Q. Gao, Z. Chen, P. Russell

Working Load and the Stroop Interference Effect

Quanying Gao, Zhe Chen, & Paul Russell University of Canterbury

Although the effect of (WM) load on the degree of distractor and Stroop interference. They found processing has been investigated in a number of paradigms, a common a negative correlation between the feature in prior research is that the target and distractors pertain to different two. Relative to those with small WM objects. The present experiments examine the effect of WM load on capacity, the participants with large WM distractor interference when the relevant and irrelevant features belong to capacity showed less Stroop interference. the same object. In Experiment 1, participants saw stimulus displays that Related findings were observed by consisted of a memory set followed by a Stroop stimulus, whose colour Unsworth, Schrock and Engle (2004). and meaning were either unrelated or incongruent. The task was to make In their study, the participants with large a speeded response to the colour of the target while holding either one WM capacity were not only faster in or six digits in memory. Although a signifi cant Stroop interference effect performing antisaccade tasks, but also was found, its magnitude was not infl uenced by WM load. Experiment 2 less likely to make refl exive saccades manipulated the size of attentional focus in addition to WM load and the to an exogenous cue on the wrong response congruency between the relevant and irrelevant features of Stroop side of the screen. Manipulating WM stimuli. Again, there was a strong Stroop interference effect, but no effect load directly, Lavie and her colleagues of WM load or attentional focus. These results suggest that the effect of (Lavie, Hirst, de Fockert, & Viding, WM load on selective may be more complex than was previously 2004) required participants to retain conceived. They also emphasize the importance of stimulus structure in either one digit (low WM load) or understanding the effect of WM load on selective attention. several digits (high WM load) while performing a letter discrimination task. Distractor interference was larger in the high WM load condition than in the Visual is intrinsically distractors can be reduced by increasing low WM load condition, suggesting selective. A typical natural scene the spatial separation between a target that high WM load impaired distractor consists of numerous objects and and distractors (B. A. Eriken & C. inhibition. events. However, only a subset of them W. Eriksen, 1974; C. W. Eriksen & These and other similar results is relevant to our behavioural goals. Hoffman, 1973), cuing participants’ are consistent with Lavie’s (Lavie, Given that the visual system has a attention to the location of the target 2005; Lavie et al., 2004) load theory limited capacity to process multiple before the target appears (C. W. Eriksen of attention. According to the theory, objects at any given time (Broadbent, & Hoffman, 1973; Yantis & Jonides, perceptual resources are limited at any 1958; Neisser, 1967), it is critical that 1990), grouping targets and distractors given time, and perception proceeds only relevant information is processed into different perceptual groups (Chen automatically until all resources are while irrelevant information is either & Cave, 2006a; Harms & Bundersen, used up. Furthermore, there are two suppressed or ignored. The question 1983; Kramer & Jacobson, 1991), or mechanisms in selective attention: a is: how does the visual system select by increasing the perceptual load of a passive perceptual selection mechanism relevant information among competing stimulus display (Lavie, 1995). that prevents distractors from being distractors? Recent research suggests that processed when the perceptual load is One way to understand selective working memory (WM) load may also high, and an active cognitive control attention is to identify the factors play an important role in modulating mechanism that requires WM to that modulate it under various the degree of distractor processing. inhibit distractor interference when the circumstances. Past research has shown Kane and Engle (2003) examined the perceptual load is low and distractors that interference from task irrelevant relationship between WM capacity

• 146 • New Zealand Journal of Vol. 36, No. 3, November 2007 Working Memory and Stroop Interference are perceived. As a result, whereas a effect of WM on distractor interference to respond to the colour of the Stroop high perceptual load reduces distractor is more complex than was proposed in stimulus on the basis of either a single processing because of the unavailability the load theory of attention. feature (e.g., whether a bar which was of resources, a high WM load results in Although WM load had been situated above or below the Stroop large distractor interference due to the manipulated directly in selective stimulus was white - the low load lack of resources to inhibit distractors. attention tasks (e.g., Lavie et al., 2004; condition) or a conjunction of features However, not all studies show Lavie & de Fockert, 2005), previous (e.g., whether the bar was white and an inverse relationship between WM studies used stimulus displays where above the Stroop stimulus - the high load and the efficiency of selective the relevant and irrelevant information load condition). She found that although attention (Chen & Chan, 2007; Logan, belonged to separate objects. There is RT was substantially slower in the high 1978; Woodman, Vogel, & Luck, reason to believe that WM load may load than in the low load condition, the 2001). For example, both Logan (1978) infl uence visual selection differently magnitude of the Stroop interference and Woodman et al. (2001) reported when the relevant and irrelevant effect was comparable in the two comparable visual search slopes when information pertain to the same object. situations. In a subsequent experiment their participants performed visual It has been shown that the effect of (Experiment 4), the size of attentional search with or without a concurrent perceptual load on distractor processing focus was manipulated by varying the memory task. Han and Kim (2004) differs as a function of the nature of size of a cue, which was a rectangle also demonstrated that the effect of stimulus displays (e.g., Chen, 2003; presented before the onset of the target WM load in the effi ciency of visual Lavie, 1995; Lavie & Cox, 1997). display. The Stroop interference effect search depended on the involvement For example, in a study conducted by was greater when the cue was large of executive WM. They showed that Lavie and Cox, participants searched rather than when it was small. These whereas increasing WM load impaired for a target among irrelevant distractors results suggest that differences in visual search effi ciency when executive that were either homogeneous (the stimulus structure infl uence participants’ WM was involved (e.g., counting low perceptual load condition) or processing strategies (Garner, 1970; backwards in threes from a three-digit heterogeneous (the high perceptual 1974; Garner & Felfoldy, 1970), which number), it had no effect when simple load condition). Distractor interference in turn modify the effect of perceptual maintenance of verbal information was from a critical incompatible distractor load on distractor interference. required (e.g., holding seven digits in was larger in the low load condition In light of the above fi ndings, it is memory). than in the high load condition. This unclear whether the effect of WM load Chen and Chan (2007) further result is consistent with the load theory on distractor processing will also be proposed that the size of attentional of attention. It suggests that a high modulated by the spatial relationship focus may have played an important perceptual load decreases distractor between the relevant and irrelevant role in many previous experiments. interference. information in a target display. In They noted that because WM load However, a similar effect was not two experiments reported here, we was usually manipulated by varying found when the target was a Stroop used Stroop stimuli to explore the the number of items held in memory, a stimulus (Chen, 2003). In a typical relationship between WM load and high WM load was typically associated experiment that uses Stroop stimuli selective attention. To control for with a wide attentional focus while a (Stroop, 1935), participants make a the extent of attentional focus, we low WM load was typically associated speeded response to the colour of a employed a spatial precue before the with a small attentional focus (e.g., word. The relationship between the presentation of the target display. Lavie et al., 2004). In other words, WM colour and meaning is manipulated In Experiment 1, participants saw a load was confounded, at least in some so that they can be congruent (e.g., memory set that consisted of either one studies, with the size of attentional the word RED written in red ink), or six digits. This was followed by a focus, which is known to infl uence the incongruent (e.g., the word GREEN brief cue, which indicated the location magnitude of distractor interference written in red ink), or neutral (e.g., the of the Stroop stimulus, whose colour (Chen, 2000; 2003; Eriksen & St. James, word SHOE written in red ink). The and meaning were either unrelated or 1986; LaBerge, Brown, Cater, Bash, & standard finding is that RT is faster incongruent. Participants made two Hartley, 1991). To determine whether in the congruent condition than in the responses on each trial. The fi rst was attentional focus could contribute to neutral condition, which in turn is faster a speeded response to the colour of the observed WM load effect, Chen and than in the incongruent condition. The the target. Upon response, a memory Chan manipulated both factors within slower RT in the incongruent condition probe appeared. Participants indicated the same paradigm. They found that a is termed the Stroop interference effect whether the probe had appeared in the high WM load resulted in ineffi cient (see MacLeod, 1991, for a review), and memory set on that trial. Experiment 2 distractor inhibition only when it was the magnitude of the Stroop interference manipulated the size of attentional focus coupled with a large attentional focus. effect is taken to indicate the degree in addition to WM load and the response When attentional focus was equated of distrator processing. Using Stroop congruency between the colour and across different experimental conditions, stimuli in a go/nogo paradigm, Chen meaning of the Stroop stimuli. Together, the WM load effect became negligible. (2003, Experiment 3) manipulated these experiments examined the role Together, these fi ndings suggest that the processing load by requiring participants of WM load and the size of attentional

New Zealand Journal of Psychology Vol. 36, No. 3, November 2007 • 147 • Q. Gao, Z. Chen, P. Russell focus on distractor inhibition when the corresponding length (“vvv”, “oooo”, by pressing one of four labelled keys distractor is part of the same object as “sssss”, “nnnnnn”) that extended on the keyboard (“z” for red, “x” for the target. between 3.060 and 6.020 horizontally. green, “,” for yellow and “.” for blue). The target always appeared at the Upon response, the memory probe Experiment 1 location indicated by the cue. Four appeared on the screen. It remained possible colours were associated there until the participants responded Method with the target stimuli. They were: by depressing one of two labelled Participants. Thirty-six University red (RGB: 100, 0, 0), blue (RGB: 0, keys (“a” for probe-present and “’” for of Canterbury students volunteered 0, 100), green (RGB: 0, 100, 0), and probe-absent responses). The memory to participate in the experiment in yellow (RGB: 100, 100, 0). Each probe was equally likely to be present 1 exchange for payment. All reported specifi c word or letter string could be or absent from the memory set. to have normal or corrected-to-normal in one of three colours except for the Each participant performed 48 vision. The participants were treated in colour that matched the meaning of practice trials. They then completed accordance with the “Ethical Principles the word (e.g. the word “red” and its three blocks of 64 trials for a total of of Psychologists and Code of Conduct” equivalent “vvv” could be blue, green, 192 trials. Whereas both speed and (American Psychological Association, or yellow, but not red). The memory accuracy were emphasized for the 1992), and the research was approved probe comprised a single black digit Stroop task, accuracy was stressed for by the University of Canterbury Human and a question mark. the memory task. Ethics Committee. Design and Procedure. The experiment Results and Discussion Apparatus and Stimuli. All stimuli was a mixed design with WM load were presented on a grey background (high vs. load) as the between-subjects Table 1 shows the data for Experiment using a Pentium-III computer with a variable and response congruency 1. For the Stroop task, RTs longer than 17-inch monitor. E-Prime (Schneider, between the meaning and colour of 2000 msec (about 1% of the total data) Eschman, & Zuccolotto, 2002), a the target (neutral vs. incongruent) were excluded from data analyses. A commercially available experimental as the within-subjects variable. There mixed analysis of variance (ANOVA) program, was used to generate stimuli were equal numbers of incongruent indicated a significant main effect and to collect responses. trials (e.g. “red” written in green ink) of response congruency [F(1,(1, 3 34)4) = η 2 Each trial consisted of a central and neutral trials (e.g. “vvv” written 14.26, p = .30, p < 0.001], with faster fi xation, a memory set, a cue, a target in green ink). RTs on the neutral trials (730.5 msec) than on the incongruent trials (755.5 display, and a memory probe (see The participants were randomly msec). Neither the main effect of WM Figure 1). The fi xation was a white assigned to one of two WM load 0 load nor the interaction between load cross that extended 1.24 horizontally conditions. Each trial started with a and response congruency reached and vertically. The memory set 1000 msec fi xation, followed by a 480 signifi cance [F(1,(1, 34)34) = 1.71,1.71, η 2 = consisted of either one or six black msec blank screen. A memory set was p .05, n.s., for load; and F(1, 34) = 1.1, digits at the centre of the screen. All then presented for either 520 ms in η 2 = .03, n.s., for the interaction]. A digits were randomly selected from 1 to the low load condition or 2000 msec p similar analysis was conducted on the 9 without replacement on a given trial, in the high load condition. Upon the and were written in 36 point Arial font. accuracy data. No signifi cant results offset of the memory set, a blank screen 2 0 were found [F(1, 34) = 1.53, η = .04, The cue, which was located 6.21 left appeared for 520 msec, followed by p n.s., for load; F(1,(1, 334)4) = 11.98,.98, η 2 = or right from the fi xation, was made a cue, which appeared on the left or p of two vertically aligned white bars .05, n.s., for response congruency; and right side of the screen for 120 msec. 2 0 F(1, 34) = 0.05, η = .001, n.s., for the separated by a gap of 2.1 . Each bar Immediately after the offset of the p 0 0 interaction]. was 1.34 in length and 0.30 in width. cue, the target display was shown for The target display consisted of either a 120 msec at the location indicated by Accuracy for the memory task coloured word (“red”, “blue”, “green”, the cue. Participants made a speeded was high, with the mean memory error or “yellow”) or a string of letters of response to the colour of the target rates being 7.1% and 5.7% for the high

Figure 1. An example of a typical trial in the high WM load condition of Experiment 1.

I + 739180 green 4? I

Memory Fixation SOA Memory Set SOA Cue Target Colour Digit Response Probe 1000 msec 480 msec 2000 msec 520 msec 120ms 120 msec Response until Response

• 148 • New Zealand Journal of Psychology Vol. 36, No. 3, November 2007 Working Memory and Stroop Interference

Table 1. (A) Mean Reaction Times (RTs, in Milliseconds) and Error Rates (%E) focus. As we noted earlier, Chen and for the Colour Task and (B) Mean Error Rates for the Memory Task in Chan (2007) observed a positive Experiment 1. Standard Errors are in the Parentheses. correlation between WM load and the I = incongruent; N = neutral. magnitude of distractor interference when WM load co-varied with the A size of attentional focus. Furthermore, Low Load High Load the effect of WM load was eliminated Dependent when the size of attentional focus was I N I N Variable held constant. Interestingly, when WM load was held constant, the magnitude RT 797 (47) 765 (46) 714 (37) 696 (35) of distractor interference differed as a % Error 4.1 (1.0) 4.7 (0.7) 5.8 (1.3) 6.6 (1.3) function of attentional focus. Because the target and distractors belonged to different objects in Chen and Chan, B it was unclear whether similar effects Dependent Low Load High Load would be found when the relevant and Variable irrelevant information pertained to the RT 908 (45.9) 1144 (35.3) same object. Method % Error 5.7 (0.79) 7.1 (1.30) Participants. Seventy-fi ve participants from the same participant pool as before took part in the experiment. and low load conditions, respectively. was due to ineffective manipulation of None had participated in Experiment However, despite the numerical the WM load. To provide converging 1, and none knew the purpose of the difference, the two conditions did not evidence to the results of Experiment experiment. differ statistically [t(34) = .90, d = 0.3, 1, we conducted Experiment 2, using Apparatus and Stimuli. The apparatus n.s.]. Although RT for the memory task a slightly different paradigm. was the same as used in Experiment was not emphasized in the instructions, 1. Only one change was made to the we conducted a t testtest toto seesee ifif therethere waswas Experiment 2 stimuli. Instead of two vertical bars, the cue display consisted of either one any difference in response latencies. Experiment 2 had two goals: To provide small white square of 0.570 (narrow The participants were faster in the low converging evidence for the results of focus) or four identical small white WM load condition (908 msec) than Experiment 1 and to examine the roles squares located at the four corners of an in the high WM load condition (1144 of WM load and the size of attentional imaginary square that extended 8.500 msec) [t(34) = 4.02, d = 1.34, p < .001]. focus on Stroop interference in the (wide focus). As in Experiment 1, the This result made sense. It is consistent same paradigm. We chose to increase cue was equally likely to appear at the with prior research (e.g., Sternberg, the sensitivity of the experiment by left or right side of the screen, with 1966) that participants took longer to increasing the number of participants its centre at the same location as the determine whether the probe digit was rather than the number of digits in the centre of the target in the subsequent in the memory set when they had to memory set for two reasons. First, a display. search through six digits rather than power analysis (Cohen, 1988) based on one digit in memory. the effect size of the participants’ RTs in Design and Procedure. WM load and Consistent with prior research, a the low and high memory load tasks of attentional focus were manipulated reliable Stroop interference effect was Experiment 1 indicated that the power in such a way that there were three found. Relative to the neutral trials, the for detecting the memory load effect different load/focus combinations: a participants were slower to respond to would be greater than .80 if more than high load with a narrow attentional the colour of the target stimulus when 20 participants were used in each group focus (the high-narrow condition), it was incompatible with the meaning. in Experiment 2. Second, keeping the a low load with a narrow attentional Interestingly, the magnitude of the same number of digits would make the focus (the low-narrow condition), and Stroop interference effect did not differ level of memory load in our experiment a low load with a wide attentional as a function of WM load, suggesting comparable to that of Lavie et al. focus (the low-wide condition). Within that WM load played a negligible role (2004), who showed that one vs. six each combination, there were an equal in the suppression of meaning in a digits was a sensitive manipulation of number of neutral and incongruent Stroop colour word stimulus. low vs. high WM load. trials. This design allowed us to compare the effect of WM load (the high-narrow However, because memory error In addition to manipulating WM condition vs. the low-narrow condition) rates did not differ signifi cantly between load and response congruency, we and the effect of attentional focus (the the two WM load conditions, it is also varied the size of the cue to low-narrow condition vs. the low-wide possible that the lack of a load effect induce participants in different groups condition) directly. on Stroop interference in Experiment 1 to adopt different sizes of attentional

New Zealand Journal of Psychology Vol. 36, No. 3, November 2007 • 149 • Q. Gao, Z. Chen, P. Russell

Figure 2. Examples of stimulus displays from Experiment 2

+ 739180 green 4?

Memory Digit Fixation Memory Set Cue Target Colour SOA SOA Probe Response 1000 msec 2000 msec 120ms 120 msec Response 480 msec 520 msec until

Response

The participants were randomly compared participants’ mean RT data A similar analysis was performed assigned 25 each to the three load/focus in the low-narrow and high-narrow on participants’ mean RT data in the conditions. Whereas the load/focus conditions, even though the condition low-narrow and low-wide conditions manipulation was a between-subjects by response congruency interaction to investigate the effect of attentional variable, response congruency was in the omnibus ANOVA was not focus on Stroop interference. The a within-subjects variable. All other signifi cant. A 2x2 mixed ANOVA with response congruency effect was aspects of the experiment were identical response congruency as the within- significant [862.5 msec for the to those of Experiment 1. subjects variable and WM load as the incongruent and 833.5 msec for the between-subjects variable revealed neutral trials, F(1,(1, 446)6) = 88.37,.37, η 2 = Results and Discussion p a significant Stroop interference .15, p < .01]. However, there was no The data are illustrated in Table 2. effect [803.5 msec and 784 msec for signifi cant effect of attentional focus Three participants’ data were not η 2 the incongruent and neutral trials, [F(1,(1, 446)6) = 22.40,.40, p = .05, n.s.] or included in the analyses because a large η 2 η 2 respectively, F(1,(1, 446)6) = 110.35,0.35, p = the interaction [F(1,(1, 446)6) = 22.22,.22, p proportion of their RT trials (over 30% .18, p < .01]. Neither the main effect = .05, n.s.]. of the total trials) were longer than the η 2 of WM load [F(1,(1, 446)6) = 00.27,.27, p = For the memory task, we conducted cut-off score of 2000 ms. .006, n.s.] nor the interaction [F(1, two separate one-way ANOVAs for the For the Stroop task, a mixed η 2 46) = 0.89, p = .02, n.s.] reached three load/focus conditions. Signifi cant ANOVA (with the three load/focus signifi cance. conditions as the between-subjects factor and response congruency as the within-subjects factor) on participants’ mean RTs revealed a signifi cant main Table 2. (A) Mean Reaction Times (RTs, in Milliseconds) and Error Rates (%E) for the Colour Task and (B) Mean Error Rates for the Memory Task in effect of response congruency [F(1,(1, Experiment 2. Standard Errors are in the Parentheses. η 2 69) = 13.55, p = .16, p < .001], with I = incongruent; N = neutral. faster RT on the neutral trials (812 msec) than on the incongruent trials A (839.7 msec). Neither the main effect High - Narrow Low - Narrow Low - Wide of condition nor the condition by response congruency interaction was Dependent η 2 Variable I N I N I N signifi cant [F(2,(2, 69)69) = 2.62,2.62, p = .07, n.s., for condition; and F(2,(2, 69)69) = 11.31,.31, η 2 RT 794 (27) 769 (28) 813 (37) 799 (38) 912 (44) 868 (38) p = .04, n.s. for the interaction]. A similar analysis was conducted on the % Error 3.8 (0.5) 3.7 (0.7) 4.2 (1.1) 3.9 (0.6) 2.4 (0.4) 2.2 (0.3) accuracy data. No signifi cant effects η 2 were found [F(2,(2, 669)9) = 11.72,.72, p = .05, η 2 B n.s. for condition; F(1,(1, 669)9) = 00.25,.25, p = .004, n.s. for response congruency; Dependent η 2 High - Narrow Low - Narrow Low - Wide and F(2,(2, 669)9) = 00.04,.04, p = .001, n.s. Variable for the interaction]. To assess the effect of WM load RT 1321 (82) 865 (53) 817 (40) on Stroop interference directly, we % Error 8.9 (1.4) 4.7 (0.8) 4.3 (0.9)

• 150 • New Zealand Journal of Psychology Vol. 36, No. 3, November 2007 Working Memory and Stroop Interference effects were found in both accuracy [F(2,(2, General Discussion A more plausible explanation for 69) = 6.53, η 2 = .16, p < .01] and RT our results is the special relationship p Using Stroop stimuli in which the relevant [F(2, 69) = 21.16, η 2 = .38, p < .001]. between the relevant and irrelevant p and irrelevant information belonged to a Post-doc Tukey’s Honestly Signifi cant single object, we investigated the effect information of the target stimulus that Differences tests further indicated of WM load on distractor processing in we used in our experiments. Stroop higher error rates and longer reaction two experiments. Our results show that stimuli are unique in that the relevant times in the high-narrow condition than the level of WM load had little effect on and irrelevant features belong to the in both the low-narrow and low-wide the magnitude of Stroop interference. same object. Research on object- conditions (all p values were smaller How do we explain these results? based attention has indicated that the than .05) with no signifi cant difference processing of one feature facilitates There are two possible interpret- between the latter two conditions. These that of other features which belong to ations, one concerning the size of results indicate that our manipulation of the same object (Chen, 1998; Duncan, attentional focus, and the other the the WM load was effective. 1984; Egly, Driver, & Rafel, 1994), and nature of the target stimulus. With regard that participants process task irrelevant The most important finding of to attentional focus, previous studies features even though doing so impairs Experiment 2 was the negligible effect showed that when the size of attentional their behavioural goals (Chen, 2005; of WM load on the magnitude of the focus was equated by a precue or by Chen & Cave, 2006b; Remington & Stroop interference effect. Although the sequential presentation of items in Folk, 2001). In terms of the present an increase in WM load led to a higher the memory set, the effect of WM load study, these results suggest that when memory error rate in the high-narrow became negligible (Chen & Chan, 2007; attention selects one dimension of an condition than in the low-narrow Logan, 1978). However, although the object, it is impossible to limit the condition, the magnitude of Stroop present results are consistent with these processing to only the task relevant interference was comparable in the two previous fi ndings, we consider it unlikely feature dimension. All the other features conditions. Given the large effect sizes that the control of attentional focus was (η 2 = .16 and η 2 = .38 for accuracy and that belong to the same objects are p p a determining factor in the present processed. RT respectively) and power (over .8) of experiments. This is because Lavie the WM load effect in the experiment, and de Fockert (2005) demonstrated In addition to meaning being an it is unlikely that our results are due to that WM load could affect distractor integral part of a Stroop stimulus, the a lack of statistical power. interference even when the size of fact that reading is a highly practised It is worth noting that the amount of attentional focus was roughly the same. activity for most people including the Stroop interference was also comparable In one experiment, their participants participants in the present experiments in the low-narrow and low-wide performed an orientation discrimination may also contribute to our observed conditions. In other words, the size of task either with or without a concurrent results. It is possible that because attentional focus had a negligible effect memory task (i.e., single- and dual-task reading is an automatic process that on the degree of distractor processing conditions, respectively). The target was requires little resources (Keele, 1972; in the present paradigm too. This always presented among a number of Posner & Snyder, 1975; but also see fi nding may seem to be inconsistent distractors. On half of the trials, one of Kahneman & Chajczyk, 1983), unless with previous research (e.g., Chen, the distractors had a unique colour (i.e., WM resources are totally depleted, 2003) which showed larger Stroop a colour singleton). On the remaining variations in WM load may not lead to interference when the attentional focus trials, all the distractors were the same observable effects on the magnitude of was wide rather than when it was narrow. colour. Because colour singletons are the Stroop interference effect. Related There was one potentially important known to capture attention (see Yantis, phenomena have been observed in feature methodological difference between the 2000, for a review), the degree of search, which does not require attention two studies. Whereas the participants in impairment in orientation discrimination under most circumstances (Treisman & Chen’s study were required to process due to the presence of the colour Gelade, 1980). However, when attention the cue because the cue indicated singleton in the presence or absence of is completely depleted by a demanding whether the target should be responded the memory task was taken to indicate concurrent task, performance in feature to or not on a given trial (the experiment the effect of WM load on distractor search is impaired (Joseph, Chun, & used a go/nogo paradigm), there was no inhibition. The results showed that the Nakayama, 1997). such requirement for the participants in irrelevant colour singleton impaired Our results lend support to a growing the present experiment.2 Perhaps this participants’ performance more in the body of literature in New Zealand and difference made the cue in Chen’s study dual-task condition than in the single- abroad that shows the complexity of a more effective stimulus in controlling task condition. Although attentional the effect of WM on selective attention the participants’ attentional focus. focus was not explicitly controlled in the tasks. Using stimulus displays where Further experiments are needed to reveal experiment, the fact that a small fi xation the relevant and irrelevant information the exact cause of this inconsistency in point was displayed for 2 seconds after belong to different objects, prior research results. the offset of the memory digits in the has shown that the effect of WM load dual-task condition makes it unlikely depends on the type of tasks involved in for the results to be contaminated by the WM (Han & Kim, 2004; Woodman & size of attentional focus. Luck, 2004) and the size of attentional

New Zealand Journal of Psychology Vol. 36, No. 3, November 2007 • 151 • Q. Gao, Z. Chen, P. Russell focus adopted by the participants (Chen References Eriksen, C. W., & St. James, J. D. (1986). Visual attention within and around the & Chan, 2007). American Psychological Association. (1992). field of focal attention: A zoom lens More recently, Sobel and colleagues Ethical principles of psychologists and model. Perception & Psychophysics, 40, code of conduct. American Psychologist, (Sobel, Gerrie, Poole, & Kane, 2007) 225-240. demonstrated that the effect of WM 47, 1597-1611. Garner, W. R. (1970). The stimulus in Broadbent, D. E. (1958). Perception and capacity in visual search was also information processing. American communication. London: Pergamon. modulated by the degree of top-down Psychologist, 25, 350-358. Chen, Z. (1998). Switching attention processing required in a task. In their Garner, W. R. (1974). The processing of within and between objects: The role experiment, participants performed information and structure. Hillsdale, of subjective organization. Canadian a speeded arrow discrimination task NJ: Erlbaum. Journal of , regarding a red horizontal arrow (facing 52, 7-16. Garner, W. R., & Felfoldy, G. L. (1970). left or right) among distractors of green Integrality of stimulus dimensions in Chen, Z. (2000). An object-based cost horizontal arrows and red arrows which various types of information processing. of visual filtering. Perception & were either vertical (distinct orientation , 1, 225-241. Psychophysics, 62, 482-495. condition) or orientated about 15 degrees Han, S-H, & Kim, M-S. (2004). Visual Chen, Z. (2003). Attentional focus, processing from horizontal (similar orientation search does not remain effi cient when load, and Stroop interference. Perception executive working memory is working. condition). Set size of the red distractors & Psychophysics, 65, 888-900. varied across trials while the overall set Psychological Science, 15, 623-628. Chen, Z. (2005). Selective attention and the Harms, L., & Bundesen, C. (1983). Color size of the display was held constant. perception of an attended nontarget object. segregation and selective attention. Relative to participants with high WM Journal of Experimental Psychology: Perception & Psychophysics, 33, 11-19. capacity, those with low WM capacity Human Perception and Performance, showed a steeper increase in RT with the 31, 1493-1509. Joseph, J. S., Chun, M. M., & Nakayama, K. (1997). Attentional requirements in a number of red distractors in the similar Chen, Z., & Cave, K. R. (2006a). Reinstating “preattentive” feature search task. Nature, orientation condition, but not in the object-based attention under positional 387, 805-807. distinct orientation condition. Assuming certainty: The importance of subjective that similar distractors require greater parsing. Perception & Psychophysics, Kahneman, D., & Chajczyk, D. (1983). Tests top-down processing, which in turn 68, 992-1003. of the automaticity of reading: Dilution of Stroop effects by color-irrelevant stimuli. requires more processing resources, Chen, Z., & Cave, K. R. (2006b). When Journal of Experimental Psychology: does visual attention select all features these results are consistent with the Human Perception and Performance, of a distractor? Journal of Experimental view that the manifestation of WM load 9, 497-509. effect may require relatively complete Psychology: Human Perception and Performance, 32, 1452-1464. Kane, M. J., & Engle, R. W. (2003). depletion of attentional resources. One Working-memory capacity and the Chen, Z., & Chan, C. (2007). Distractor way to test this hypothesis directly control of attention: The contributions of interference stays constant despite would be to conduct an experiment goal neglect, response competition, and variation in working memory load. task set to Stroop interference. Journal similar to that of Joseph, Chun, and Psychonomic Bulletin & Review, 14, of Experimental Psychology: General, Nakayama (1997) to see if participants 306-312. with low WM capacity would be 132, 47-70. Cohen, J. (1988). Statistical power analysis Keele, S. W. (1972). Attention demands disproportionably impaired in feature for the behavioural sciences. Hillsdale, of memory retrieval. Journal of search by a demanding concurrent NJ: Erlbaum. Experimental Psychology, 93, 245-248. attention task relative to those with a Duncan, J. (1984). Selective attention and Kramer, A. F., & Jacobson, A. (1991). high WM capacity. We are currently the organization of visual information. Perceptual organization and focused conducting experiments to investigate Journal of Experimental Psychology: attention: The role of objects and this issue. General, 113, 501-517. proximity in visual processing. Perception In conclusion, the present Egly, R., Driver, J., & Rafal, R. D. (1994). & Psychophysics, 50, 267-284. Shifting visual attention between objects experiments indicate that the LaBerge, D., Brown, V., Carter, M., Bash, D., and locations: Evidence form normal manifestation of the WM load effect & Hartley, A. (1991). Reducing the effects and parietal lesion participants. Journal on selective attention may depend on a of adjacent distractors by narrowing of Experimental Psychology: General, number of factors that include but are attention. Journal of Experimental 123, 161-177. not limited to the type of tasks involved Psychology: Human Perception and in WM, the extent of attentional focus Eriksen, B. A., & Eriksen, C. W. Performance, 17, 65-76. (1974). Effects of noise letters upon adopted by the participants, and the Lavie, N. (1995). Perceptual load as the identification of a target letter a necessary condition for selective nature of stimulus structure of the in a nonsearch task. Perception & attention. Journal of Experimental target. Psychophysics, 16, 143-149. Psychology: Human Perception and Eriksen, C. W., & Hoffman, J. E. (1973). The Performance, 21, 451-468. extent of processing of noise elements Lavie, N. (2005). Distracted and confused?: during selective encoding from visual Selective attention under load. Trends in displays. Perception & Psychophysics, Cognitive Science, 9, 75-82. 14, 155-160.

• 152 • New Zealand Journal of Psychology Vol. 36, No. 3, November 2007 Working Memory and Stroop Interference

Lavie, N., & Cox, S. (1997). On the effi ciency Woodman, G. F., Vogel, E. K., & Luck, S. of attentional selection: Effi cient visual J. (2001). Visual search remains effi cient search results in inefficient distractor when visual working memory is full. rejection. Psychological Science, 8, Psychological Science, 12, 219-224. 395-398. Yantis, S. (2000). Control of visual attention. Lavie, N., & de Fockert, J. (2005). The In S. Monsell & J. Driver (Eds.), Control role of working memory in attentional of cognitive processes: Attention and capture. Psychonomic Bulletin & Review, performance XVIII (pp. 73-104). 12, 669-674. Cambridge, MA: MIT Press. Lavie, N., Hirst, A., de Fockert, J. W., Yantis, S., & Jonides, J. (1990). Abrupt & Viding, E. (2004). Load theory of visual onsets and selective attention: selective attention and cognitive control. Voluntary versus automatic allocation. Journal of Experimental Psychology: Journal of Experimental Psychology: General, 133, 339-354. Human Perception and Performance, Logan, G.. D. (1978). Attention in character- 10, 121-134. classification tasks: Evidence for the automaticity of component stages. Journal Notes: of Experimental Psychology: General, 1. The majority of the participants were 107, 32-63. between the ages of 20 and 28. MacLeod, C. M. (1991). Half a century Approximately 67% were female. of research on the Stroop effect: An 2. Practical consideration of the already integrative review. Psychological Bulletin, high demand of processing load in the 109, 163-203. high-narrow condition prevented us from using a similar go/nogo task as that in Neisser, U. (1967). Cognitive Psychology. Chen (2003). New York, NY: Appleton. Posner, M. I., & Snyder, C. R. R. (1975). Facilitation and inhibition in the processing Address for correspondence: of signals. In P. M. A. Rabbitt & S. Dornic (Eds.), Attention and performance V (pp. Zhe Chen 669-682). New York: Academic Press. Department of Psychology University of Canterbury Remington, R. W., & Folk, C. L. (2001). A dissociation between attention and Private Bag 4800 selection. Psychological Science, 12, Christchurch, New Zealand 511-515. email: [email protected] Schneider, W., Eschman, A., & Zuccolotto, A. phone: 0064 3-3642987 ext. 7987 (2002). E-Prime user’s guide. Pittsburgh: fax: 0064 3-3642181 Psychology Software Tools, Inc. Sobel, K. V., Gerrie, M. P., Poole, B. J., & Kane, M. J. (2007). Individual differences in working memory capacity and visual search: The roles of top-down and bottom- up processing. Psychonomic Bulletin & Review, 16, 840-845. Sternberg, S. (1966). High-speed memory scanning in human memory. Science, 153, 652-654. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643-662. Treisman, A., & Gelade, G. (1980). A feature- integration theory of attention. Cognitive Psychology, 12, 97-136. Unsworth, N., Schrock, J. C., & Engle, R. W. (2004). Working memory capacity and the antisaccade task: Individual differences in voluntary saccade control. Journal of Experimental Psychology: , Memory, and Cognition, 30, 1302-1321. Woodman, G. F., & Luck, S. J. (2004). Visual search is slowed when visuospatial working memory is occupied. Psychonomic Bulletin & Review, 11, 269-274.

New Zealand Journal of Psychology Vol. 36, No. 3, November 2007 • 153 •