.
165
Stimulus
Music-like targets were designed to resemble musical notes. They consisted of 5-lines with a square located in either the 2nd space, on the middle line or in the 3rd space and were presented in groups of 3 (see Figure 21a). A pilot study found that meaningful results were not possible when there were more than 3 symbols for either the musicians or the non-musicians. Therefore, it was decided to keep the pattern more easily recognizable for the non-musicians simply consisting of a square below, on or above the centre line. In the interests of consistency, the Landolt C target consisted of figures with randomly assigned orientation - the gap facing up, down, left or right - also presented in groups of 3 (see Figure 21b).
Figure 21a: Musical target in C/B/A configuration.
Figure 21b: Landolt C target in Left/Up/Down configuration.
Figure 21: Visual Search Targets showing sample Musical Target (Figure 21a) and Landolt C Target (21b).
Each square in the music-like target and the gaps in the Landolt C target were designed to subtend the same angle at 60cm corresponding with Snellen letter size of 6/24 or 20/80.
166
Procedure
The images were generated using a custom written programme for MATLAB and presented on a linearized 27-inch Mitsubishi Diamond Pro monitor driven at a frame rate of 80Hz. driven by a Hewlett Packard ‘Elitebook 8470p’ PC (Intel Core i5 2.60GHz processor/8.00GB RAM/16-bit Operating System).
The participants were seated in front of the monitor at a chin and headrest assembly that was mounted on an instrument table. The table was set so that the viewing distance to the screen was 60cm. The participant’s height was carefully aligned using a canthus mark that was level with the centre of the computer screen. One second after a preparatory tone was sounded, the search target was presented in the centre of the screen for 100 milliseconds. (This time was chosen to ensure that the target was viewed in a single fixation with little chance of a saccadic EM being generated before the target disappeared.) This was immediately followed by the display of 8 possible matching targets presented on the screen in a circle of radius 7.75 cm for 4.5 seconds’ duration.
Participants were instructed to search these 8 targets for a possible match to the test target that had just been presented. A 2 alternative, forced choice method was employed – target present or target absent – with the instruction given to respond by pressing the corresponding computer key as quickly as possible (see Figure 22).
167
A B C
Target Present =
‘a’ Target Present = ‘d’
4.5 seconds 100 ms
Figure 22: Visual Search Experimental design. A) Search target presented on the computer screen for 100ms. B): 8 potential matching targets presented on screen for 4.5 seconds. C): 2AFT paradigm – press ‘a’ on the computer keyboard if the search target was present: ‘d’ if the target was absent.
The study consisted of 4 test conditions: Landolt C targets with a single beep warning tone, musical targets with single beep warning tone, musical targets with congruent tones and musical target with non-congruent tones. A ‘congruent tone’ referred to a sound- pattern that corresponds to the shape of the musical pattern. For example, the congruent tone corresponding to Figure 21a would be 3 descending note pitches. A ‘non-congruent’ tone would be 3 tones at the same pitch or 2 tones ascending. Each of the 4 test conditions had 40 trials comprising 20 target present and 20 target absent trials apiece.
Results
Non-Musical Target - Landolt C
The average time taken to complete the task and the accuracy (d’) of search performance in each of the 4 conditions was calculated and the results analyzed using Graphpad
168
‘Prism’TM. The measure d’, as defined by signal detection theory is a measure of the sensitivity of detecting a stimulus - hits, misses, false positives and false negatives were calculated. Due to the data not being normally distributed between the groups, Mann-
Whitney U and Kruskal-Wallis tests were chosen and performed to determine if specific effects existed for musicians/non-musicians and non-musicians/non-expert/expert sight readers respectively and the results summarized in Figure 23. Musicians/Non-Musicians
(Fig 23a) and Non-Musicians/Non-Expert/Expert sight readers (Fig 23b) were plotted against Time (seconds) and Musicians vs Non-Musicians (Fig 23c) and Non-
Musicians/Non-Expert/Expert sight readers (Fig 23d) were plotted against Accuracy (d’) when performing visual search. The error bars signify one standard error of the mean
(SEM).
The time taken to perform visual search of Landolt C targets was not significant different between musicians and non-musicians: U (34) = 0.4077, p = 0.69. Nor was an effect found between across non-musicians, non-expert or expert sight readers for time using the Kruskal-Wallis test: 2 (37) = 0.6969, p = 0.71 with a mean rank FM score of 21.12 for non-musicians, 18.58 for non-experts and 17.67 for experts.
Sensitivity, d’, was found to be significant between non-musicians and musicians: U (35)
= 2.653, p = 0.01 with musicians being more accurate than non-musicians in Landolt C visual search.
A post-hoc Kruskal-Wallis test comparing experts, non-experts and non-musicians for
Landolt C sensitivity (d’) was significant for the effect of musicianship: 2 (35) = 6.162, p
= 0.045 with a mean rank FM score of 14.30 for non-musicians, 18.67 for non-experts
169
and 25.28 for experts. Applying a Bonferroni correction, expert sight readers were found to be significantly more accurate than non-musicians in Landolt C search: p = 0.04.
Musical Targets
The average time taken to complete and the accuracy (d’) of performance in each of the 4 conditions were calculated as for the Landolt C experiment and the results analyzed using
Graphpad ‘Prism’TM. Mann-Whitney u and Kruskal-Wallis tests were separately performed to determine if specific effects existed for musicians/non-musicians and non- musicians/non-expert/expert sight readers respectively for each of the sound conditions and the results summarized in Figure 24 for Time and Figure 25 for d’. Musicians/Non-
Musicians were plotted against Time (seconds) for Beep only (Figure 24a), Congruent
Tones (Figure 24c) and Non-Congruent Tones (Figure 24e) when performing visual search. Non-Musicians/Non-Expert/Expert Sight readers were plotted against Time
(seconds) for Beep only (Figure 24b), Congruent Tones (Figure 24d) and Non-Congruent
Tones (Figure 24f) when performing visual search. No significant Time effects were found in any condition for any group.
Similarly, Musicians/Non-Musicians were plotted against Accuracy (d') for Beep only
(Figure 25a), Congruent Tones (Figure 30c) and Non-Congruent Tones (Figure 25e) when performing visual search. Non-Musicians/Non-Expert/Expert Sight Readers were plotted against Accuracy (d') for Beep only (Figure 25b), Congruent Tones (Figure 25d) and Non-Congruent Tones (Figure 25f) when performing visual search. The error bars signify one standard error of the mean (SEM).
170
L a n d o lt C _ T im e L a n d o lt C _ T im e
4 .0 4 .0
)
) s
s 3 .5 3 .5
c
c
e
e
s
s
(
(
e
e
m
m i
i 3 .0 3 .0
T T
2 .5 2 .5 M u s ic ia n s N o n -M u s ic ia n s N o n - N o n - E x p e r t M u s ic ia n s E x p e r t S ig h t S ig h t R e a d e rs R e a d e rs F ig u r e 2 3 a . F ig u r e 2 3 b .
L a n d o lt C _ d ' L a n d o lt C _ d '
1 .5 1 .5
)
)
'
'
d
d
(
(
1 .0 1 .0
y
y
t
t
i
i
v
v
i
i
t
t
i
i
s
s n
n 0 .5 0 .5
e
e
S S
0 .0 0 .0 M u s ic ia n s N o n -M u s ic ia n s N o n - N o n - E x p e r t M u s ic ia n s E x p e r t S ig h t
F ig u r e 2 3 c . S ig h t R e a d e rs R e a d e rs F ig u r e 2 3 d .
F ig u r e 2 3 : M u sician s/N o n -M u sician s (F ig 2 3 a) an d o n -M u sician s/N o n -E x p ert/E x p ert S ig h t R ead ers (F ig u re 2 3 b ) w ere p lo tted ag ain st T im e (seco n d s) an d M u sician s v s N o n -M u sician s (F ig u re 2 3 c) an d N o n -M u sician s/N o n -E x p ert/E x p ert S ig h t R ead ers (F ig u re 2 3 d ) w ere p lo tted ag ain st A ccu racy (d ') w h en p erfo rm in g v isu al searc h . E rro r b ars = S E M .
The results for time in all musical target conditions are reported together as follows:
Music – beep only
The time taken to perform visual search of musical target with a beep only between musicians and non-musicians: U (35) = 0.5670, p = 0.57. The effect between the musicianship groups using the Kruskal-Wallis test: 2 (30) = 1.971, p = 0.37 with a mean rank time score of 14.33 for non-musicians, 19.09 for non-experts and 14.25 for experts.
171
Music – congruent tones
The time taken between musicians and non-musicians: U (36) = 1.593, p = 0.12. The effect between the musicianship groups (Kruskal-Wallis test): 2 (36) = 3.740, p = 0.15 with a mean rank time score of 22.41 for non-musicians, 17.82 for non-experts and 14.00 for experts.
Music – non-congruent tones
The time taken between musicians and non-musicians: U (36) = 1.544, p = 0.13. The effect between the musicianship groups (Kruskal-Wallis test): 2 (30) = 2.403, p = 0.30 with a mean rank time score of 19.36 for non-musicians, 14.64 for non-experts and 13.56 for experts.
172
M u s ic _ B e e p _ T im e M u s ic _ B e e p _ T im e
3 .6 3 .6
3 .4 3 .4
)
)
s
s c
c 3 .2 3 .2
e
e
s
s
(
(
e
e 3 .0 3 .0
m
m
i
i T 2 .8 T 2 .8
2 .6 2 .6
M u s ic ia n s N o n -M u s ic ia n s N o n - N o n - E x p e r t M u s ic ia n s E x p e r t S ig h t F ig u re 2 4 a . S ig h t R e a d e rs R e a d e rs F ig u re 2 4 b .
M u s ic _ C o n g r u e n t T o n e s _ T im e M u s ic _ C o n g r u e n t T o n e s _ T im e
3 .6 3 .6
3 .4 3 .4
)
)
s
s c
c 3 .2 3 .2
e
e
s
s
(
(
e
e 3 .0 3 .0
m
m
i
i T T 2 .8 2 .8
2 .6 2 .6
M u s ic ia n s N o n -M u s ic ia n s N o n - N o n - E x p e r t M u s ic ia n s E x p e r t S ig h t F ig u re 2 4 c . S ig h t R e a d e rs R e a d e rs F ig u re 2 4 d .
M u s ic _ N o n -C o n g r u e n t T o n e s _ T im e M u s ic _ N o n -C o n g r u e n t T o n e s _ T im e
3 .6 3 .6
3 .4 3 .4
)
)
s
s c
c 3 .2 3 .2
e
e
s
s
(
(
e
3 .0 e 3 .0
m
m
i
i T 2 .8 T 2 .8
2 .6 2 .6
M u s ic ia n s N o n -M u s ic ia n s N o n - N o n - E x p e r t M u s ic ia n s E x p e r t S ig h t F ig u re 2 4 e . S ig h t R e a d e rs R e a d e rs F ig u re 2 4 f.
F ig u r e 2 4 : M u sician s/N o n -M u sician s w ere p lo tted ag ain st T im e (seco n d s) fo r B eep o n ly (F ig u re 2 4 a), C o n g ru en t T o n es (F ig u re 2 4 c) an d N o n -C o n g ru en t T o n es (F ig u re 2 4 e) w h en p erfo rm in g v isu al search . N o n -M u sician s/N o n -E x p ert/E x p ert S ig h t R ead ers w ere p lo tted ag ain st T im e (seco n d s) fo r B eep o n ly (F ig u re 2 4 b ), C o n g ru en t T o n es (F ig u re 2 4 d ) an d N o n -C o n g ru en t T o n es (F ig u re 2 4 f) w h e n p erfo rm in g v isu a l search . E rro r B ars = S E M .
173
The results for d’ in all musical target conditions are reported together as follows:
Music – beep only
Between musicians and non-musicians, d’ for the visual search of musical targets with a beep only was: U (36) = 2.774, p = 0.009. The effect between the musicianship groups using the Kruskal-Wallis test: 2 (36) = 11.71, p = 0.003 with a mean rank time score of
13.94 for non-musicians, 18.08 for non-experts and 29.22 for experts.
Music – congruent tones
Between musicians and non-musicians, d’ for the visual search of musical targets with congruent tones was: U (36) = 2.410, p = 0.02. The effect between the musicianship groups using the Kruskal-Wallis test: 2 (36) = 8.815, p = 0.01 with a mean rank time score of 15.32 for non-musicians, 17.18 for non-experts and 28.17 for experts.
Music – non-congruent tones
Between musicians and non-musicians, d’ for the visual search of musical targets with non-congruent tones was: U (36) = 2.811, p = 0.008. The effect between the musicianship groups using the Kruskal-Wallis test: 2 (36) = 10.23, p = 0.006 with a mean rank time score of 14.06 for non-musicians, 18.54 for non-experts and 28.39 for experts. As a
Kruskall-Wallis test (which is equivalent to a 1 way ANOVA) was employed and is a non-parametric analysis, no interaction can be determined. However, the trends in data regardless of expertise was similar across congruency conditions, whilst showing a main effect of expertise, suggests no interaction effect (see Figure 25).
174
M u s ic _ B e e p _ d ' M u s ic B e e p : d '
2 .5 2 .5
2 .0 2 .0
)
)
'
'
d
d
(
(
y
y 1 .5 1 .5
t
t
i
i
v
v
i
i
t
t
i
i s
s 1 .0 1 .0
n
n
e
e
S S 0 .5 0 .5
0 .0 0 .0 M u s ic ia n s N o n -M u s ic ia n s N o n - N o n - E x p e r t M u s ic ia n s E x p e r t S ig h t F ig u re 2 5 a . F ig u re 2 5 b . S ig h t R e a d e rs R e a d e rs
M u s ic _ C o n g ru e n t T o n e s _ d ' M u s ic _ C o n g ru e n t T o n e s _ d '
2 .5 2 .5
2 .0 2 .0
)
)
'
'
d
d
(
(
y
y 1 .5 1 .5
t
t
i
i
v
v
i
i
t
t
i
i s
s 1 .0 1 .0
n
n
e
e
S S 0 .5 0 .5
0 .0 0 .0 M u s ic ia n s N o n -M u s ic ia n s N o n - N o n - E x p e r t M u s ic ia n s E x p e r t S ig h t
F ig u re 2 5 c . F ig u re 2 5 d . S ig h t R e a d e rs R e a d e rs
M u s ic _ N o n -C o n g ru e n t T o n e s _ d ' M u s ic _ C o n g ru e n t T o n e s _ d '
2 .5 2 .5
2 .0 2 .0
)
)
'
'
d
d
(
(
y
y 1 .5 1 .5
t
t
i
i
v
v
i
i
t
t
i
i s
s 1 .0 1 .0
n
n
e
e
S S 0 .5 0 .5
0 .0 0 .0 M u s ic ia n s N o n -M u s ic ia n s N o n - N o n - E x p e r t M u s ic ia n s E x p e r t S ig h t F ig u re 2 5 e . F ig u re 2 5 f. S ig h t R e a d e rs R e a d e rs
F ig u r e 2 5 : M u sician s/N o n -M u sician s w e re p lo tted ag ain st A ccu ra cy (d ') fo r B eep o n ly (F ig u re 2 5 a), C o n g ru en t T o n es (F ig u re 2 5 c) an d N o n -C o n g ru en t T o n es (F ig u re 2 5 e) w h en p erfo rm in g v isu al search . N o n -M u sician s/N o n -E x p ert/E x p ert S ig h t R ea d ers w ere p lo tted ag ain st A ccu racy (d ') fo r B e ep o n ly (F ig u re 2 5 b ), C o n g ru en t T o n es (F ig u re 2 5 d ) an d N o n -C o n g ru en t T o n es (F ig u re 2 5 f) w h en p erfo rm in g v isu al se a rc h . E rro r B a rs = S E M .
175
V is u a l S e a rc h - A ll G ro u p s v s A ll C o n d itio n s
E x p e rt S ig h t R e a d e r s 2 .5 N o n -M u s ic ia n s N o n -E x p e rt S ig h t R e a d e r s
2 .0
)
'
d
(
y 1 .5
t
i
v
i
t i
s 1 .0
n
e S 0 .5
0 .0
C p s s lt e e e o e n n d B o o n T T a g g L n n o o C -C n o N F ig u r e 2 6 : L an d o lt C , B eep , C o n g ru en t T o n es an d N o n -C o n g ru en t T o n es p lo tted ag ain st S en sitiv ity (d ') fo r N o n -M u sician s, N o n -E x p ert an d E x p ert M u sic S ig h t R ead ers. E rro r b ars = S E M .
The between group comparison Mann-Whitney test was performed with the results presented in Table 9a below. Analysis of the effects of sound on visual search was carried out using a Kruskal-Wallis test and the results are shown in Figure 26 above. Multiple comparisons for within groups tests were performed and the resultant p-values presented in Table 9 below.
A significant effect for sound was found in all groups. Non-musicians: F (3, 60) = 18.82, p <0.0001; non-expert: F (3, 43) = 5.750, p = 0.002 and expert sight readers: F (3, 32) =
6.780, p = 0.001.
It can be seen from Table 9a that the significant difference between the non-musicians and the musicians is mainly due to expertise. Experts owed significantly superior sensitivity in all sound conditions when compared with the non-musicians.
176
Table 9a
Landolt C Beep Only Congruent Tones Non-Congruent Tones
Musician vs 0.01* 0.009** 0.02* 0.008** Non-Musician Expert vs Non- 0.46 0.06 0.07 0.11 Expert Expert vs Non- 0.04* 0.002** 0.01* 0.004** Musician Non-Expert vs 0.84 0.94 >0.9999 0.83 Non-Musician
Table 9b
NON-MUSICIAN Landolt C Beep Only Congruent Tones Non-Congruent Tones
Landolt C 0.62 <0.0001**** <0.0001****
Beep Only <0.0001**** <0.0001****
Congruent Tones 0.68
NON-EXPERT Landolt C Beep Only Congruent Tones Non-Congruent Tones
Landolt C 0.92 0.04* 0.03*
Beep Only 0.03* 0.01*
Congruent Tones 0.92
EXPERT Landolt C Beep Only Congruent Tones Non-Congruent Tones
Landolt C 0.78 0.005** 0.02*
Beep Only 0.02* 0.045*
Congruent Tones 0.78
Table 9: p-values for sensitivity in visual search across sound and expertise conditions. Table 9a – between groups comparison and Table 9b - within groups comparison.
177
In summary, the results suggest that musicians are significantly better at visual search for non-musical and musical targets. However, this effect appears to be the result of expertise. None of the groups displayed a difference in d’ between Landolt C and Beep or between congruent and non-congruent tones.
Discussion
Visual search using the Landolt C targets might be expected to serve as a control task when comparing the search capabilities of musicians and non-musicians using a non- musical target. It seems reasonable that all adults should perform at ceiling for such a task
(Redford et al., 2011) because it does not require any specific expertise to choose the orientation of a gap in a circle. This was not found to be the case in this study: musicians were significantly more accurate in the Landolt C task than non-musicians: p=0.01. This is in agreement with other findings that musicians are much more accurate with a non- musical task (Brochard, Dufour, & Despres, 2004). The current study did not find them to be significantly faster. This is because the task was designed specifically to look for sensitivity under time pressure rather than measure the time taken to get an accurate result. Nevertheless, the time differences were still analyzed should differences have become apparent even within these parameters.
Expertise in sight-reading, as defined by this thesis, was not significant when compared with the performance of non-experts for Landolt C visual search. However, it came close and may be a result of some experts being present in the non-expert group as discussed previously. Similar research has found accuracy in disease detection correlating with EM patterns of expert radiologists over less the experienced (Pietrzyk, McEntee, Evanoff,
178
Brennan, & Mello-Thoms, 2014) and EMs indicating expertise in visual processing has also been demonstrated in professional chess players (Bilalic, Langner, Erb, & Grodd,
2010). Unlike the current study, Bilalic’s study did not test for any difference between the expert and non-expert chess players on a non-chess related task which would have been instructive in the present context. Nevertheless, it is possible that a slightly less stringent definition of expertise may have resulted in significant differences between the experts and non-experts and would warrant reassessing with the modified criterion.
Considering music sight-reading as a sophisticated form of pattern recognition (Wolf,
1976) led to the hypothesis that musicians should outperform non-musicians in a visual search task that involved musical-like targets. The design of the target in this study (see
Figure 25), was such that it could be reliably interpreted by non-musicians as a pattern with a square in the space above, centered on or in the space below the middle line of a group of 5 parallel horizontal lines. The same pattern, however, had meaning for musicians in that the squares correspond to the locations of the notes C, B and A respectively on a musical stave in the treble clef. When music is being sight-read, it is essential that the musician be able to recognize where the note heads are on the stave without having to count the individual lines. Therefore, musicians might be expected to be faster and/or more accurate at searching for and matching music-like targets than non- musicians. Sight-reading experts might also demonstrate an expertise effect over the non- experts due to the reduced peripheral crowding effects for music notes that experts have been found to have (Wong & Gauthier, 2012) and as suggested by the results of the EM experiments in this thesis.
179
The ‘beep only’ condition - where there was no musical information involved in the preparatory tone other than a single beep immediately preceding the target presentation - resulted in no significant time differences for any condition. Not surprisingly, the musicians were significantly more accurate for searching a music-like target than the non-musicians: p=0.009 and the experts close to significance over non-experts: p=0.06.
However, it is noteworthy than none of the groups perform significantly differently between their d’ for Landolt C and d’ for music targets with beep only. This may be that each group performs the task at a particular level on first pass. Repeating the task on both
Landolt C and beep only or randomizing the presentations would enable this issue to be explored.
It was not known if the congruent or non-congruent tones might enhance, interfere with or be irrelevant to the visual search sensitivity of a music-like target within each subject group. Following the trials with the beep only tone, the participants were advised that, rather than a single beep tone, they would hear three tones that would either match or not match the relative direction of the squares on their target pattern for the next and subsequent group of trials respectively. All groups’ sensitivity increased significantly from the beep only condition to the congruent tones. The non-musicians increased markedly: p <0.0001, the non-experts: p = 0.03 and experts: p = 0.02. An explanation for this finding may be twofold. Firstly, there may be a learning effect involved, as all groups were performing their second trial of music-like target searches and all groups improved.
Secondly, this learning effect may be confounded by the congruent and incongruent tones as they appear to affect the musicians’ response, but not the non-musicians.
180
This may be manifestation of the so-called ‘Posner effect’ (Posner, 1980). It was found that when cuing a spatial location by sound, the reaction times to the appearance of an object in space of could be affected positively if the sound and location were compatible, or negatively if not compatible. This interference was the result of a drop in performance that corresponded to the time taken to process, reject and then reprocess a stimulus appropriately. In the current study, it could be argued that the musicians encounter a level of interference relative to the non-musicians when congruent or incongruent tones are presented because they are processing the music-like targets as music. On the other hand, the non-musicians are processing these targets simply as a target or pattern - the tones are irrelevant and afford no help or hindrance to them. Therefore, they improve in the congruent condition from the beep condition as a result of learning and appear to show little change in the non-congruent condition as ceiling may have been reached for the task.
On the other hand, learning appears to occur for the musicians, but the tones seem to interfere with the magnitude of their responses. For the experts, the beep/congruent tone effect is more sensitive: p = 0.02 while the non-experts’ is p = 0.03. For the beep/non- congruent condition, the experts’ sensitivity is significantly better: p = 0.045 and the non- experts: p – 0.01. One interpretation of this result is that, relative to their beep only condition, the experts show more enhancement by the congruent tones but also more interference by the non-congruent tones when compared with the non-experts.
These results suppose the recent findings of Fiveash and Pammer – who challenged semantic working memory by comparing word lists and music – across domain effects
181
(Fiveash & Pammer, 2014). They found that when performing a language memory task musicians were distracted by music playing in the background. Specifically, that the effect was heightened when unexpected notes were encountered by the expert group. The overall effect of music playing was less for non-experts and negligible for non-musicians.
This could arise as a challenge to WMC which might be more sensitive to interference.
That is, experts might only be capable of demonstrating expertise when all parameters are presented optimally. When challenged, whether by non-congruous tones in relation to visual search or unexpected shaping and notational complexities in the score, the rules are less able to be followed automatically. As a consequence, the ability to ‘chunk’ is compromised and ultimately accuracy and/or speed of performance can be affected.
Research into cross domain sensory influences is an evolving area and, like any other form of processing, expertise should be taken into account. The current study has suggested that certain cross-domain influences have occurred, but a more precise relationship between the sounds and expertise would be possible in further studies with the randomization of the conditions rather than individual blocks of stimulus types. In that was, the role that learning might play in the present context could be more decisively determined in relation to the enhancement or interference effects of the sounds.
Other research has indicated that training groups of letters in the periphery, known as perceptual learning, can improve letter recognition and reading speed (S. Chung, Legge,
& and Cheung, 2004). The musicians in the present study showed an increased ability to accurately search musical as well as non-musical targets when compared with non- musicians. This result is unexpected if it is assumed that any processing benefit derived
182
from the ability to read music is domain specific. Over time, musicians may be engaging in ‘peripheral learning’ related to musical symbols which leads to a reduction in crowding for these symbols in the periphery with the flow-on effect of increased accuracy in visual search for musical targets. Greater accuracy for visual search of musical targets would be expected and was found to be significantly so. However, recent research found that while expert musicians have reduced peripheral crowding effects for music stimuli, they did not have reduced crowding for non-musical targets (Wong &
Gauthier, 2012). This involved identifying targets in the peripheral visual field while fixation is held constant. The present study obviously used central vision in a serial visual search, but peripheral vision was required to direct central attention. If music-like targets were simply more easily recognized (Redford et al., 2011), it would explain musicians’ superior performance in the music-like search task, but not the Landolt C target unless cross-modal integration – where skills attained in one task influence another - has occurred (Meyer & Wuerger, 2001; Repp & Penel, 2002; Zimmer, Roberts, Harshbarger,
& Woldorff, 2010). The improved visual span capabilities of musicians for non-musical targets, as found earlier in this thesis, supports such a conclusion - that peripheral processing might generally be superior in musicians regardless of the target.
The failure of time to reach significance despite it being the hallmark of expertise is due to the design of the experiment. If the search task was simply the time taken to get the correct target present or target absent response, then the normal effect would have been found. However, the purpose of this experiment was to look for sensitivity under time pressure – the array of 8 targets from which to compare the test target was only on display for 4.5 seconds. Beyond that time, should the search not have been completed, the
183
sensitivity reverts to chance. Significant sensitivity and sensory enhancement or interference findings under such conditions are, therefore, a valid measure for analysis.
The summary of findings table can be upgraded to include the current findings in Table
10 below. Superior visual search for a non-musical target was found to be a feature of musical experience and a significant domain specific feature of sight reading expertise.
This final model will form the basis of the general discussion and conclusions in relation to EMs, cognitive performance and sight reading expertise.
CATEGORY Non-Musician Non-Expert Sight Expert Sight Reader Reader
SKILL Literacy Literacy + Music Literacy + Music Sight Sight Reading Reading
OUTCOME Text Reading Expert Text Reading Expert Text Reading Expert
FEATURES LSB DP DP LSB WMC Visual Span LSB FD/FM @ 1000 Hz Visual Span Visual Search WordRAN non-musical target FD/FM @ 1000 Hz FM @ ? Hz Visual Search non-musical target Visual Search domain specific target
Table 10: Summary of Processing Features (Version 5) including visual search results.
Conclusion
Musicians were found to have significantly enhanced visual search capabilities for non- musical as well as musical targets. This may be attributed to reduced crowding effects of
184
musicians for musical stimuli due to target specific perceptual learning. Expert music sight readers showed domain specific expertise for visual search of music-like targets over non-experts that was significantly affected by input from a congruent, non- congruent sound source. Non-musicians were unaffected by the tones when searching the musical targets and appeared to demonstrate visual task learning.
185
CHAPTER 6 General Discussion
The aim of this thesis has been to explore characteristics of piano sight reading expertise.
Deliberate care has been taken to refer to the experts as ‘expert music sight readers’ rather than ‘expert musicians’. Sight reading is a learned skill and is different from expertise in music composition, improvisation or ‘playing by ear’. This thesis has investigated expertise in the visual processing of notated music of the western art music idiom and has not sought to imply that sight reading experts are superior musicians in any way other than visual processing.
Defining the expert sight reader has traditionally been a somewhat fluid term, often based on self-reporting or assigned to individuals studying music at a tertiary level. The results of past EM studies have suggested that reading music is akin to reading text and that expertise is demonstrated by the experts’ ability to ‘chunk’ groups of notes together as a unit in order to improve their speed of processing. Sight reading experts are said to have either an innate ability or that they manifest expertise only after years of deliberate practice. The ability to identify a sight-reading expert by their performance of a particular level of sight reading assessment would allow researchers to easily be able to assign expertise and confidently proceed to compare their visual processing characteristics with those of non-expert sight readers and non-musicians. It then becomes a less complicated step to explore whether such visual processing expertise crosses domains to include non- musical targets and whether relationships might exist between the cross-modal integration of sight, sound and expertise. Being able to discriminate between the expert and the non-expert makes it possible to explore whether there are aspects of musical
186
experience, not just of expertise, that might confer a benefit towards enhancing visual processing characteristics considered necessary for successful text reading - like visual span for example.
The preceding experimental chapters explored features of visual processing expertise and how this is demonstrated in the music sight reading expert. Chapter 3 demonstrated that the three key features of expertise – extensive hours of deliberate practice, speed of performance and superior WMC – were each significantly present in those who could successfully perform a 6th grade AMEB sight reading test piece.
Having assigned expertise in this way, the EM experiments of Chapter 4 showed that expert and non-expert music sight readers reacted differently to blur and size manipulations of the music stimuli and that expert sight readers demonstrated characteristic of ‘chunking’. Experts and non-experts’ EM patterns were also different in relation to the challenges of increased speed and changes to the expected layout of the score. Chapter 5 explored a number of perpetual, cognitive and auditory processing skills by comparing expert music sight readers with non-musicians and non-expert sight readers. Experts, in addition to having demonstrated superior WMC, were also found to display expertise across domains by demonstrating visual processing expertise for non- musical targets.
Insofar as some questions have been answered regarding the nature of the expert sight reader, more intriguing questions are now raised regarding the nature of the non-expert.
This is because it was found that formal music training significantly related to some general visual processing advantages outside of the musical domain. That is, non-expert
187
music sight readers were found to be better than non-musicians for visual span, visual search for a non-musical target and FD/FM thresholds @ 1000 Hz. While the last point might be expected, the others are not.
Before accepting these findings as unequivocal evidence of the benefits of formal music education, assumptions basic to the methodology need to be more comprehensively scrutinized. The following section addresses these points, followed by implications for future research and the proposal that formal music education be recommended in text reading remediation.
Eye Movement Studies
Those individuals who could perform a 6th Grade AMEB sight reading assessment at or near perfection were flagged as potential visual processing experts. This claim was tested by measuring the experience of Deliberate Practice (DP), speed of performance and
Working Memory Capacity (WMC) found in those participants. These findings were significantly superior to those musicians unable to perform the sight-reading assessment and labelled as being the expert sight readers for the ensuing experiments. This does not, however, guarantee that some experts may be part of the non-expert group. Either the level may have been slightly too high or certain individuals may not have performed at their best due to performance anxiety or lack of recent practice on the piano. As a result, some findings of no significant effect between experts and non-experts or between non- experts and non-musicians may be misleading – particularly if a clear trend was evident.
Assessing sight reading at a variety of levels would be necessary to determine if expertise exists at a different grade level. Sample size might also be a factor. Future studies may
188
wish to increase the number of participants to further examine those trends that were not significant. However, based on the numbers of participants involved, a story of sight reading expertise does still emerge, the essence of which would not change.
Nevertheless, the participants who did achieve 6th Grade sight reading performance were visual processing experts for the purposes of this thesis. It is for this reason that, despite possible ‘leakage’ of experts into the non-expert group, the categorisation of expertise into these two groups is acceptable on the basis of measurable processing superiority. The results of this thesis suggest that these experts, as assigned, possessed significantly superior processing skills: speed of performance, WMC, years of deliberate practice, visual search for musical targets and the ability to read random lists of words significantly faster than the non-experts. Ultimately, while it is possible that experts were in the non-expert group, it is highly unlikely that there would be non-experts in the expert group. As such, a continuum of non-expertise would exist rather than a continuum of sight reading skill in general. A large variation in ability exists in the non-expert group whereas the experts are much more homogenous in their skills. However, the thesis aim was “to investigate unknown characteristics and unresolved traits of piano music sight reading expertise” (p43). As such, the thesis has had some measures of success. To look for how non-experts might differ amongst themselves, while an opportunity for further investigation, is beyond of the findings of this thesis and the current literature: that expertise is the marriage of WMC with DP. In addition, the researcher is not suggesting that all non-experts are terrible sight readers, or that they are not able to be improved.
However, the results suggest that unique processing superiority exists in the expert group as defined and that increasing DP will not attain it without the requisite WMC.
189
The blur and size levels used in this thesis were sufficient to elicit significant EM changes that corresponded with the literature on text reading. However, finer gradations between the N10 and N5 sizes as well as levels of blur greater than 3.2 would be required to produce more nuanced effects. Even so, the it was suggested that the CPS may be more of an expertise-dependent measure than a blanket measure of the functionality of the visual system.
The disruption to the spacing of music score was found to affect the saccadic latency in expert sight readers – an indication of processing pressure. Likewise, the 120MM metronome setting revealed processing stress in the non-expert music sight readers and demonstrated a strategy of increased EMs with shorter fixation durations to cope with an excessive speed demand – the speed/accuracy trade-off. Consequently, small percentage increments above the individual’s fastest, most accurate playing speed was suggested as an appropriate prescription for the use of the metronome for instrumental learning generally, not just for percussionists.
The accuracy of performance in the EM studies might be questioned as being important.
The subjects were requested to play ‘as quickly and as accurately as possible’. It has been shown that the speed chosen to ensure the accuracy of a typing task was approximately
20% below potential (K. A. Ericsson et al., 2007). This suggests that speed is controlled by each individual to accommodate the request for accuracy - the so-called
‘speed/accuracy trade-off’ (Cameron, 1995). The non-expert sight readers did perform slowly, but were, nevertheless, substantially accurate. The audio recordings of the test pieces confirmed that this is the case.
190
Additional justification for the selection of 6th Grade sight-reading assessment as the benchmark of expertise may be warranted. Recent papers dealing with expertise in relation to fluency of music reading (Wong & Gauthier, 2009,2012) and expertise in object recognition (McGugin, Richler, Herzmann, Speegle & Gauthier, 2012) use self- reporting as the basis from which to assess their respective expertise measuring tools. The
Vanderbilt Expertise Test (McGugin, Richler, Herzmann, Speegle, & Gauthier, 2012) tests for object expertise by having the subject judge if a test object is present or absent in a group of 3 similar objects. For example, cars, birds or planes. This test would be a worthwhile addition to the test battery for expertise in this thesis. However, it may not necessarily elicit object processing expertise as the responses are not forced to make judgements at speed. The authors claim to be able to measure domain-specific and general object recognition skills. This is after stating, on p16, that “……self-reports of expertise/interest may be inadequate predictors Additional justification for the selection of 6th Grade sight reading assessment as the benchmark of expertise may be warranted.
Recent papers dealing with expertise in relation to fluency of music reading (Wong &
Gauthier, 2009, 2012) and expertise in object recognition (McGugin, Richler, et al.,
2012) use self-reporting of perceptual expertise” (McGugin, Richler, et al., 2012).
Furthermore, on p11, the selection of objects used in testing was termed ‘arbitrary’. It was from this point that the researchers performed their testing on self-reported experts.
In this thesis, the selection of 6th Grade was not arbitrary. A recommendation of this level has been reported in the literature (A. J. Waters, Townsend, E., & Underwood, G. ,
1998) and it was from this point that expertise was assessed. Regarding the fluency of music reading test, this was determined on the basis of detecting which of 2 presentations
191
of 4 musical notes had 1 note changed from an original presentation (Wong & Gauthier,
2012). This test could have been performed as pattern matching task by a non-musician; as was the musical search target used in this thesis. Choosing not to use these tests was made on the basis of domain specificity. The AMEB assessments are graded, standardised tests of sight reading ability and are, therefore, very specific to the task being investigated. The other tests of expertise would certainly be worthwhile comparisons of the method chosen. Such future comparisons would provide further validation of each of these tests of expertise and further eliminate the bias of self- reporting expertise.
The criteria of ‘perfect or near to perfect performance’ was subjectively judged by the researcher and was not recorded or audited by an independent observer. As such, a level of bias may have been a factor. However, the researcher observed that the subjects were either very good or perfect at this standard or they failed to come even close to making musical sense of the piece; often giving up after a few bars into their attempt. Also, the subjects were not permitted to preview the test piece, but play by sight without any pre- planning or pre-rehearsal. As a result, the 6th Grade AMEB level was considered an acceptable level to benchmark sight reading expertise. In each case, the experimentation was undertaken before the sight-reading assessment was made. This ensured that any anxiety that might arise from having failed to perform the expertise judgement piece was minimised. The participants were aware that sight reading ability was the basis for analysis. The following wording appeared on the recruitment poster (see Appendix 1).
‘Our aim is to see what effect this has on your eye movements. There will also be a short questionnaire about your past and present musical life and a check of your sight-reading
192
capacity.’ It may, therefore, be of concern that the test results might be influenced by motivational factors as well as anxiety. For example, performance has long been known to be affected consciously or subconsciously, according to an individual’s self-interest in the outcome or perception of self (Sears, 1943). In the same way, expectation of failure or success in a task based on past successes and failures can influence performance
(Gebhard, 1949). Performing the sight-reading assessment after all testing procedures helps to minimise this issue also. Participants were not aware of the group they were assigned to while performing the experiments.
While these factors may have been an influence in the EM studies, the subjects involved in the cognitive tests were a mixture of musicians and non-musicians performing predominantly non-musical tasks. The non-musicians knew they would be tested on non- musical targets and not on musical targets (see Recruitment Poster, Appendix 3). These non-musical tasks – WordRAN, visual and perceptual span for letters and Landolt C visual search – were expected to act as control tasks. It was not expected that there would be a difference would be found between groups. If anything, the non-musicians might be motivated to outperform the musicians in non-musical tasks. The same could be said for
WMC or WordRAN utilising numbers and words respectively. However, significant differences were found when the successful achievement of 6th Grade sight reading assessments was used as the point of demarcation. The aim of the thesis was to find an objective measure at which point a degree of confidence can be placed when labelling a particular musician as a sight-reading expert based on visual processing. The results suggest that if an individual can perform at this standard, they can be categorised as a visual processing expert. From such a perspective, the thesis has achieved that aim.
193
Comparing the visual angle of a musical note with that of a letter was problematic. While the size of the note head can be approximated to that of a letter, a note is sometimes a solid object and at other times an outline, while the letters are always an outline or even a single line. One sits on a line and meaning is taken from the angle or shape of lines or other shapes attached to it; the other draws meaning form its spatial location on a stave with the attachments (beams) sometimes adding meaning and sometimes not. How these differences impacted upon visual processing is beyond the scope of this thesis.
Nevertheless, observable and often significant trends were revealed that address the nature of EMs being investigated in this thesis. For example, chunking skills were demonstrated by sight reading experts and are consistent with the literature. The N10 and
N5 size levels chosen were sufficient to elicit several significant responses in the non- expert group - increased fixation durations when processing became difficult, saccade latency changes and evidence of smaller visual span and increased crowding effects for visual targets in peripheral vision relative to the experts. However, the amount of blur used for testing was insufficient to elicit marked effects. Using more significant levels of blur would be required to examine further differences between the two groups. The same can be said for the possibly insufficient complexity of the disrupted score or using only a single speed to test metronome effects. Nevertheless, sufficient differences were revealed between expert and non-expert sight readers to suggest that score reading and text reading share many processing characteristics and establish a point from which to start more detailed observations.
Regardless of the experimental condition imposed upon the participants, expert music sight readers consistently performed significantly faster. Speed is the hallmark of the
194
expert (Bilalic et al., 2011; K.A. Ericsson et al., 1993; Farrington-Darby & Wilson, 2006;
I. Gauthier & Bukach, 2007) and the results suggested that those in the expert sight reader category in this thesis were indeed experts using this definition. Future research projects that wish to include only those musicians with visual expertise, using the method described in this thesis, can be employed with confidence. As previously stated, what remains uncertain is whether or not some expert sight readers may have been assigned to the non-expert group and this could account for the failure of some of the experiments to reach significance despite a clear trend being evident.
Cognitive Processing
Given that some experts may be present in the non-expert group, the finding that expert sight readers performed significantly better than non-experts in WordRAN is of great importance - experts were faster than non-experts in a non-musical domain: p=0.0005. It is necessary to further validate this claim in another area of expertise. For example, chess players would be an ideal group as they have been known to show many of the same features of expertise as musicians, such as WMC (Campitelli & Gobet, 2011; D.
Hambrick et al., 2014; Ruthsatz & Urbach, 2012). If expertise effects were to be found in chess ranked chess players for WordRAN, WordRAN could be developed as a screening tool for vocational selection where visual processing expertise might be a requirement – for example, air traffic controllers or aerial surveillance experts in the military.
FD @ 1000Hz and visual search accuracy for musical targets were reliable means of discriminating expertise between musicians. FM @ 1000Hz, visual span and visual search (Landolt C) were more difficult to accurately categorise for expertise but were
195
significant between non-musicians and non-expert sight readers. Finding other significant differences between experts and non-experts may be subject to the carrier frequency being tested. Therefore, firm conclusions can only be reached if further testing is done over a wider range of carrier frequencies. This may also be unclear due to the possible
‘leakage’ of experts into the non-expert group.
The method for testing visual spans in this thesis may be influenced by reduced peripheral crowding for letters as it was found to be significantly broader than non- musicians at the non-expert level. This led to the hypothesis that there might be an advantage to learning to read music in that non-experts have a significantly enhanced visual span and will be discussed in greater detail in the following sections. However, a thought-provoking finding was that LSB was found to be not significant for musicians at any level. This was using the same data as the visual span experiment. It appears that all participants use holistic processing for letters and it is not related to visual processing expertise for music sight readers or music experience generally. This conforms to the conclusions of other researchers (Hsiao & Cottrell, 2009; Tso et al., 2014). However, the key difference between the studies is that the present study does not involve targets from the domain of expertise. The LSB characteristics would need to be tested using musical targets to more exactly replicate those findings. Nevertheless, the current results suggest
LSB for letter detection across all participants regardless of musical involvement.
Musicians were found to have superior visual search skills for non-musical targets. This may be another manifestation of the enhanced visual spans present in those with formal music education, not just the processing experts. Expert sight readers had further
196
enhanced visual search for musical targets further heightened by congruent auditory input.
In summary, despite the possibility of the non-expert group containing some experts, the cognitive experiments in this thesis revealed that WordRAN was related significantly to sight reading expertise, as was WMC and the speed of performance. These experts showed associated EM patterns akin to those known found in text reading experts and demonstrated related expertise in features of sound perception and the interaction of sound with vision in ways quite different to non-experts and non-musicians. Therefore, for the purposes of the ensuing discussion, the unpopular proposal is being made that visual processing expertise, in this case music sight reading expertise, is something that might only occur in an individual that is born with superior WMC and has the opportunity and sufficient interest to engage in the requisite amount of DP. It is not being asserted that good sight reading cannot be attained with practice, but that ‘expertise’ as it relates to the psychophysics of input and response is not possible without WMC and DP together. All study participants in all of the studies were either postgraduate or undergraduate university students with above average intelligence by definition. Non- musician, expert and non-expert groups all showed LSB when processing groups of random letters. However, despite being well matched for intelligence, of specific interest was the difference in visual span found between musicians and non-musicians. It is to be emphasised that this was measured using a non-musical target. This was a musician effect; not related to expertise and invites additional scrutiny.
197
Other findings of note were superior FD and FM thresholds @ 1000 Hz and visual search for non-musical targets. When it is considered that a reduced visual span is characteristic of a novice reader (G. E. Legge et al., 1997; Risse, 2014; Underwood, Hubbard, et al.,
1990), that FM thresholds in particular are deduced in those with reading difficulties
(Boets et al., 2011; Talcott et al., 1999), that FM may be simply a less accurate measure of FD (Demany & Semal, 1989; Horst, 1989) and that FD is trainable with improvement transferring to other tones (Grimault et al., 2003; Micheyl et al., 2006), it is reasonable to propose two possible reasons for this finding. There is either a processing benefit to be obtained by formal music training of this kind or that there is another innate factor, ’X’, that is the pre-requisite for a given amount of DP in music to develop these enhanced processing skills for non-musical targets.
Solving for ‘X’
Solving for X, the entity that allows musical experience, not expertise, to confer visual processing advantage, is relevant when suggesting that training in one domain might cause an improvement in another. Should ‘X’ be found to exist however, formal music training would be of no benefit from the point of view of improving performance in another area. In fact, some never achieve expertise despite the accumulation of many hours of DP (Hambrick et al., 2014) and that text and music reading difficulties are known to frequently co-exist (Gaser & Schlaug, 2003). Therefore, can more generalised improvements in cognitive processing be made with DP in music or is the ability to benefit from DP in music the result of another innate ability?
198
The existence or otherwise of this factor re-ignites decades of controversy over music and intelligence – the so called ‘Mozart Effect’. The Mozart effect is the short term improvement in spatial-temporal reasoning, and not spatial only, after 10 minutes of listening to Mozart’s Sonata for Two Pianos in D Major (K448) (F. Rauscher & Shaw,
1998). Spatial-temporal reasoning is being able to mentally manipulate objects without a visual representation of physical prototype. It was claimed that listening to the Mozart piece caused the excitation of certain cortical firing patterns that share a similar ‘neural coding’ to that of spatial-temporal tasks. However, this response can also be elicited using other styles of music exhibiting structure and predictable form as opposed to ambient and minimalist music, for example. Consideration of the musical training of the subject, their age and expertise in relation to the test task, as well as musical preferences
(F. Rauscher & Shaw, 1998), should be factored in before making such conclusions.
This effect has been extended by some music educators to encapsulate the entire ‘music makes you smarter’ claim. Some insist that music only makes you smarter at music, saying that when the experiments have been repeated, similar improvements were evident in groups who were exposed to silence (Demerest & Morrison, 2000). Perhaps music study simply helps create better learners. There is evidence of higher academic performance as a result of attending acting or musical appreciation classes, not just music performance lessons (Demerest & Morrison, 2000).
Further attempts to reinforce the ‘music makes you smarter’ argument have surfaced again recently (Schellenberg, 2006). This research specifically looks at intellectual development when related to having taken music lessons, rather than just the heightened
199
functioning that listening to music can evoke, and whether these skills are lasting and/or transferable to other areas. Schellenberg found a small, positive correlation with general intelligence after long term music lessons but not with social interaction measures, so it seemed to be specific to IQ. It was also found that playing regularly was a predictor of lasting higher academic achievement, but the results were not subject-area specific and the effect was stronger in early childhood than later. The author made no attempt to explain the cause of this relationship and it was unclear what type of music lessons were involved. This study did not control for any type of ‘schooling effect’ that may have occurred or it may just be that smarter children take music lessons because they like the intellectual stimulation. Irrespective, the key premise of the ‘Mozart Effect’ is that it involves listening to music rather than reading it. However, perhaps there is a subconscious benefit in seeing and reading music that leads to a general improvement in seeing and reading other objects over time. Such an effect might involve visual perceptual learning.
Visual Perceptual Learning (VPL), also known as statistical learning, is the training of a specific sensory output in terms of discriminating differences, such as orientation, position or texture, and exists in the auditory domain with the training of Frequency
Discrimination (Tsodyks & Gilbert, 2004). VPL occurs when specific neurons are tuned due to repeated exposure to a particular stimulus and occurs subconsciously over time. Of specific interest to the current discussion is whether or not learning can transfer from one task to another or occur spontaneously and be a source of improving perception in normal populations by targeted training (Lu, Huab, Huanga, Zhouc, & Dosher, 2011). That is, could the constant presence of the lines of the music staff in the periphery of the music
200
reader’s field of vision be subconsciously be learned and then ignored. As a consequence, the variable features of the music – the notes – would become more discernible. Over time, this ability could be transferred to other targets – letters – and manifest as a broader visual span. Research has shown that subjects learned and improved performance in relation to details in the background of a visual presentation while being trained on targets in the foreground (Fiser & Aslin, 2001). This indicates that subconscious VPL had taken place. Another study showed that simply being exposed to a target enhanced learning in a related task some time after the original exposure (Folstein, Gauthier, &
Palmeri, 2010). The role of VPL has been discussed by other researchers, albeit in a less than sympathetic attitude as if it may be too simple an explanation at the same time acknowledging that others support this view (Wong & Gauthier, 2012). The suggestion that a learning mechanism is involved is also suggest by McGugin et al., 2012 on p21 by referring to an “underlying potential that is domain-general but which becomes expressed in domain-specific skills through experience” (McGugin, Richler, et al., 2012).
Two types of VPL have been proposed: Relevant VPL (R-VPL) and Irrelevant VPL (I-
VPL). R-VPL is thought to occur in a number of stages. The early stage causes changes to V1 and is specific to location, orientation and eye with no feedback reinforcement being required. The mid-stage, V2, V3, V4, V5 (MT), is less specific for orientation and location. Late stage involves higher level cognition and attention. I-VPL is when enhancement occurs for a background feature while being trained on the foreground feature. For example, background motion detection improved when a subject was being trained on a letter recognition task (T. Watanabe & Sasaki, 2015). This indicates that attention is not required and learning can occur with only exposure to an object. As
201
suggested, perhaps exposure to the parallel lines of the stave (background) is sufficient to enhance learning of position of the notes (foreground). Interestingly, it has been shown that peripheral crowding effects for identification of gabor patch orientation was reduced when the target gabor was flanked by gabors of the same orientation arranged vertically, one above the other (Maniglia et al., 2011). Perhaps the horizontal lines of a music stave may be behaving in this way – musical note perception may be subconsciously enhanced by sets of horizontal lines above and below the stave being viewed. This is may be feasible as training groups of letters in the peripheral visual field has been shown to improve general letter recognition and reading speed (S. Chung et al., 2004).
The changes in perception as a result of VPL are thought to involve the transfer of information from the brain’s right hemisphere to the left via the corpus callosum. A recent study showed this association in both behavioural and functional fMRI studies
(Roser, Fiser, Aslin, & Gazzaniga, 2011). Roser et al. proposed that the left side of the brain is more responsible for hypothesis creation and conceptual knowledge. The lower order/right brain sends its information from the statistical learning of basic shape relationships to the higher order/left brain where the visual material can be ‘chunked’.
New ‘rules’ are created over time as explicit domain learning occurs so that increasingly more complex objects can be interpreted quickly. Therefore, the transfer of information between the hemispheres of the brain, via the corpus callosum, occurs in higher learning.
Recalling that holistic processing and LSB are associated with right hemisphere processing and/or FFA (I. Gauthier et al., 2014; Hsiao & Cottrell, 2009; Tso et al., 2014), is it a coincidence that I-VPL is also associated with right hemisphere processing – the origins of statistical learning according to Roser et al.? Might this be the mechanism
202
responsible for musicians’ reduced crowding for objects in the periphery? Is it also a coincidence that musicians have a larger corpus callosum than non-musicians, with greater myelination of nerve fibres increasing the speed of interhemispheric communication (Schlaug, Janke, Huang, Staiger, & Steinmetz, 1995)? It is known that musicians are not born that way. Children with an interest in learning music have equivalent relevant brain structures when compared with those who do not and when neither group has had any musical training. This indicates that these structures develop over time with music training (A. Norton et al., 2005). Plasticity, or the ability for the brain to change, may only be relevant in an immature brain rather than in an adult, but there is a difference nonetheless in the brain areas relevant to reading (Gaser & Schlaug,
2003). The question arises whether the same cortical areas are involved in processing text and music. While adults have distinct neural areas for language and music processing, it is not clear whether or not this is the case with a child as increased modularly may occur over time (McMullen & Saffran, 2004).
If VPL is not found to be the mechanism whereby musicians develop reduced peripheral confusion and wider visual spans, then perhaps another innate factor needs to be identified – ‘X’. That is, an individual needs to possess ‘X’ in order to engage in the DP of sight-reading, just as the innate factor of superior WMC is required for the development of expertise (Grabner, 2014; Meinz & Hambrick, 2010). A full scale assessment of the IQs of non-musicians compared with those of the non-expert music sight readers might reveal ‘X’. The existence of ‘X’ could be possible as the non- musicians in this study - university undergraduates, postgraduates and academic staff - did not display a broader visual span for random letters despite having high intelligence.
203
It was found that the non-musicians in this study had holistic processing characteristics according to LSB ratio. Therefore, if ‘X’ exists, it is unlikely that deliberate formal music education has any part to play in the remediation of text reading. On the other hand, if
‘X’ doesn’t exist, then definitive causal relationships between music training and improvements in IQ would no doubt have already been fully exploited by those most interested in doing so (Schellenberg, 2006). Perhaps ‘X’ exists but, unlike WMC, is trainable. A possible candidate could be fluid intelligence: Gf. It has been said to be fixed and related to working memory, but recent work has shown that Gf may be trainable and transferable across domains (Jaeggi, Buschkuehl, Jonides, & Perrig, 2008). Research into the benefits of Gf and learning is an exciting area of future investigation.
Locating ‘X’ earlier in the learning hierarchy may explain the claim that reading dyslexia and music reading dyslexia have similar processing components and commonly co-exist in the same individual (Gaser & Schlaug, 2003). The reason for this not being seen frequently could be that those taking formal music lessons and struggle to succeed with music sight reading are free to discontinue lessons or change to a different mode of music instruction. A child is not free to leave school and discontinue trying to learn to read.
Should this be the location of ‘X’, any speculation of a causative relationship between music education and improvement in IQ or other areas of academic performance may not be legitimate. It does not rule out what effects might be found with targeted music VPL at an early age. This is because it is not clear whether the distinct neural areas for language and music processing that are found in adults are the same in children as it might develop over time (McMullen & Saffran, 2004). The concept of brain plasticity permits
204
speculation that targeted training might change the brain in such a way that expertise and/or holistic processing might arise and manifest in the activation of the FFA.
The FFA, as the name implies, has to do with expertise in facial recognition. Some researchers argue for it being an innate facial recognition system (McKone et al., 2012).
Others claim that it is a more generalised area of object recognition as it is difficult to detect differences in high resolution fMRI brain responses to faces or objects of expertise
(I. Gauthier et al., 2014; I. Gauthier & Tarr, 2002; Wong & Gauthier, 2009). Whether faces are the only objects to be processed ‘holistically’ is not clear. According to Tanaka and Farrah, the holistic nature of object processing may be at one end of a processing continuum, with ‘featural’ processing at the other (Tanaka & Farah, 1993). They define holistic processing as occurring when an object is interpreted as a single representation
‘without an internal part structure’. On the other hand, featural processing utilises the individual parts of an object for interpretation. Their experiments showed that when a facial feature was altered, it was identified more accurately as having been changed when presented within a whole face than when presented on its own. This was not the case for faces that had been either inverted or scrambled or for an object such as a house.
The development of expertise in object processing is thought to involve a critical period of development where stimulation of neurons is required in order to achieve function
(Sengpiel, 2007). Specific neurons become tuned to the features of a particular object or
‘exemplar’. Ongoing exposure to the exemplar causes these neurons to become less sensitive to its presence (Wallis, 2013). This results in ‘perceptual narrowing’ as the neurons become fine-tuned to the habitual stimuli (Kelly et al., 2009). Kelly et al. found
205
that perceptual narrowing had occurred in human infants around 3-9 months enabling reliable recognition of same-race faces.
According to Wallis, expertise for the recognition of non-face objects that activate the
FFA also involves the process of perceptual narrowing (Wallis, 2013). Wallis asserts that this type of learning will result in holistic processing - only less commonly encountered features will cause neurons to respond rather than the ones habituated to a particular object. That is, if neurons are highly adapted, whether to a face or an object, only a difference will result in a change in the neurons’ pattern of activity. This relates to the findings of Tanaka and Farrah when a change in the shape of a nose, for example, was more noticeable in the context of a full face, i.e. a difference was detected, than when presented in isolation. This may explain the expert sight readers’ EM response when encountering unexpected spacing in the score. If expert sight-readers process the score more towards the holistic end of the continuum, then the disruption to the score may have caused a change in the firing pattern of the neurons because the appearance of the
‘exemplar’ had changed. This may in turn be responsible for the increase in saccadic latency that was found as increased saccadic latency is a known feature of visual uncertainty (Cameron, 1995).
Wallis also claims that holistic processing is a function of expertise. While this may help to explain the response of the expert sight readers to visually disrupted score, what other evidence has been found to suggest holistic processing of objects other than faces?
Research into the magnitude of the N170 during object processing has produced some noteworthy results. The N170 is an event related potential that emanates from brain areas
206
associated with face processing (Bentin, Truitt, Puce, Perez, & McCarthy, 1996). Its delay and increase in magnitude being indicative of face processing. A recent study found that when expertise training was given for a specific object, the magnitude of the N170 was reduced if a face was presented at the same time as the object (Rossion, Kung, &
Tarr, 2004). These findings suggest processing competition between faces and objects and further strengthen the view that facial recognition might be the earliest specialisation of the object recognition system.
Other research has showed that the N170 was delayed and enhanced for inverted faces only and not for objects (Rossion et al., 1999). However, more recent research has found the N170 to be delayed and enhanced for a non-face object of expertise: inverted fingerprints. When presented to fingerprint experts, inverted fingerprints produced a delayed and increased N170 (Busey & Vanderkolk, 2005).
Therefore, face and expert object processing may be indistinguishable in the FFA because both have undergone similar development due to perceptual narrowing. The exact nature of this process remains elusive and not without controversy. The links between expertise, perceptual learning and holistic processing will no doubt provide a fascinating and fruitful area of future research. Music sight reading experts, now able to be objectively identified, are an easily accessible group of processing experts for such purposes.
Conclusions
Successful performance of a 6th grade sight reading assessment task appeared to be related to the benchmarks of processing expertise – speed of performance, superior WMC
207
and extensive DP. EM studies demonstrated a number of differences between the so- defined expert and non-expert sight readers that corresponded to known features of EM patterns with text reading experts. The cognitive processing experiments permitted comparisons not only between musicians and non-musicians, but between expert and non-expert sight readers. Without the delineation of expertise in this way, differences in processing that might be found between musicians and non-musicians might be thought to be due to musicianship alone when in fact they were due to processing expertise.
WordRAN and visual search involving a visual target being pertinent examples. As a result, expertise was proposed to be the result of an innate potential – WMC – that is developed and manifested by the addition of appropriate DP.
Of greater interest were the non-musical processing enhancements found in the non- expert sight readers when compared with the non-musicians. The question of being able to use targeted perceptual learning, perhaps including the use of music-like stimuli as a means of improving peripheral crowding effects and its relationship with text reading remediation was raised. This is the area to which research interests are now focused along with the search for the innate factor, ‘X’. However, should ‘X’ exist and be found to be untrainable, remediation of this type becomes a moot point.
208
APPENDICES
APPENDIX 1 MUSICAL EXPERIENCES SURVEY
HREA Approval #13140 Below is a short questionnaire about your musical life: past and present. 1. At what age did your formal instrumental training commence? <5 years 5-7 years 8-10 years 10-12 years >12 years
2. How many years of formal training have you undertaken? <2 years 2-5 years 5-8 years 8-10 years >10 years 3. What grade of Practical have you attained? (AMEB or equivalent)
4. What level of Tertiary Music study have you obtained? Certificate Diploma Undergraduate Degree Postgraduate Degree
5. What grade of Theory or Musicianship have you attained?
6. Do you play any other musical instruments? Yes No
7. Have you group or ensemble playing experience? Yes No
8. I can improvise on my instrument. Yes, I love it Yes, but only if I have to No, I avoid it Never tried
9. I can read/write another language. Yes, fluently Yes, but only if I have to No, not at all
10. How frequently do you play at the moment? 11. How frequently do you sight read? 209 APPENDIX 2 RECRUITMENT POSTER 1 HREA Approval #: 13140 Volunteers for Eye Movement Research Wanted!! IF YOU CAN PLAY THIS ON THE ……. THEN WANT 2 2 !! If you can sight read and play that melody on a keyboard, then I want to talk to you! If you can sight read and play that melody on a keyboard, then I want to talk to you! You will be required to play little pieces of music, like the one above, on the piano. Some of them will be big, some will be small: some will be blurry and some will be clear. Our aim is to see what effect this has on your eye movements. There will also be a short questionnaire about your past and present musical life and a check of your sight reading capacity. The process will take approximately 1 hour. 210 APPENDIX 3 RECRUITMENT POSTER 2 HREA Approval No.: ______ Volunteers for Music Cognition Research Wanted!! The Psychophysical Characteristics of Musicians in Relation to Music Sight Reading and Aural Expertise and other Parameters. Musicians AND Non-Musicians required! You will be required to listen to a series of notes and determine which is the higher of the two or if the pitch/level of the note changes. You will then be asked to detect a visual target among other targets and identify an increasing number of targets in a line – numbers and letters for the non-musicians. Musicians tested with musical notes as well! A short IQ, Non-Word reading and Eye Movement test will follow. For musicians, there will also be a short questionnaire about your past and present musical life, an evaluation of your aural ability and of your music sight reading on a piano keyboard. The process will take approximately 80 minutes for non-musicians with an extra 10 minutes for musicians. 211 BIBLIOGRAPHY Ackman, P. L. (2014). Nonsense, common sense, and science of expert performance: Talent and individual differences. Intelligence, 45, 6-17. Ahissar, M., Lubin, Y., Putter-Katz, H., & Banai, K. (2006). Dyslexia and the failure to form a perceptual anchor. Nature Neuroscience, 9(12), 1558-1564. Akutsu, H., Bedell, H. E., & Patel, S. S. (2000). Recognition thresholds for letters with simulated dioptric blur. Optom Vis Sci, 77(10), 524-530. Amitay, S., Ahissar, M., & Nelken, I. (2002). Auditory Processing Deficits in reading Disabled Adults. Journal of the Association for Research in Otolaryngology, 3, 302-320. Anderson, D. E., Vogel, E. K., & Awh, E. (2013). A Common Discrete Resource for Visual Working Memory and Visual Search. Psychological Science, 24(6), 929- 938. doi:10.1177/0956797612464380 Anvari, S., Trainor, L., Woodside, J., & Levy, B. (2002). Relations among musical skills, phonological processing, and early reading ability in preschool children. J. Experimental Child Psychology, 83, 111-130. Ashby, J., Rayner, K., & Clifton, C. (2005). Eye movements of highly skilled and average readers: differential effects of frequency and predictability. Q J Exp Psychol A, 58(6), 1065-1086. doi:10.1080/02724980443000476 Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: A proposed system and its control processes. The psychology of learning and motivation: advances in research and theory (Vol. 2, pp. 89-195). New York: Academic Press. Atterbury, B. (1983). A Comparison of Rhythm Pattern Perception and Performance in Normal and Learning-Disabled Readers Aged 7 and 8. Journal of Research in Music Education, 31(4), 259-270. Baddeley, A. (2007). Working memory, Thought, and Action. Oxford: Oxford University Press. Baddeley, A., & Hitch, G. (1974). Working Memory. The Psychology of Learning and Motivation, 8. Balota, D. A., Pollatsek, A., & Rayner, K. (1985). The interaction of contextual constraints and parafoveal visual information in reading. Cogn Psychol, 17(3), 364-390. Banai, K., & Yifat, R. (2011). Perceptual Anchoring in Preschool Children: Not Adultlike, but There. PLoS ONE, 6(5), 1-9. Banai, K., & Yifat, R. (2012). Anchoring in 4- to 6-year-old children relates to predictors of reading. . Journal of Experimental Child Psychology, 112, 403-416. Beidas, H., Khateb, A., & Breznitz, Z. (2013). The cognitive profile of adult dyslexics and its relation to their reading abilities. Reading and Writing, 26, 1487-1515. Bentin, S., Truitt, A., Puce, A., Perez, E., & McCarthy, G. (1996). Electrophysiological Studies of Face Perception in Humans. Journal of Cognitive Neuroscience, 8(6), 551-565. Bilalic, M., Langner, R., Erb, M., & Grodd, W. (2010). Mechanisms and Neural Basis of Object and Pattern Recognition: A Study With Chess Experts. Journal of Experimental Psychology: General, 139(4), 728-742. 212 Bilalic, M., Langner, R., Ulrich, R., & Grodd, W. (2011). Many faces of chess expertise: Fusiform face area in chess experts and novices. The Journal of Neuroscience, 31(28), 10206-10214. Binder, K. S., Pollatsek, A., & Rayner, K. (1999). Extraction of information to the left of the fixated word in reading. J Exp Psychol Hum Percept Perform, 25(4), 1162- 1172. Boets, B., Vandermosten, M., Poelmans, H., Luts, H., Wouters, J., & Ghesquie, P. (2011). Preschool impairments in auditory processing and speech perception uniquely predict future reading problems. Research in Developmental Disabilities, 32, 560-570. Brochard, R., Dufour, A., & Despres, O. (2004). Effect of musical expertise on visuospatial abilities: Evidence from reaction times and mental imagery. Brain and Cognition, vol. 54, pp. 103–109. Brain and Cognition, 54, 103-109. Busey, T. A., & Vanderkolk, J. R. (2005). Behavioral and electrophysiological evidence for configural processing in fingerprint experts. . Vision Research, 45, 431–448. Cameron, E. L. (1995). Factors Determining the Latency of Saccadic Eye Movements. (PhD), University of Rochester, New York. Campitelli, G., & Gobet, F. (2011). Deliberate Practice Necessary But Not Sufficient. Current Directions in Psychological Science, 20(5), 280-285. Chace, K. H., Rayner, K., & Well, A. D. (2005). Eye movements and phonological parafoveal preview: effects of reading skill. Can J Exp Psychol, 59(3), 209-217. Chin, T., & Rickard, N. S. (2010). The Music USE (MUSE) Questionnaire: An instrument to measure engagement in music. . Music Perception, 29(4), 429-446. Chung, S., Legge, G., & and Cheung, S.-H. (2004). Letter-recognition and reading speed in peripheral vision benefit from perceptual learning. Vision Research, 44, 695- 709. Chung, S. T. (2010). Detection and identification of crowded mirror-image letters in normal peripheral vision. Vision Res, 50(3), 337-345. doi:10.1016/j.visres.2009.11.017 Chung, S. T., & Legge, G. E. (2009). Precision of position signals for letters. Vision Res, 49(15), 1948-1960. doi:10.1016/j.visres.2009.05.004 Corrigall, K. A., & Trainor, L. J. (2011). Associations Between Length of Music Training and Reading Skills in Children. Music Perception: An Interdisciplinary Journal, 29(2), 147-155. Curby, K., & Gauthier, I. (2010). To the Trained Eye: Perceptual Expertise Alters Visual Processing. Topics in Cognitive Science, 2, 189-201. Daniels, R. D. (1986). Relationships among Selected Factors and the Sight-Reading Ability of High School Mixed Choirs. Journal of Research in Music Education, 34(4), 279-289. doi:10.2307/3345261 Dee-Lucas, D., Just, M. A., Carpenter, P., & Daneman, M. (1982). What Eye Fixations tell us about Time Course if text Integration. In R. a. F. Groner, P. (Ed.), Cognition and Eye Movements (pp. 155-168). Amsterdam: North Holland Publishing Company. Deguchia, C., Boureux, M., Sarlo, M., Besson, M., Grassi, M., Schön, D., & Colombo, L. (2012). Sentence pitch change detection in the native and unfamiliar language in 213 musicians and non-musicians: Behavioral, electrophysiological and psychoacoustic study. . Brain Res, 1455, 75-89. Demany, L., & Semal, C. (1989). Detection thresholds for sinusoidal frequency modulation. J. Acoust. Soc. Am. 1295-1301, 85(3). Demerest, S., & Morrison, S. (2000). Does Music Make You Smarter? . Music Educators’ Journal., 87(2), 33-39+58. Detterman, D. (2014). Introduction to the intelligence special issue on the development of expertise: is ability necessary? Intelligence, 45, 1-5. Ehrle, N., & Samson, V. (2005). Auditory discrimination of anisochrony: Influence of the tempo and musical backgrounds of listeners. Brain and Cognition, 58, 133-147. Ehrlich, S., & Rayner, K. (1981). Contextual Effects on Word Perception and Eye Movements during Reading. Journal of Verbal Learning and Verbal Behaviour, 20, 641-655. Ekwall, E. E., & Shanker, J. L. (1988 ). Diagnosis and remediation of the disabled reader (3rd ed.). Boston: Allyn & Bacon. Elliott, C. A. (1982). The Relationships among Instrumental Sight-Reading Ability and Seven Selected Predictor Variables. Journal of Research in Music Education, 30(1), 5-14. doi:10.2307/3344862 Engbert, R., Longtin, A., & Kliegl, R. (2002). A Dynamic Model of Saccade Generation in reading Based on Spatially Distributed Lexical Processing. Vision Research, 42, 621-636. Ericsson, K. A., Krampe, R. T., & Tesch-Romer, C. (1993). The role of deliberate pracrice in the aquisition of expert performance. Psychology Review, 100(3), 363- 406. Ericsson, K. A., & Lehmann, A. (1996). Expert and exceptional performance: evidence of maximal adaptation to task constraints. Annual Review of Psychology, 47, 273- 305. Ericsson, K. A., Roring, R., & Nandagopal, K. (2007). Giftedness and evidence for reproducibly superior performance: An account based on the expert performance framework High Ability Studies, 18(1), 3-56. Falle, B. (2011). Proposed Methods For Efficiently Attaining The Skills Required For Accurate Time Keeping In Music. (Bachelor of Music Honours Honours), Edith Cowan University, WA. Farrington-Darby, T., & Wilson, J. (2006). The nature of expertise: A review. Applied Ergonomics, 37, 17-32. Fiser, J., & Aslin, R. (2001). Unsupervised Statistical Learning of Higher-Order Spatial Structures from Visual Scenes. Psychological Science, 12(6), 499-504. Fiveash, A., & Pammer, K. (2014). Music and language: Do they draw on similar syntactic working memory resources? Psychology of Music, 42(2), 190-209. doi:10.1177/0305735612463949 Fleisher, A. G. (1986). Control of Eye Movements by Working Memory Load. Biological Cybernetics, 55, 227-238. Folstein, J., Gauthier, I., & Palmeri, T. (2010). Mere exposure alters category learning of novel objects. Frontiers in Psychology, 1(40), 1-6. 214 Formby, C. (1985). Differential sensitivity to tonal frequency and to the rate of amplitude modulation of broadband noise by normally hearing listeners. J. Acoust. Soc. Am., 78(1), 70-77. Furneaux, S., & Land, M. F. (1999). The effects of skill on the eye-hand span during musical sight-reading. Proc Biol Sci, 266(1436), 2435-2440. doi:10.1098/rspb.1999.0943 Gaser, C., & Schlaug, G. (2003). Brain structures differ between musicians and non- musicians. J Neurosci, 23(27), 9240-9245. Gauthier, I., & Bukach, C. (2007). Should we reject the expertise hypothesis? Cognition, 103, 322-330. Gauthier, I., McGugin, R., Richker, J., Herzmann, G., Speegle, M., & Van Gulick, A. (2014). Experience moderates overlap between object and face recognition, suggesting a common ability. J Vis, 14(8), 1-12. Gauthier, I., & Tarr, M. (2002). Unraveling Mechanisms for Expert Object Recognition: Bridging Brain Activity and Behavior. Journal of Experimental Psychology: Human Perception and Performance, 28(2), 431-446. Gebhard, M. (1949). Changes in the attractiveness of activities: the effect of expectation preceding performance. . Journal of Experimental Psychology, 39(3), 404-413. Gernsbacher, M., & Robertson, R. (1995). Reading Skill and Suppression Revisited. Psychological Science, 6(3), 165-169. Gobet, F., Lane, P. C., Croker, S., Cheng, P. C., Jones, G., Oliver, I., & Pine, J. M. (2001). Chunking mechanisms in human learning. Trends in Cognitive Science, 5(6), 236-243. Goolsby, T. (1987). The Parameters of Eye Movements in Vocal Music Reading. Grabner, R. (2014). Going beyond the expert-performance framework in the domain of chess. Intelligence, 45, 109-111. Grimault, N., Micheyl, C., Carlyon, R., Bacon, S., & Collet, L. (2003). Learning in discrimination of frequency or modulation rate: generalization to fundamental frequency discrimination. Hearing Research, 184, 41-50. Gudmundsdottir, H. R. (2010). Advances in music-reading research. . Music Education Research, 12(4), 331-338. Hallam, S., Rinta, T., Varvarigou, M., & Creech, A. (2012). The development of Practice strategies in young people. Psychology of Music, 40(5), 652-680. Halverson, D. L. (1974). A Biometric Analysis of Eye Movement Patterns of Sight- Singers. (PhD Doctoral Dissertation), The Ohio State University. Hambrick, Oswald, F., Altmann, E., Meinz, E., Gobet, F., & Campitelli, G. (2014). Deliberate practice: Is that all it takes to become an expert? Intelligence, 45, 34- 45. Hambrick, D., Altmann, E., Oswald, F., Meinz, E., Gobet, F., & Campitelli, G. (2014). Accounting for expert performance: The devil is in the details. Intelligence, 45, 112-114. Hambrick, D., & Meinz, E. (2011). Limits on the predictive power of domain-specific experience and knowledge in skilled performance. Current Directions in Psychological Science, 25(5), 275-279. Heller, D. (1982). Eye Movements in Reading. In R. a. F. Groner, P. (Ed.), Cognition and Eye Movements (pp. 139-154). Amsterdam: North Holland Publishing Company. 215 Horst, J. W. (1989). Detection and discrimination of frequency modulation of complex signals. J. Acoust. Soc. Am., 85(5), 2022-2030. Howe, M., Davidson, J., & Sloboda, J. (1998). Innate talents: reality or myth? Behaviour and Brain Sciences, 21, 399-422. Hsiao, J., & Cottrell, G. (2009). Not all visual expertise is holistic: but it may be leftist. Psychological Science, 20(4), 455-463. Jacobsen, O. (1941). An Analytical Study of Eye Movements in Reading Vocal and Instrumental Music. Journal of Musicology, 3. Jaeggi, S., Buschkuehl, M., Jonides, J., & Perrig, W. (2008). Improving fluid intelligence with training on working memory. Proceedings of the National Academy of Sciences of the USA, 105(19), 6829-6833. Juhasz, B. J., & Rayner, K. (2003). Investigating the effects of a set of intercorrelated variables on eye fixation durations in reading. J Exp Psychol Learn Mem Cogn, 29(6), 1312-1318. doi:10.1037/0278-7393.29.6.1312 Just, M. A., & Carpenter, P. A. (1976). Eye Fixations and Cognitive Processes. Cogn Psychol, 8, 441-480. Kanwisher, N., McDermott, J., & Chun, M. (1997). The fusiform face area: A module in human extrastriate cortex specialized for face perception. The Journal of Neuroscience, 17(11), 4302-4311. Kelly, D., Liu, S., Lee, K., Quinn, P., Pascalis, O., Slater, A., & Ge, L. (2009). Development of the other-face effect during infancy: Evidence to ward universality? . Journal of Experimental Child Psychology, 104, 105-114. Kennison, S. M., & Clifton, C., Jr. (1995). Determinants of parafoveal preview benefit in high and low working memory capacity readers: implications for eye movement control. J Exp Psychol Learn Mem Cogn, 21(1), 68-81. Kidd, J. C., Shum, K. K. M., Wong, A. M. Y., Ho, C. S. H., and , & Au, T. K. F. (2015). Auditory Perception and Word Recognition in SLI. In Press. Kinsler, V., & Carpenter, R. (1995). Saccadic eye movements while reading music. Vision Research, 35(10), 1447-1458. Kinsler, V., & Carpenter, R. H. S. (1995). Saccadic eye movements while reading music. Vision Research, 35(10), 1447-1458. doi:http://dx.doi.org/10.1016/0042- 6989(95)98724-N Kopiez, R., & Lee, J. I. (2008). Towards a general model of skills involved in sight reading music. Music Education Research, 10(1), 41-62. Kowler, & Anton, S. (1987). Reading Twisted Text: Implications for the role of Saccades. Vision Research, 27(1). Kowler, E. (2011). Eye movements: the past 25 years. Vision Res, 51(13), 1457-1483. doi:10.1016/j.visres.2010.12.014 Krause, V., Pollok, B., & Schnitzler, A. (2010). Perception in action: The impact of sensory information on sensorimotor synchronization in musicians and non- musicians. Acta Psychologica, 133, 28-37. Kulasegaram, K., Grierson, L., & Norman, G. (2013 ). The roles of deliberate practice and innate ability in developing expertise: evidence and implications. Medical Education, 47, 979-989. Lang, M. (1961). An Investigation of Eye Movements Involved in the Reading of Music. Paper presented at the The International Ophthalmic Optical Congress, London. 216 Legge, G. (2007). Psychophysics of Reading in Normal and Low Vision. Mahwah, New Jersey, USA.: Lawrence Erlbaum Associates, Inc., Publishers. Legge, G., Cheung, S., Yu, D., Chung, S., Lee, H. W., & Owens, D. (2007). The case for the visual span as a sensory bottleneck in reading. J Vis, 7(2), 9 1-15. doi:10.1167/7.2.9 Legge, G. E. (2007). The Psychophysics of Reading. New Jersey, USA: Lawrence Erlbaum Associates, Inc. Legge, G. E., Ahn, S. J., Klitz, T. S., & Luebker, A. (1997). Psychophysics of reading-- XVI. The visual span in normal and low vision. Vision Res, 37(14), 1999-2010. Legge, G. E., Rubin, G. S., & Luebker, A. (1987). Psychophysics of reading--V. The role of contrast in normal vision. Vision Res, 27(7), 1165-1177. Lervag, A., & Hulme, C. (2009). Rapid Automatised naming Taps a Mechanism That Places Constraints on the Development of Early Reading Fluency. Psychological Science, 20(8), 1040-1048. Levi, D. M. (2008). Crowding--an essential bottleneck for object recognition: a mini- review. Vision Res, 48(5), 635-654. doi:10.1016/j.visres.2007.12.009 Lombardo, M., & Deaner, R. (2014). You can't teach speed: sprinters falsify the deliberate practice model of expertise. PeerJ, DOI 10.7717/peerj.445, 1-31. Lu, Z.-L., Huab, T., Huanga, C.-B., Zhouc, Y., & Dosher, B. A. (2011). Visual perceptual learning. Neurobiology of Learning and Memory, 95, 145-151. Luck, S., & Vogel, E. (1997). The capacity of visual working memory for features and conjunctions. Nature, 390, 279-281. Madell, M., & Hebert, S. (2008). Eye Movements and Music: Where do we Look Next? Music Perception, 26, 157-170. Maniglia, M., Pavan, A., Cuturi, L., Campana, G., Sato, G., & Casco, C. (2011). Reducing Crowding by Weakening Inhibitory Lateral Interactions in the Periphery with Perceptual Learning. PLoS ONE, 6(10), 1-9. Marrs, H., & Patrick, C. (2002). A Retrun to Eye Movement Training? An Evaluation of the Reading Plus Program. Reading Psychology, 23(4), 297-322. McAnany, J. J., Shahidi, M., Applegate, R. A., Zelkha, R., & Alexander, K. R. (2011). Contributions of optical and non-optical blur to variation in visual acuity. Optom Vis Sci, 88(6), 716-723. doi:10.1097/OPX.0b013e3182147202 McConkie, G. W., & Rayner, K. (1975). The span of effective stimulus during a fixation in reading. . Perception & Psychophysics, 17, 578-586. McGugin, R., Gatenby, J., Gore, J., & Gautier, I. (2012). High-resolution imaging of expertise reveals reliable object selectivity in the FFA related to perceptual performance. Proc Natl Acad Sci U S A, 109(42), 17063-17068. McGugin, R., Richler, J., Herzmann, G., Speegle, M., & Gauthier, I. (2012). The Vanaderbilt Expertise test reveals domain-general and domain-specific sex effects in object recognition. Vision Research, 69, 10-22. McKone, E., Crookes, K., Jeffery, L., & Dilks, D. (2012). A critical review of the development of face recognition: Experience is less important than previously believed. Neuropsychology, 29(1-2), 174-212. McMullen, E., & Saffran, J. (2004). Music and Language: A Developmental Comparison. Music Perception: An Interdisciplinary Journal, 21(3), 289-311. 217 Megreya, A., & Havard, C. (2011). Left face matching bias: Right hemisphere dominance or scanning habits? Laterality: Asymmetries of Body, Brain and Cognition, 16(1), 75-92. Meinz, E., & Hambrick, D. (2010). Deliberate Practice Is Necessary but Not Sufficient to Explain Individual Differences in Piano Sight-Reading Skill: The Role of Working Memory Capacity. Psychological Science, 21(7), 914-919. Meseguer, E., Carreiras, M., & Clifton, C., Jr. (2002). Overt reanalysis strategies and eye movements during the reading of mild garden path sentences. Mem Cognit, 30(4), 551-561. Meyer, G., & Wuerger, S. (2001). Cross Modal Integration of Auditory and Visual Motion Signals. NeuroReport, 12(11), 2557-2560. Micheyl, C., Delhommeau, K., Perrot, X., & Oxenham, A. (2006). Influence of musical and psychoacoustical training on pitch discrimination. Hearing Research, 219, 36-47. Miellet, S., & Sparrow, L. (2004). Phonological codes are assembled before word fixation: evidence from boundary paradigm in sentence reading. Brain Lang, 90(1-3), 299-310. doi:10.1016/S0093-934X(03)00442-5 Miller, B. A. (2012). Rhythmic Instruction From Square One: A Constructivist Teacher and Her Metronome. General Music Today, 26(1), 19-26. Milner, B. (1971). Interhemispheric differences in the localization of psychological processes in man. British Medical Bulletin, 27, 272–7. British Medical Bulletin, 27, 272-277. Mishra, J. (2013). Improving sightreading accuracy: A meta-analysis. Psychology of Music, Vol. 42(2), 131–156. Mishra, J. (2014). Factors Related to Sight-Reading Accuracy: A Meta Analysis. Journal of Research in Music Education, 61(4), 452-465. Mishra, J. (2014). Improving sightreading accuracy: A meta-analysis. Psychology of Music, 42(2), 131-156. doi:10.1177/0305735612463770 Misra, M., Katzir, T., Wolf, M., & Poldrack, R. (2004). Neural Systems for Rapid Automatized Naming in SkilledReaders: Unraveling the RAN/Reading Relationship. Scientific Studies of Reading, 8(3), 241-256. Moreno, S., & Besson, M. (2006). Musical training and language-related brain electrical activity in children. Psychophysiology, 43, 287-291. Murray, D. J. (1967). The role of speech responses in short-term memory. . Canadian Journal of Psychology, 21, 263-276. Nergard-Nilssen, T., & Hulme, C. (2014). Developmental Dyslexia in Adults: Behavioural Manifestations and Cognitive Correlates. Dyslexia, 20, 191-207. Norton, A., Winner, E., Cronin, K., Overy, K., Lee, D., & and Schlaug, G. (2005). Are there pre-existing neural, cognitive, or motoric markers for musical ability? Brain and Cognition, 59, 124-134. Norton, T., Corliess, D. and Bailey, J. (2002). The psychophysical measurement of visual function. Amsterdam: Butterworth. Nystrom, M., & Holmqvist, K. (2010). An adaptive algorithm for fixation, saccade, and glissade detection in eyetracking data. behaviour Research Methods, 42(1), 188- 204. 218 O’Regan, J. K., Levy-Schoen, A., & Jacobs, A. M. (1983). The effect of visibility on eye- movement parameters in reading. . Perception and Psychophysics., 34(5), 457- 464. Oganian, Y., & Ahissar, M. (2012). Poor Anchoring Limits Dyslexics’ Perceptual, Memory and reading Skills. Neuropsychologia, 50, 1895-1905. Patston, L., Kirk, I., Mei, H., Rolfe, M., Corballis, L., & Tippett, J. (2007). The unusual symmetry of musicians: Musicians have equilateral interhemispheric transfer for visual information. . Neuropsychologia, 45, 2059–2065. Pavlidis, G. (1981). Do Eye Movements Hold the Key to Dyslexia? Neuropsychologia, 19, 57-64. Pelli, D. G., Tillman, K. A., Freeman, J., Su, M., Berger, T. D., & Majaj, N. J. (2007). Crowding and eccentricity determine reading rate. J Vis, 7(2), 20 21-36. doi:10.1167/7.2.20 Perea, M., & Acha, J. (2009). Space information is important for reading. Vision Research, 49(15), 1994-2000. doi:http://dx.doi.org/10.1016/j.visres.2009.05.009 Pietrzyk, M. W., McEntee, M. F., Evanoff, M. E., Brennan, P. C., & Mello-Thoms, C. R. (2014). Direction of an initial saccade depends on radiological expertise. Paper presented at the Medical Imaging Plomin, R., Shakeshaft, N., McMillan, A., & Trzaskowski, M. (2014). Nature, nurture, and expertise. Intelligence, 45, 46-59. Poll, G., Miller, C., Mainela-Arnold, E., Donnelly Adams, K., Misra, M., & Park, J. (2013). Effects of children’s working memory capacity and processing speed on their sentence imitation performance. International Journal of Language and Commnication Disorders, 48(3), 329-342. Posner, M. I. (1980). Orienting of Attention. Q J Exp Psychol (Hove), 32(1), 3-25. Rauscher, F., & Shaw, D. (1998). Key components of the Mozart effect. Perceptual and Motor Skills, 86, 835-841. Rauscher, F. H., & Hinton, S. C. (2011). Music Instruction and its Diverse Extra-Musical Benefits. Music Perception, 29(2), 215-226. Rayner, K. (1986). Eye movements and the perceptual span in beginning and skilled readers. J Exp Child Psychol, 41(2), 211-236. Rayner, K. (1986). Eye Movements and the Perceptual Span in Beginning and Skilled Readers. . Journal of Experimental Child Psychology., 41, 211-236. Rayner, K. (1998). Eye movements in reading and information processing: 20 years of research. Psychol Bull, 124(3), 372-422. Rayner, K., Chace, K., Slattery, T., & Ashby, J. (2006). Eye Movements as Reflections of Comprehension Processes in Reading. Scientific Studies of Reading, 10(3), 241- 255. Rayner, K., Fischer, M., & Pollatsek, A. (1998). Unspaced text interferes with both word identification and eye movement control. Vision Research, 38(8), 1129-1144. doi:http://dx.doi.org/10.1016/S0042-6989(97)00274-5 Rayner, K., & McConkie, G. (1976). What Guides a Reader's Eye Movements. Vision Research, 16, 829-837. Rayner, K., Pollatsek, A., & Binder, K. S. (1998). Phonological codes and eye movements in reading. J Exp Psychol Learn Mem Cogn, 24(2), 476-497. 219 Rayner, K., Well, A. D., & Pollatsek, A. (1980). Asymmetry of the effective visual field in reading. . Perception & Psychophysics, 27(6), 537-544. Redford, J., Green, S., Geer, M., Humphrey, M., & Thiede, K. (2011). Exploring metacognitive accuracy in visual search. Memory & Cognition, 39(8), 1534-1545. doi:10.3758/s13421-011-0123-y Reichle, E. D., Pollatsek, A., Fisher, D., & Rayner, K. (1998). Toward a model of eye movement control in reading. Psychol Rev, Vol. 105(1), 125-157. Repp, B. (2010). Sensorimotor synchronization and perception of timing: Effects of music training and task experience. Human Movement Science, 29, 210-213. Repp, B., & Penel, A. (2002). Auditory Dominance in Temporal Processing: New Evidence From Synchronization With Simultaneous Visual and Auditory Sequences. Journal of Experimental Psychology: Human Perception and Performance, 28(5), 1085–1099. Restle, F. (1972). Serial Patterns: The Role of Phrasing. Journal of Experimental Psychology, 92(3), 385-390. Risse, S. (2014). Effects of visual span on reading speed and parafoveal processing in eye movements during sentence reading. J Vis, 14(8), 1-13. Rogalsky, C., Rong, F., Saberi, K., & Hickok, G. (2011). Functional Anatomy of Language and Music Perception: Temporal and Structural Factors Investigated Using Functional Magnetic Resonance Imaging. The Journal of Neuroscience, 31(10), 3843–3852 Roser, M., Fiser, J., Aslin, R., & Gazzaniga, M. (2011). Right Hemisphere Dominance in Visual Statistical Learning Journal of Cognitive Neuroscience, 23(5), 1088-1099. Rossion, B., Gauthier, I., Tarr, M., Despland, P., Bruyer, R., Linotte, S., & Crommelink, M. (1999). The N170 occipito-temporal component is delayed and enhanced to inverted faces but not to inverted objects: an electrophysiological account of face- specic processes in the human brain. Cognitive Neuroscience, 11(1), 69-74. Rossion, B., Kung, C.-C., & Tarr, M. (2004). Visual Expertise with Nonface Objects Leads to Competition with the Early Perceptual Processing of Faces in the Human Occipitotemporal Cortex. Proc Natl Acad Sci U S A, 101(40), 14521-14526. Ruthsatz, J. (2014). The Summation Theory as a multivariate approach to exceptional performers. Intelligence, 45. Ruthsatz, J., Ruthsatz, K., & Ruthsatz-Stephens, K. (2014). Putting practice into perspective: Child prodigies as evidence of innate talent. Intelligence, 45, 60-65. Ruthsatz, J., & Urbach, J. (2012). Child prodigy: A novel cognitive profile places elevated general intelligence, exceptional working memory and attention to detail at the root of prodigiousness. Intelligence, 40, 419-426. Salthouse, T., & Ellis, C. (1980). Determinants of Eye Fixation Duration. The American Journal of Psychology, 93(2), 207-234. Schellenberg, E. G. (2006). Long-Term Positive Associations Between Music Lessons and IQ. Journal of Educational Psychology, 98(2), 457-468. Schlaug, G., Janke, L., Huang, Y., Staiger, J., & Steinmetz, H. (1995). Increased Corpus Callosum Size in Musicians. Neurophysiologica, 33(8), 1047-1055. Schmidt, F. (1981). Movement Patterns of Woodwind Instrument Performers while Sight Reading Music. 220 Sears, R. (1943). Motivational factors in aptitude testing. American Journal of Orthopsychiatry, 13(3), 468. doi:10.1111/j.1939-0025.1943.tb06014.x Sengpiel, F. (2007). The Critical Period. Current Biology, 17(17), R742-743. Simonton, D. K. (2007). Talent and expertise: the empirical evidence for genetic endowment. High Ability Studies, 18(1), 83-84. Simonton, D. K. (2014). Creative performance, expertise acquisition, individual differences, and developmental antecedents: An integrative research agenda. Intelligence, 45, 66-73. Sisson, E. D. (1938). Eye-Movement Training as a Means of Improving Reading. Journal of Education Research, 32. Sloboda, J. (1974). The Eye-Hand Span: An Approach to the Study of Sight Reading. Psychology of Music, 2(4), 4-10. Sloboda, J. (1977). Phrase Units as Determinants of Visual Processing in Music Reading. Br J Psychol, 68, 117-124. Sloboda, J. (1985). The Musical Mind: The Cognitive Psychology of Music. Oxford: Clarendon Press. Slotnick, S., & White, R. (2013). The fusiform face area responds equivalently to faces and abstract shapes in the left and central visual fields. Neuroimage, 83, 408-417. Sperling, G. (1960). The information available in brief visual presentations. Psychological Monographs, 74(11), 1-29. Spiegel, M., & Watson, C. (1984). Performance on frequency-discrimination tasks by musicians and nonmusicians. J. Acoust. Soc. Am., 76(6), 1690-1695. Stainthorp, R., Stuart, M., Powell, D., Quinlan, P., & Garwood, H. (2010). Visual Processing Deficits in Children With Slow RAN Performance. Scientific Studies of Reading, 14(3), 266-292. Talcott, J., Witton, C., McClean, M., Hansen, P., Rees, A., Green, G., & Stein, J. (1999). Can sensitivity to auditory frequency modulation predict children's phonological and reading skills? NeuroReport, 10, 2045-2050. Tanaka, J., & Farah, M. (1993). Parts and Wholes in Face Recognition. The Quarterly Journal of Experimental Psychology, 46A(2), 225-245. Tatler, B. W., Wade, N. J., Kwan, H., Findlay, J. M., & Velichkovsky, B. M. (2010). Yarbus, eye movements, and vision. Iperception, 1(1), 7-27. doi:10.1068/i0382 Tinker, M. A. (1946). The study of eye movements in reading. Psychol Bull, 43, 93-120. Truitt, F., Clifton, C. J., Pollatsek, A., & Rayner, K. (1997). The Perceptual Span and the Eye-Hand Span in Sight Reading Music. Visual Cognition, 4(2), 143-161. Tso, R., Au, T., & Hsiao, J. (2014). Perceptual Expertise: Can Sensorimotor Experience Change Holistic Processing and Left-Side Bias? Psychological Science, 25(9), 1757-1767. Tsodyks, M., & Gilbert, C. (2004). Neural networks and perceptual learning. Nature, 431, 775-781. Underwood, G., Clews, S., & Everatt, J. (1990). How do readers know where to look next? Local information distributions influence eye fixations. Q J Exp Psychol A, 42(1), 39-65. Underwood, G., Hubbard, A., & Wilkinson, H. (1990). Eye fixations predict reading comprehension: the relationships between reading skill, reading speed, and visual inspection. Lang Speech, 33 ( Pt 1), 69-81. 221 Van Nuys, K. a. W., H,. (1943). Memory Span and Visual Pauses in Reading Rhythms and Melody. Psychological Monographs, 55, 31-50. Vandierendonck, A., Kemps, E., Fastame, M., & Szmalec, A. (2004). Working memory components of the Corsi blocks task. Br J Psychol, 95, 57-79. Wade, N. J. (2010). Pioneers of eye movement research. Iperception, 1(2), 33-68. doi:10.1068/i0389 Wai, J. (2014). Experts are born, then made: Combining prospective and retrospective longitudinal data shows that cognitive ability matters. Intelligence, 45, 74-80. Wallis, G. (2013). Towards a unified model of face and object recognition in the human visual system. Frontiers in Psychology, Vol. 4, Article 497, pp. 1-25. Frontiers in Psychology, 4, 1-25. Watanabe, D., Savion-Lemieux, T., & Penhune, V. (2007). The effect of early musical training on adult motor performance: evidence for a sensitive period in motor learning. Exp Brain Res, 176, 332-340. Watanabe, T., & Sasaki, Y. (2015). Perceptual Learning: Toward a Comprehensive Theory. Annual Review of Psychology, 66, 197-221. Waters, Townsend, E., & Underwood, G. (1998). Expertise in musical sight reading: A study of pianists. Br J Psychol, 89, 123-149. Waters, A., Underwood, G., & Findlay, J. (1997). Studying expertise in music reading: Use of a pattern-matching paradigm. Perception & Psychophysics, 59(4), 477- 488. doi:10.3758/BF03211857 Waters, A. J., Gobet, F., & Leyden, G. (2002). Visuo-spatial abilities in chess players. . Br J Psychol, 30, 303-311. Waters, A. J., Townsend, E., & Underwood, G. . (1998). Expertise in musical sight reading: a study of pianists. Br J Psychol, 89(1), 123–149. Watkins, A. (2000). Intuitive Overlays Test. London, U.K. Weaver, H. E. (1943). A Survey of Visual Processes in Reading Differently Constructed Musical Selections. Psychological Monographs, 55, 1-30. Werner, P. D., Swope, A. J., & Heide, F. J. (2006). The Music Experience Questionnaire: development and correlates. . The Journal of Psychology, 140, 329-345. Wetherill, G. B. a. L., H. (1965). Sequential Estimation of Points on a Psychometric Function. The British Journal of Mathematical and Statistical Psychology, 18(1), 1-10. Wolf, T. (1976). A cognitive model of musical sight-reading. J Psycholinguist Res, 5(2), 143-171. Wong, Y. K., & Gauthier, I. (2009). A Multimodal Neural Network Recruited by Expertise with Musical Notation. Journal of Cognitive Neuroscience, 22(4), 695- 713. Wong, Y. K., & Gauthier, I. (2012). Music-reading expertise alters visual spatial resolution for musical notation. Psychon Bull Rev, 19(4), 594-600. doi:10.3758/s13423-012-0242-x Woods, A. J., Göksun, T., Chatterjee, A., Zelonis, S., Mehta, A., & Smith, S. E. (2013). The development of organized visual search. Acta Psychologica, 143(2), 191-199. doi:http://dx.doi.org/10.1016/j.actpsy.2013.03.008 Woody, R. H. (2012). Playing by Ear: Foundation or Frill? . Music Educators Journal, 99(2), 82-88. 222 Wurtz, P., Mueri, R., & Wiesendanger, M. (2009). Sight-reading of violinists: eye movements anticipate the musical flow. Exp Brain Res, 194(3), 445-450. doi:10.1007/s00221-009-1719-3 Yang, S.-N., & McConkie, G. W. (2001). Eye movements during reading: a theory of saccade initiation times. Vision Research, 41, 3567–3585. York, R. (1952). An Experimental Study of Vocal Music Reading using Eye Movement Photography and Voice Recording. Young, L. (1971). A Study of the Eye Movements and Eye-Hand Temporal Relationships of Successful and Unsuccessful Piano Sight Readers while Piano Sight reading. Yu, D., Cheung, S. H., Legge, G. E., & Chung, S. T. (2007). Effect of letter spacing on visual span and reading speed. J Vis, 7(2), 2 1-10. doi:10.1167/7.2.2 Zimmer, U., Roberts, K., Harshbarger, T., & Woldorff, M. (2010). Multisensory conflict modulates the spread of visual attention across a multisensory object. Neuroimage, 52, 606-616. 223