35030ournal ofNeurology, Neurosurgery, and Psychiatry 1995;58:350-356

Ettlinger revisited: the relation between J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.58.3.350 on 1 March 1995. Downloaded from and sensory impairment

Edward H F De Haan, Charles A Heywood, Andrew W Young, Nicki Edelstyn, Freda Newcombe

Abstract tinction between apperceptive and associative The concept of agnosia as a higher order "Seelenblindheit", where the first results from functional impairment, which can occur visuosensory deficits and the second was in the absence of low level visual percep- thought to represent a difficulty in associating tual deficits, continues to provoke debate. meaning with a relatively intact percept. A few This controversy is complicated by the years later, Freud4 coined the term "agnosia" fact that, on close examination, agnosic to describe these recognition disturbances. It patients do tend to have some perceptual is the viability of the concept of associative difficulties. Thus the issue centres agnosia or as Teuber5 put it "a normal per- around the question as to whether these cept stripped of its meaning"-that has been deficits play a causal part in the aetiology questioned. Notably, Bay67 claimed that the of agnosia or whether they are function- socalled higher order recognition deficits are ally independent, with both impairments secondary to sensory impairment, general resulting from the substantial cerebral intellectual loss, language problems, or a com- lesions involved in agnosia. In 1956, bination of these factors. Ettlinger published a study' in which he More recent investigations have convinc- compared the performance of patients ingly shown that visual recognition disorders with visual recognition deficits and can occur in patients with normal or even patients with posterior brain lesions above average intelligence.8 10 Also, it has now whose recognition abilities were intact. been clearly established that language difficul- He argued that visual perceptual prob- ties are not instrumental in causing associative lems could not explain the recognition .I112 The question whether subtle sen- deficit in agnosia as he saw far worse per- sory impairment (Bay's "Funktionswandel"), ceptual impairments in patients who did or a certain constellation of sensory deficits, not experience any problems in visual can produce the clinical symptoms of associa- recognition. Although the logic of tive agnosia, however, remains controversial. Ettlinger's argument is not disputed, The debate is fuelled by the fact that even the some criticisms have arisen concerning most pure cases of agnosia often show some Russell-Cairns Unit, the study, such as the fact that his experi- mild problems on tasks of visual . Radcliffe Infirmary, In a seminal study, Ettlinger' set out to test http://jnnp.bmj.com/ Woodstock Road, mental group did not include a truly Oxford OX2 6HE, UK object agnosic patient. In addition, Bay's hypothesis. He carried out a careful E H F De Haan Ettlinger's visual-sensory assessment can assessment of sensory status in patients with N Edelstyn no longer be considered comprehensive and patients without recognition deficits, and F Newcombe in the light of present day knowledge of he argued that sensory status alone could not Psychology Department, Durham the cerebral visual apparatus. This study explain the presence or absence of recognition University, Science therefore investigated three disorder. The crux of his argument was that Laboratories, Durham with a deficit (prosop)agnosic patients and five although patients recognition on October 1, 2021 by guest. Protected copyright. DH1 3LE, UK have other C A Heywood patients with unilateral brain lesions may sensory impairments, patients on an exten- who do not experience recognition problems MRC Applied without recognition deficits Psychology Unit, 15 sive battery of visual sensory tests. The can show equal or worse impairments on the Chaucer Road, results support Ettlinger's original claim sensory tests. Cambridge CB2 2EF, that some cases) agnosia cannot be Recently, Campion'3 has reopened the dis- UK (in A W Young explained as resulting from lower level cussion, pointing out that Ettlinger's study include a Department of visual impairments. did not truly object agnosic patient, Psychonomics, and that Ettlinger's tests might not have cov- Utrecht University, (7NeurolNeurosurg Psychiatry 1995;58:350-356) ered all the relevant sensory abilities for visual Heidelberglaan 2, recognition. It is noted, however, that one Utrecht, The Netherlands patient in Ettlinger's population was fully E H F De Haan Keywords: agnosia; recognition deficits; lower level prosopagnosic (see Pallis14). Although visual impairments; visual sensory tests prosopagnosia the inability to recognise Correspondence to: familiar faces'- is often considered function- Dr Edward De Haan, Psychological Laboratory, Visual recognition disorders as a result of ally separate from object agnosia,'6 17 the Utrecht University, brain disease or injury have created theoretical dichotomy of apperceptive v associative is still Heidelberglaan 2, 3584 CS Utrecht, The Netherlands. controversies since they were first described in applicable. Therefore, we suggest that there is Received 25 April 1994 and the second half of the 19th century.2 a prima facie case that Ettlinger's line of rea- in final revised form Regarding the more selective impairment of soning also applies to object agnosia. 2 August 1994. Accepted 5 August 1994 object recognition, Lissauer3 suggested a dis- Campion's second criticism is more serious, Ettlinger revisited: the relation between agnosia and sensory impairment 351

as with hindsight Ettlinger did not test for all name or the voice. His visual recognition J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.58.3.350 on 1 March 1995. Downloaded from sensory abilities that later knowledge has problems were not confined to faces. Other shown might possibly affect visual recogni- types of within class recognition," such as for tion. Also, Ettlinger used a composite sensory cars or flowers, were very poor, and on object score, and did not look for subtle problems on naming tasks he showed a mild degree of specific sensory abilities or constellations of object agnosia. The severity of his impairment sensory impairments. of face recognition is well illustrated by his Recent advances in neuropsychological and performance on a forced choice task in which neurophysiological understanding of the visu- he was required to choose which one of two osensory apparatus have put us in a much bet- simultaneously presented faces looked famil- ter position to consider this issue. There is iar. Each trial involved a famous and an unfa- now strong evidence for relatively indepen- miliar face presented on photographic slides, dent functional components in human visual and he was asked to guess if he was unsure. processing. 18-20 Neuropsychological research His performance on this task was not signifi- has shown highly selective disorders of colour cantly different from chance, whereas he per- (but not grey) discrimination,2' movement formed much better on a parallel task with perception,22 and luminance (not colour) dis- written names." Recent MRI showed bilateral crimination.2324 Moreover, Warrington2' has abnormal signal in the temporo-occipital shown the existence of dissociable impair- junction, mainly in the inferior surface. ments affecting visual acuity, shape discrimi- Patient MS-This patient was first exam- nation, location, and colour perception in a ined in 1971 and has been studied at regular small group of patients who had made a par- intervals thereafter.263' 3335 In 1970, MS was a tial recovery from blindness. This advance in 21 year old, left handed police cadet, when he knowledge not only indicates in a more had presumptive herpes encephalitis. The detailed way which sensory abilities to test for; permanent neurological and clinical features it has also resulted in much more refined include a left sided homonymous hemianopia, assessment techniques.26 central , amnesia, agnosia, and The aim of the present study is therefore to prosopagnosia. He has never presented with reconsider the issue raised by Ettlinger by sensorimotor loss, , or apraxia. His investigating sensory status in three severely verbal IQ has remained in the average range prosopagnosic or object-agnosic patients and and he is a regular reader of a daily newspa- five exservicemen with unilateral brain lesions per. He is interested in and better than aver- who showed no recognition deficits. age at mental arithmetic (WAIS-R scale score 14). His visual acuity is excellent on clinical examination. At the age of 42, he was still able Methods to read, uncorrected, with either left or right SUBJECTS eye, the smallest print (N5) of ophthalmologi- Case descriptions cal text. His difficulty in recognising objects Patient PH-In 1982, at the age of 19, PH (and faces) was very severe and had not was involved in a road traffic accident in changed significantly on formal testing since which he sustained a severe closed head 1971. Using subsets of the Snodgrass and injury. The physical, ophthalmological, and Vanderwart36 line drawings, he only recog- neuropsychological sequelae have been nised seven from a sample of 72 on one occa- http://jnnp.bmj.com/ described in detail by De Haan and cowork- sion and 25 out of 80 pictures on another ers,27-29 and only the essential points will be occasion. The recognition disorder was not reiterated here. On examination in 1985 his confined to the visual modality: MS also intellectual abilities were reduced on the per- made errors in tactual object identification. In formance subtests, but were in the average 1989 MRI showed bilateral temporo-occipital range on the verbal subtests of the revised damage and enlarged ventricles, particularly

Wechsler adult intelligence scale. Language pronounced on the right and in its posterior on October 1, 2021 by guest. Protected copyright. functions were relatively preserved, and he sector.26 could read without difficulty. His short term Patient NR-In 1982, at the age of 25, NR memory was intact with verbal and visual was involved in a head on car collision in material, but long term memory was severely which he sustained a severe closed head reduced in both domains. He was well ori- injury. The details concerning his case are in ented, however, and could give a coherent the report by De Haan et al.37 His intellectual account of his daily activities. The patient had abilities were in the "below average" range on a variety of problems in visual and spatial per- verbal subtests and his performance abilities ception apparent on tasks that included visual were inferior. He could read (small) printed closure and perception of three dimensional text without difficulty, but silent reading for structure. By contrast, he could read very comprehension was very slow. Spontaneous small print. Faces were recognised as such, speech was technically fluent but hampered and easily distinguished from animal faces. by word finding problems. He had no prob- Matching of photographs of unfamiliar lem understanding spoken instructions and faces,30 recognition of emotional facial expres- his writing was legible. Short term verbal and sions, and race and sex discriminations were spatial memory were normal, but long term impaired but still possible. Overt recognition retention of verbal and non-verbal material of familiar faces was nearly always unsuccess- were grossly impaired. Colour perception was ful, but people whose faces remained unidenti- moderately impaired; he made five errors on fied were usually easily recognised from the the Ishihara colour test. Copying of complex 352 Haan, Heywood, Young, Edelstyn, Newcombe

line drawings such as the Rey-Osterrieth fig- evidence of dementia on the mini mental state J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.58.3.350 on 1 March 1995. Downloaded from ure (34/36) and line drawings of objects was questionnaire. satisfactory. His visuospatial perception was impaired as was perception of facial stimuli. EXPERIMENTAL PROCEDURE He had problems inferring the age and sex Recognition tasks from photographs of faces. Simultaneous Two tasks were used to assess the ability to matching and recognition of unfamiliar faces recognise visual stimuli. The familiar face was essentially at chance level, and he per- recognition task involves 20 photographs of formed at the lower end of the normal range media celebrities (politicians, TV personali- on a matching task of facial expressions. ties, etc) that were individually presented for a Identification of familiar faces, whether by recognition response. A trial was scored cor- naming them or giving appropriate semantic rect if the subject was able to recall the name of information, was virtually impossible. He per- the celebrity, or could produce any particular formed at chance level on a task where he had information that identified that person (see to decide whether a face looked familiar or not De Haan et al'9: line up task, highly familiar (19/32 correct), but was within the normal condition). The Oldfield-Wingfield object range (30/32 correct) on a parallel version of naming test consists of 36 line drawings of this task, which used written names as stimuli. objects, ranging in familiarity from "shoe" to Visual recognition difficulties for objects were "gyroscope", that are individually presented apparent on a selection of 40 Snodgrass line and the subject is required to name them.40 drawings in which NR recognised only 10 The control data were taken from 25 control items. Within category discrimination of visu- subjects for the gunshot wound group. ally similar items was also very poor. In April and May 1982 CT showed a shallow extra- Visual sensory tasks cerebral collection of blood and CSF over the Visual-sensory assessment was carried out convexity of the right parietal lobe. Repeat with the following apparatus. Stimuli were CT in June 1983 showed an area of infarction presented on a 20 inch colour monitor in the left parietotemporal region of the brain. (Mitsubishi 3922ELP) with long persistence phosphors. The Commision Internationale de Normal controls for the agnosic patients l'Eclairage coordinates for the three phos- Two separate groups of control subjects were phors were: red, x = 0-625, y = 0 34; green, selected on the basis of age. To assess the per- x = 0-205, y = 0 7; blue, x = 0 15, y = 0*06. formance of PH and NR six healthy volun- The screen was controlled by a Pluto II teers in the age range of 20-29 years were graphics device (Electronic Graphics Ltd) tested, and to evaluate MS's performance providing a screen resolution of 768 pixels another six subjects in the age range of 40-49 horizontally by 576 pixels vertically. Eight bit years were assessed on the experimental tasks. resolution for each of the red, blue, and green guns allowed for the presentation of a choice Unilateral brain injured patients of 256 simultaneous colours from a palette of Five exservicemen who had sustained a pene- 256'. For achromatic stimuli, the choice was trating missile injury to the brain during the restricted to 256 grey levels. The patient second world war participated in this study. responded to the visual stimuli by touching a All injuries involved unilateral lesions in the glass window placed immediately in front of http://jnnp.bmj.com/ posterior half of the brain, as confirmed by the monitor. A grid of infrared beams tra- operation reports, or CT, or both. These versed the surface of the window at 1 cm patients are particularly valuable for scientific intervals. Each beam was detected by a photo- research because of the well defined, trau- cell, which was tuned to the high frequency matic lesions incurred at a similar age in what infrared signal. The response was recorded by were previously young, healthy brains.38 Two the interruption of one or more of these

exservicemen had a lesion in the right hemi- beams. This touch sensitive screen and the on October 1, 2021 by guest. Protected copyright. sphere, and three had left sided damage. graphics controller were interfaced with a These five patients were selected from a Tandon Plus microcomputer. cohort of 35 men with gunshot wounds, who All tasks employed an equivalent oddity participated in a parallel study on sensory sta- paradigm in which three stimuli were pre- tus, on the basis of their performance on the sented in a vertical alignment. Two stimuli experimental tasks. In line with Ettlinger's were identical and the third differed in the logic, the inclusion criteria were a normal per- appropriate dimension and only appeared at formance on visual recognition tasks and an the top or bottom of the row. Subjects were impaired performance on one or more sensory required to touch the odd one out (top or bot- status tests. Given that the exservicemen are tom stimulus). Each task consisted of 50 tri- now in their 60s and 70s, it was thought pru- als, and a titration procedure was designed to dent to screen them with the mini mental reach a reliable threshold with a minimum of state questionnaire; none showed any evi- trials. Therefore, three consecutive correct dence of disorientation or dementia. choices resulted in an increase in task diffi- culty, whereas a single error made the dis- Normal control subjectsfor the exservicemen crimination easier. Twenty five healthy subjects who were com- parable with the experimental group in age, Shape discrimination educational background, and general intellec- Shape discrimination was assessed with a tual abilities served as controls. None showed modification of the task of Efron.41 Two black Ettlinger revisited: the relation between agnosia and sensory impairment 353

squares and a rectangle or two black rectan- Grey discrimination J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.58.3.350 on 1 March 1995. Downloaded from gles and a square were presented in a vertical The same procedure as that used for colour row, 5 cm apart on a white background (24 discrimination was adopted. Three squares candelas (cd) m-2), and subjects were asked (5 x 5 cm) were presented against a black to pick the odd one out. The black squares background, two of which were of identical were always 5 cm x 5 cm and initially the luminance and the third different. Each trial rectangles had a width to height ratio of consisted of a randomly selected grey of 2-25:1. To establish threshold, this ratio was between 5-5 cdm-2 and 10 0 cdm-2 and one titrated over 50 trials as described earlier, that was randomly either darker or lighter by where the difference in the width between the the appropriate luminance difference. The square and the rectangle was increased or difficulty of the discrimination was succes- decreased by 50% for one error or three con- sively altered by a 50% increase or decrease in secutive correct choices respectively. The area the difference in luminance. The first discrim- of the squares was always identical to that of ination was between greys that differed by 5-0 the rectangles. To prevent the task being cdm-2, followed by 50 trials where the diffi- solved by vernier offset the lateral position of culty level was titrated as described earlier. each discrimination was randomly varied by Performance was measured in cdm-2. up to 2 5 cm, with the constraint that the edges of the shapes were never aligned. Shapefrom movement Performance was measured in mm. The discrimination of shape from movement was assessed by a procedure identical to the Location discrimination shape discrimination task described earlier. Subjects were required to discriminate the Shape was now generated, however, by the position of a dot in a circle. Three black outline coherent lateral movement of every pixel in a circles, radius 7-5 cm, on a white background random dot pattern, which defined a shape, (24 cdm- ), each containing a 3 mm diameter against a random dot background where each black dot were used. In one circle the dot was pixel composing the background moved in a located centrally; in the others it was dis- random fashion. The shape and background placed along either the vertical or horizontal were composed of random dot patterns with a axis. The three circles were presented in a ver- density of 25%. Performance was measured in tical row, 1 cm apart, where each was dis- mm. placed laterally in the random manner described earlier (shape discrimination) to Correlated movement prevent the task being solved by the vertical Three stationary vertically aligned squares (5 alignment of the central dot. Subjects received x 5 cm) were presented against a black back- 50 trials and the degree of displacement was ground. Each square contained a number of titrated as described earlier. Performance was randomly arranged white dots, each dot corre- measured in mm. sponding to a single pixel. All the pixels within the squares could be either switched on Colour discrimination and off randomly, giving the appearance of The ratio ofthe output ofthe red gun to that of stationary white dots twinkling on and off; or a the green gun at isoluminance was deter- proportion of the pixels, against this random mined after gamma correction. Luminance switching, could be switched on and off in http://jnnp.bmj.com/ measurements were made with a Minolta such a manner as to give the percept of white luminance meter LS-1 10. Chromatic stimuli dots moving laterally across the square. Two were then produced by the addition of red squares could be filled with laterally drifting and green, in varying proportions, while white dots, and the remaining square with maintaining a constant luminance of 7-8 "stationary twinkling" dots; or vice versa. The cdm-2. This resulted in 80 isoluminant hues dot density remained constant at 10%. After

that ranged, in roughly equal steps on the CIE three consecutively correct responses, the per- on October 1, 2021 by guest. Protected copyright. chromaticity diagram, between the red and centage of drifting dots compared with sta- green produced by the red and green guns tionary dots within a single square decreased alone. A vertical row of three square patches by a factor of 5%. A single incorrect response (5 x 5 cm) was presented against a white increased the amount of drifting dots com- background (8-8 cdm-2), where two patches pared with stationary dots by a similar factor. were identical and the third differed in hue. Each trial consisted of a central, randomly Texture discrimination selected hue with an equivalent one at the top For texture discrimination, squares (5 x 5 or the bottom of the display, and one that dif- cm) of random white dots, each a single pixel, fered in chromaticity by the appropriate num- against a black background were used, where ber of hue steps towards either the red or the the texture was defined by the density of ran- green, which was again randomly determined. dom dots: The initial discrimination was The difficulty of the discrimination was suc- between densities of 5% and 14%. Task diffi- cessively altered by a 50% increase or culty was varied according to the titration pro- decrease in the number of chromaticity steps. cedure described earlier. An increase in The first discrimination was between hues difficulty was accomplished by increasing the that differed by 20 hue steps, followed by 50 less dense pattern while decreasing the more trials where the difficulty level was titrated as dense pattern by 1%. The reverse procedure described earlier. Performance was measured decreased the difficulty. The textures were in arbitrary hue steps. again separated by 2-5 cm in a vertical row. 354 Haan, Heywood, Young, Edelstyn, Newcombe

Table 1 Performnance of three (prosop) agnosic patients no nosic, with overt recognition of familiar J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.58.3.350 on 1 March 1995. Downloaded from and 25 normal control subjects on standardface and object recognition tasks faces. Patient PH also has a milder form of object agnosia, and MS has severe problems Familiarface recognising objects visually. Over the years recognition Object naming (max score = 20) (max score = 36) this object naming task has been given several times to and he is consistent in the PH 0 20 MS, items NR 0 13 that he can and those that he cannot identify. MS 0 8 Patient NR's object recognition ability is Control subjects 17-9 (1 6) 34-1 (1-8) somewhere between that of PH and MS. Values in parentheses are SD. Individual thresholds for each of the sen- sory tasks were calculated by taking the mean of the last three peaks (best score) in the per- formance curve. To evaluate the performance Five examples of each density ensured that on of patients, means and SDs of the control a single trial the two identical discriminanda groups were calculated subsequently for all were different examples of identical density. eight tests. For each patient the deviation This made it impossible to select the odd one from the appropriate control mean is indi- out merely by noting that the position of the cated at three levels of significance: 0 05 (z > dots was identical in two of the patterns. 1-65), 0-01 (z > 2 33), and 0-001 (z > 3-10). Performance was measured in % density. From table 2 it is apparent that PH performs within the normal range on the tasks for Line orientation shape, colour, grey, line orientation, and cor- The resolution of the colour monitor pre- related movement discrimination. In compari- vented its use for this task, as the degree of son with the age matched control subjects he jaggedness of the lines contained information is impaired at the perception of location, tex- about the angle. Therefore, lines were drawn ture, and movement. This is clearly a minimal with a printer and assembled in a series of reduction, however, in perceptual ability as books. Three lines were presented, vertically his performance on these tasks fall in the aligned down the centre of the page. The pro- range of the normal subjects used as controls cedure of line offset was also used (as previ- for MS who are, on average, some 20 years ously described). The same oddity paradigm older than PH. Patient NR's performance on and binary response procedure were used. the tasks for the perception of colour, grey, Randomisation and presentation of individual line orientation, and correlated movement are angles was controlled by the computer. Each all within the normal range. He has some diffi- response was entered into the computer, and culty on tasks for shape, location, and texture on that basis, the next presentation was discrimination, but as was the case with PH selected and displayed. Three isolated lines, these problems are relatively mild, as his per- each of 40 mm in length, were presented. formance often falls in the range of the older Two of the lines made the same angle with the (aged 40-49) control subjects. Patient NR is, (imaginary) horizontal and one line made a however, much more severely impaired in the larger or smaller angle. Three consecutively perception of movement; his score is more correct responses resulted in the angular dif- than 39 SDs above the control mean. ference being decreased by 50%, whereas a Regarding MS, we note that he has cerebral http://jnnp.bmj.com/ single incorrect response resulted in this being achromatopsia. This problem has been exten- increased by 50%. Eight sizes of angle were sively investigated in several studies,2642 and used (32, 16, 14, 12, 10, 8, 7, 6, 5, 4, 3, 2, we refer to these publications for details con- and 1 degree). cerning his colour perception deficit. On all other sensory tasks, MS performs within the range of the age matched controls. Results Results in table 3 show that the five on October 1, 2021 by guest. Protected copyright. Table 1 summarises the performance of the selected men with gunshot wounds have no three agnosic patients on the visual recogni- problems in visually identifying familiar faces tion tasks. As noted in previous investigations, or objects. On the whole, their performance is it is clear that all three are densely prosopag- either equal to, or better than, that of healthy control subjects. This is not surprising as the absence of visual recognition problems was Table 2 Performance on visual-sensory tasks for three agnosic patients and two control one of the selection criteria. groupsfrom different age ranges PH NR Controls 1 MS Controls 2 Table 3 Performance offive exservicemen with unilateral Age band 28 29 20-29 42 40-49 brain lesions and 25 control subjects on standard face and No of subjects 1 1 6 1 6 object recognition tasks Shape (mm) 1.1 2.5*** 0 74 (0 36) 1-4 1-28 (0-78) Location (mm) 2-8*** 4-1*** 0 99 (0-49) 2 1 2-64 (2 44) Familiarface Colour (arbitrary: see text) 1-0 1-0 1-0 (-) A 1-0 (-) recognition Object naming Grey (candelas) 0-98 0 95 1-10 (0-36) 1-21 1-03 (0 68) (max score = 20) (max score = 36) Movement (mm) 1-8** 10 9*** 1-06 (0 25) 1-2 1-84 (0 93) Correlated movement AT 19 33 (% correct) 5 0 5 0 50 (-) 5-0 7 50 (5 59) EH 18 34 Texture (% difference) 1-7* 2.3*** 1-14 (0-31) 3-0 2-33 (1-63) SJ 18 34 Line orientation (degrees) 2-7 2-0 1-60 (0 89) 2-7 1-83 (1-17) RR 18 34 Jo 17 34 *p < 0-05 (z > 1-65); **p < 0-01 (z > 2 33); ***p < 0-001 (z > 3-10). Control subjects 17 9 (1-6) 34-1 (1-8) In all cases larger scores indicate poorer performance; values in parentheses are SD; A = achro- matopsic. Values in parentheses are SD. Ettlinger revisited: the relation between agnosia and sensory impairment 355

Table 4 Dataforfive selected exservicemen with unilateral missile wounds and appropriate control group on visual- J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.58.3.350 on 1 March 1995. Downloaded from sensory tasks Patient Gunshot wound AT EH SJ RR _o controls Shape (mm) 5-4*** 2.9*** 1-8 2-3** 2-1* 1 12 (0 44) Location (mm) 5-8*** 4.9*** 12-5*** 5.3*** 3.4** 1-53 (0 80) Colour (arbitrary: see text) A 4-3*** 1-7* 1.0 1-3 1 10 (0 30) Grey (candelas) 2.0* 3-4*** 2-6*** 1-5 1 1 1-09 (0 54) Movement (mm) 24 0*** 3-8** 9-3*** 4.0** 3.4* 1-66 (0-85) Correlated movement (% correct) 41-7*** 15-0 NT NT NT 7-53 (544) Texture (% difference) 6-3*** 2-3 6-3*** 3.0* 1 0 1-48 (0 66) Line orientation (degrees) 2-7 2-7 5-7*** 3-7 6-7*** 2-30 (0 95) *p<005 (z> 1-65); **p < 0-01 (z> 233); ***p < 0001 (z> 310). In all cases larger scores indicate poorer performance; values in parentheses are SD; NT = not tested; A = achromatopsic.

The other selection criterion was a severe recent advances in knowledge of the organisa- impairment on one or more of the sensory tion of the visual pathways,'8 19 indicating perception tasks. As stated in the subjects sec- which aspects of sensory processing can be tion, these five exservicemen were selected disrupted selectively. Therefore, the experi- from a group of 35 men with unilateral missile mental tasks form a much more comprehen- injuries to the brain, but this does not imply sive assessment of sensory status than was that the other 30 men did not have any prob- available to Ettlinger. We have also used a dif- lems at all on these tasks. The point is only ferent procedure to analyse the data, which that these exservicemen suffice to demon- allows for individual assessment of separate strate our case. Inspection of table 4 shows functions. that for each of the eight tasks there is at least Overall, the results strongly support one man with a severe impairment. For Ettlinger's original claim. For the agnosic instance, AT performed 9 7 SDs above the patients, in comparison with the appropriate control mean on the shape perception task; SJ age matched control group, PH showed sub- performed 13-7 SDs above the control mean tle perceptual problems on some of the sen- on the location task; AT was completely sory tests, NR was in the main slightly worse achromatopsic; EH performed 4-3 SDs above than PH but only severely impaired in the the control mean on the grey perception task; perception of movement, and MS performed AT and SJ performed 7-3 SDs above the con- in the normal range on all experimental tasks trol mean on the texture perception task; AT apart from colour perception. These problems performed 26&3 SDs above the control mean are, however, insufficient to explain the pro- on the movement perception task; JO nounced recognition problems, as the exser- performed 4-6 SDs above the control mean vicemen with unilateral brain lesions who do on the line orientation task; and AT per- not have recognition deficits are, at least, just formed 6-3 SD above the control mean on the as impaired on these sensory tasks as the correlated movement task. Moreover, all five agnosic patients. For example, the patient AT of these exservicemen showed impairments on is also achromatopsic but this does not have at least four of the eight sensory tasks, and in any detrimental effects on his ability to recog- http://jnnp.bmj.com/ one case (AT) seven out of eight tasks were nise objects. Only in the case of NR could it performed at an impaired or grossly impaired be argued that his movement perception level. Yet all the five were able to recognise problem is excessive even compared with the faces and objects. patients with gunshot wounds. A possible objection that could be raised is that the discriminative power of some of the

Discussion sensory status tests might have been too small on October 1, 2021 by guest. Protected copyright. In this study we assessed visuosensory status to show subtle perceptual deficiencies. For in three agnosic patients and five exservice- instance, all three agnosic patients perform at men with unilateral brain injuries. Our aim ceiling on the test for correlated movement. was to reconsider the issue raised by Ettlinger The data presented in table 4, however, argue in the 1 950s that, contrary to suggestions clearly against this possibility. At least one made by Bay et al6 7 and others, sensory patient with gunshot wounds does show a impairments are insufficient to explain the severe problem on this task and the control recognition deficits in certain cases. subjects, as a group, are not perfect either. Yet Ettlinger's study has been criticised for two none of these subjects have any problem reasons. Firstly, he did not include a truly recognising visually presented stimuli. The object agnosic patient in his series. Our pre- same point applies to all of our tasks. sent study was carried out with three patients, Another objection that could be put for- all of whom were fully prosopagnosic and at ward is that the screening tasks for visual least one of whom (MS) was severely object recognition problems are too easy to allow agnosic. The second criticism concerned the detection of subtle impairments in the gun- limited range of sensory functions assessed by shot wound population. Although we did not Ettlinger and the use of a composite score report any examples here, there are, however, instead of evaluating the separate sensory abil- several cases from the gunshot wound group ities individually. The sensory functions tested who do show clear evidence of recognition in this study were selected on the basis of problems on these tasks. In another study 356 Haan, Heywood, Young, Edelstyn, Newcombe

14 Pallis CA. with the gunshot wound population,43 we Impaired identification of faces and places with J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.58.3.350 on 1 March 1995. Downloaded from agnosia for colours. 7 Neurol Neurosurg Psychiatry 1955; have reported impairments with the familiar 18:218-24. face recognition task. 15 Bodamer J. Die Prosopagnosie. Archiv fur Psychiatrie und Nervenkrankheiten 1947;179:6-53. A final point is the possibility that one 16 De Renzi E. Current issues on prosopagnosia. In: Ellis selective sensory deficit is, as such, not suffi- HD, Jeeves MA, Newcombe F, Young A, eds. Aspects of face processing. Dordrecht: Martinus Nijhoff, 1986: cient to cause agnosia but that a certain con- 243-52. stellation of sensory impairments does result 17 Ellis HD, Young AW. Are faces special? In: Young AW, Ellis HD, eds. Handbook of research on face processing. in a visual recognition disorder. This sugges- Amsterdam: North Holland, 1989:1-26. tion seems highly unlikely, however, in view of 18 Cowey A. Sensory and non-sensory visual disorders in man and monkey. Philos Trans R Soc Lond Biol 1982; a gunshot wound patient like AT. He per- B298:3-13. forms very badly on most of the sensory tasks 19 Cowey A. Aspects of cortical organization related to selec- tive attention and selective impairments ofvisual percep- but has no visual recognition problems. tion. In: Posner MI, Marin OSM, eds. Attention and Thus we conclude that, whereas patients performance. Vol XI. New Jersey: Lawrence Erlbaum, 1985:41-62. with often experience visual 20 Desimone R, Albright TD, Gross CG, Bruce C. Selective perceptual problems, these impairments are properties ofinferior temporal neurons in the macaque. J Neurosci 1984;4:2051-68. not always sufficient to explain the recogni- 21 Heywood CA, Wilson B, Cowey A. A case study of cortical tion deficit. In this particular study, it is sug- "blindness" with relatively intact achromatic discrimina- tion. J Neurol Neurosurg Psychiatry 1987;50:22-9. gested that at a functional level both PH and 22 Zihl J, Von Cramon D, Mai N. Selective disturbances of MS have a higher order visual recognition movement vision after bilateral brain damage. Brain 1983;106:313-40. problem. We do not claim that their problems 23 Hyvarinen J, Rovamo J. Acquired blindness for achromatic are the same. For example, several studies stimuli. Documenta Ophthalmologica Proceedings Series 198 1;30:94-9. have shown covert recognition for PH27 28 but 24 Rovamo J, Hyvarinen J, Hari R. Human vision without not for MS.34 Thus there may well be differ- luminance contrast system: selective recovery of the red- green colour contrast from acquired blindness. ences between higher order deficits. Our point Documenta Ophthalmologica Proceedings 1982;33:457-66. is only that low level problems, singly or in 25 Warrington EK. Visual deficits associated with occipital lobe lesions in man. In: Chagas C, Gattass R, Gross C, combination, do not create the phenomena eds. Handbook of clinical neurology. Vol 1 (45), Clinical found. Our re-examination, with modem Neuropsychology. Amsterdam: Elsevier, 1987:247-61. 26 Heywood CA, Cowey A, Newcombe F. Chromatic dis- methods, of the issues raised by Ettlinger' crimination in a cortically colour blind observer. thus confirms and underlines the conclusion European Journal ofNeuroscience 199 1;3:802-812. 27 De Haan EHF, Young AW, Newcombe F. Faces interfere he reached. with name classification in a prosopagnosic patient. Cortex 1987;23:309-16. This work was supported by MRC grant G 8804850 to F N 28 De Haan EHF, Young AW, Newcombe F. Face recogni- and AWY and MRC grant G 9114877N to FN, Richard tion without awareness. Cognitive Neuropsychology Kerr, and E H F De H. 1987;4:385-415. 29 De Haan EHF, Young A, Newcombe F. Covert and overt recognition in prosopagnosia. Brain 1991;114:2575-91. 30 Benton AL, Van Allen MW. Tests of visualfunction. Iowa: Oxford University Press, 1973. 31 Young AW, De Haan EHF. Boundaries of covert recogni- tion in prosopagnosia. Cognitive Neuropsychology 1988; 1 Ettlinger G. Sensory deficits in visual agnosia. J Neurol, 5:317-36. Neurosurg Psychiatry 1956;19:297-308. 32 Ratcliff G, Newcombe F. Object recognition: some deduc- 2 Jackson JH. Case of large cerebral tumour without optic tions from the clinical evidence. In: Ellis AW, ed. neuritis and with left hemianopia and imperception. Normality and pathology in cognitive functions. London: Royal London-Ophthalmology Hospital Reports 1876; Academic Press, 1982:147-71. 8:434-444. Reprinted in: Taylor J, ed. Selected writings of 33 Newcombe F, Ratcliff G. Agnosia: a disorder of object John Hughlings Jackson Vol II; London: Hodder and recognition. In: Michel F, Schott B, eds. Les Syndromes Stoughton, 1932:146-52. de disconnexion calleuse chez l'homme. Lyon: Collogue 3 Lissauer L. Ein Fall von Seelenblindheit nebst einem International de Lyon, 317-41. http://jnnp.bmj.com/ Beitrag zur theorie derselben. Archiv fur Psychiatrie und 34 Newcombe F, Young AW, De Haan EHF. Prosopagnosia Nervenkrankheiten 1889;21:222-70. and object agnosia without covert recognition. 4 Freud S. On aphasia: a critical study. London: International Neuropsychologia 1989;27: 179-91. University Press, Imago, 1953. 35 Young AW, Newcombe F, Hellawell D, De Haan EHF. 5 Teuber H-L. Alteration of perception and memory in Implicit access to semantic information. Brain Cogn man. In: Weiskrantz L, ed. Analysis of behaviour change. 1989;11: 186-209. New York: Harper and Row, 1986. 36 Snodgrass JG, Vanderwart M. A standardized set of 260 6 Bay E. Disturbances of and their exami- pictures: norms for name agreement, image agreement, nation. Brain 1953;76:515-51. familiarity, and visual complexity. J Exp Psychol Hum 7 Bay E. Agnosie und Funktionswandel. Monographien aus Learn Mem 1980;6:174-215. dem Gesamtgebiet der Neurologie und Psychiatrie 1950; 37 De Haan EHF, Young AW and Newcombe F. on October 1, 2021 by guest. Protected copyright. 73: 1-94. Neuropsychological impairments of face recognition 8 Habib M. Visual hypoemotionality and prosopagnosia units. QJ Exp Psychol 1992;44A:141-75. associated with right temporal lobe isolation. 38 Newcombe F. The processing of visual information in Neuropsychologia 1986;24:577-82. prosopagnosia and acquired dyslexia. In: Osborne DJ, 9 Bruyer R, Laterre C, Seron X, Feyereisen P, Strypstein E, Gruneberg MM, Eiser JR, eds. Research in psychology and Pierrard E, Rectem D. A case of prosopagnosia with medicine, Vol 1. London: Academic Press, 1979:315-22. some preserved covert remembrance of familiar faces. 39 De Haan EHF, Young AW, Newcombe F. A dissociation Brain Cogn 1983;2:257-84. between sense of familiarity and access to semantic 10 Tiberghien G, Clerc I. The cognitive locus of prosopag- information concerning familiar people. European nosia. In: Bruyer R, ed. The neuropsychology offace percep- Journal of Cognitive Psychology 199 1;3:51-67. tion and facial expression. Hillsdale: Lawrence Erlbaum, 40 Newcombe F, Oldfield RC, Ratcliff GG, Wingfield A. 1986:39-62. Recognition and naming of object-drawings by men with 11 Damasio AR, Damasio H, van Hoesen GW. focal brain wounds. J Neurol Neurosurg Psychiatry 197 1; Prosopagnosia: anatomic basis and behavioral mecha- 34:329-40. nisms. Neurology 1982;32:331-4. 41 Efron R. What is perception? Boston studies in the philosophy of 12 Davidoff JB, Matthews WB, Newcombe F. Observations science. New York: Humanities Press, 1968:137-73. on a case of prosopagnosia. In: Ellis HD, Jeeves MA, 42 Mollon JD, Newcombe F, Polden PG, Ratcliff G. On the Newcombe F, Young A, eds. Aspects offace processing. presence of three cone mechanisms in a case of total Dordrecht: Martinus Nijhoff, 1986:279-90. achromatopsia. In: Colour vision deficiencies, Vol V. 13 Campion J. Apperceptive agnosia: the specification and Bristol: Hilger, 1980:130-5. description of constructs. In: Humphreys GW, Riddoch 43 Young AW, Newcombe F, De Haan EHF, Small M, Hay MJ, eds. Visual object processing: a cognitive neuropsycho- DC. Face perception after brain injury: selective impair- logical approach. London: Lawrence Erlbaum, 1987: ments affecting identity and expression. Brain 1993;116: 197-232. 941-59.