Microsoft word - booth & sharma, m&e.doc

STRESS AND STROOP PROPORTION-CONGRUENT EFFECTS Stress Reduces Attention to Irrelevant Information: Evidence from the Stroop Task Centre for Cognitive Neuroscience and Cognitive Systems Stroop interference can be reduced by stress, and this has been taken as evidence that stress decreases the attention paid to irrelevant information, a theory known as ‘Easterbrook’s hypothesis’. This contradicts more recent theories, which state that attentional control deteriorates in stress. Fifty-five participants undertook a Stroop task under high stress (loud white noise) and low stress conditions. Attention to the irrelevant word information was assessed by manipulating the proportion of congruent trials (e.g. the word RED in the colour red); it is known that Stroop interference increases when many such trials are presented. This effect was reduced when participants were stressed, which is evidence that stress does indeed reduce attention to irrelevant information. This pattern of results was not present in participants with low working memory spans, presumably because these participants had less attentional control. These findings highlight an important weakness in contemporary Keywords: Stroop, Selective Attention, Stress, Working Memory, Easterbrook’s Stress Reduces Attention to Irrelevant Information: Evidence from the Stroop Task Easterbrook’s hypothesis (Easterbrook, 1959) states that emotional individuals pay less attention to ‘peripheral’ (i.e., less relevant) information. Easterbrook did not himself specify the mechanism behind this effect; presumably stress overloads the attention system to some extent, and so reduces the attention resources available for allocation to less relevant information (c.f. perceptual load, Lavie, Hirst, de Fockert, & Viding, 2004). Easterbrook’s hypothesis has been supported by results indicating that stressed participants show reduced Stroop interference. In the Stroop task (Stroop, 1935), participants ignore the meaning of words and respond to the colour in which they are presented; Stroop interference is the difference in response time or accuracy between incongruent trials (e.g., the word RED in green print) and control trials (e.g. +++ in green print). The word information is irrelevant, so its influence should be reduced under stress because less attentional resources are allocated to it (Easterbrook did not specifically posit a spatial narrowing of attention, as is This prediction has been repeatedly supported. For example, O’Malley & Poplawsky (1971) used a traditional Stroop task, with multiple stimuli presented on cards. Participants who were stressed by hearing loud bursts of white noise were generally faster at this task, and showed less interference, compared to participants who performed the task in silence. O’Malley & Gallas (1977) replicated this finding with an improved control condition: participants exposed to bursts of white noise at 85dB showed reduced interference compared to those exposed to noise at 75dB. Chajut and Algom (2003, Experiment 2) also replicated this in a similar experiment using a more modern computer-controlled single-trial Stroop task, and also showed that Stroop interference could be reduced by a very difficult fake IQ test. Substances that increase physiological arousal have also been found to reduce interference (Callaway, 1959; Kenemans, Wieleman, Zeegers, & Verbaten, 1999). Easterbrook’s (1959) hypothesis has been neglected in more recent theories of cognition in emotion, which tend to claim that attentional selection deteriorates with stress or anxiety (e.g. Eysenck, Derakshan, Santos, & Calvo, 2007; Mathews & Mackintosh, 1998): Eysenck et al. predict that Stroop interference should increase in anxious states (see Pallak, Pittman, Heller, & Munson, 1975). Therefore, the fact that interference is usually found to decrease seems to present a problem for recent theories of cognition in anxiety. If this problem is to be accepted, it must be shown that stress’s effects on Stroop interference do indeed result from improved selectivity, and not from some strategy change made by the participants. For example, it is possible that stress simply increases the participants’ motivation levels (c.f. McFall, Jamieson, & Harkins, 2009), so that they wish to complete the task (and escape the stress manipulation) as quickly as possible. This will lead to an overall decrease in RTs. Responding more quickly could mean a smaller Stroop effect because less time has elapsed for interference to build up to affect performance. It is therefore important to discount a strategic explanation of stress’s effects on Stroop interference. One way to address whether stress-related decreases in Stroop interference are strategic is to simultaneously assess the attention paid to the irrelevant word information with another method. For example, if many congruent trials (e.g., BLUE in blue print) appear in the Stroop task, interference increases (Kane & Engle, 2003; Lindsay & Jacoby, 1994). Although the cause of this ‘proportion-congruent’ effect is still controversial (Blais, Robidoux, Risko, & Besner, 2007), it is clear that such a manipulation can only affect participants’ behaviour if they attend to the irrelevant word information to some degree. If stressed participants attend less to the irrelevant word information, they should be less likely to notice the congruent trials and their behaviour should not change. In other words, the Stroop interference shown by stressed participants should be unaffected by the proportion of congruent trials This pattern would not be expected if stress’s reduction of Stroop interference was due to some strategic factor. Presenting congruent trials increases the salience of the distracter words, because a correlation is introduced between the colour information and the word information (see Melara & Algom, 2003). This makes the irrelevant word information harder for the participant to ignore, Melara and Algom argue, because the attention system is predisposed to seek out such correlations. If stress’s reduction in Stroop interference is caused by some strategic change – i.e., if the irrelevant word information is still processed in stress – then the correlation between word information and colour information will still decrease the participants’ ability to inhibit the distracter words, and will still increase interference. In other words, the proportion-congruent effect in stress and non-stress conditions should be equal. On the other hand, if stress does indeed decrease attention to the irrelevant word information as Easterbrook’s (1959) hypothesis implies, then this correlation is more likely to go unnoticed, and the proportion-congruent effect should be small in To increase the power of the experiment, participants’ working memory (WM) spans were assessed. Although WM span ostensibly relates to the storage capacity of memory, it is also related to the control of attention, and the maintenance of task goals (De Fockert, Rees, Frith, & Lavie, 2001; Kane & Engle, 2003; Lavie et al., 2004). This attention control aspect of working memory has been present since its inception (see Baddeley, 1992), and some theorists have suggested that individual differences in working memory span may be largely or entirely attributable to attention control, rather than storage capacity per se (Kane, Conway, Hambrick, & Engle, 2007). It therefore seems reasonable that WM span and stress, if they both relate to attention control and/or selectivity, will interact in their effects on Stroop interference. Specifically, at any given moment, participants with higher WM spans (‘high spans’) are more likely to be treating as irrelevant information which actually is irrelevant to their task, and those with lower WM spans (‘low spans’) are more likely to be suffering a ‘slip’ of attention and mistakenly treating target colour information as irrelevant. Easterbrook’s hypothesis suggests that attention resources are withdrawn from irrelevant stimuli, but if the aspect of a Stroop stimulus that is treated as irrelevant is less consistent for low spans, then they would be expected to show less consistent stress effects. High spans on the other hand are more likely to maintain attentional control to selectively attend to the target colour so that stress decreases Stroop interference. For this reason, high and low spans were identified after the experiment so that their performance could be compared. To summarise, the present experiment investigated whether Stroop interference was reduced in stress, particularly in groups with high WM spans. A constant white noise at 84dBC was used to induce stress; this was found to be effective by Chajut and Algom (2003), and a similar manipulation was used by O’Malley and colleagues (O'Malley & Gallas, 1977; O'Malley & Poplawsky, 1971). The proportion of congruent trials was also manipulated. If stress reduces attention to irrelevant information as described by Easterbrook (1959), stressed participants should not notice this manipulation, and their behaviour should not change in response to it. WM span was assessed, as it was predicted that the above pattern of results should be more clear in high spans. Sixty-three participants (50 females), aged 17 to 59 years (M = 22) took part in the experiment. All were native English speakers with normal or corrected hearing and vision. They received £5 or course credit for their participation. The Stroop task employed a 2x3x2 within-participants design. The factors were Trial Type (incongruent or control), Proportion-Congruent (proportion of congruent trials within the block: 0%, 25% or 50%), and the Volume of the noise (stressful, 84dBC, or unstressful, 65dBC). The experiment was run using a Dell Inspiron 510M computer running E- prime v1.0.1, a CRT monitor, and a PST response box. A chin rest was positioned one metre from the screen. Noise was created using Philips SBC HC202 headphones with the transmitter switched off (this produces a broad-range hiss). Volumes were set As a manipulation check for stress, skin conductance data were collected using a BIOPAC MP35. This recorded at 10Hz via 10mm-diameter Ag-AgCl disposable electrodes treated with isotonic gel, attached to the distal phalanges of the third and Stimuli subtended 0.9° vertically and averaged 3.4° horizontally, and were presented in red, blue, yellow or green print on a black background. Incongruent stimuli consisted of a colour word presented in a contrasting colour, e.g. BLUE in green print. Congruent stimuli consisted of a word in a matching colour, e.g. BLUE in blue print. Control stimuli consisted of strings of three to six +’s in one of the four colours above. Neutral stimuli consisted of a neutral word presented in one of the four colours. Words were randomly selected for each participant from lists of mono- and bi-syllabic words of up to eight letters, which were matched for letter length, syllable length and CELEX frequency (measured using Lexicus, Frankish, 2002). These later appeared as targets in a recognition test: no significant effects were recorded, so Participants were seated in front of the screen, and the procedure was explained. They were reassured about the safety of the noise. The skin conductance electrodes were applied and tested, and recording was initiated. During a practice phase, participants familiarised themselves with the task. Participants learnt the colour to which each button corresponded, and were advised that speeded responding was unnecessary. The experimenter left the room during all phases of the Stroop task. The practice phase consisted of the words ‘HOUSE’, ‘TREE’ and ‘CAT’, randomly presented 25 times in each of the four print colours. Each trial began with a 1000 ms blank black screen before the stimulus, which remained on-screen until a response was made. The practice phase also served as a habituation period for the skin At the end of the practice phase the experimenter reiterated the nature of the task, stressing that the participant should now respond as quickly and accurately as possible, and explaining that three blocks of trials would be presented. The experimenter then set the headphones to the required volume. The order of the two volume conditions was counterbalanced across participants. The first experimental phase consisted of three blocks, one for each Proportion-Congruent condition. Each block included 48 critical trials (24 control and 24 incongruent). The 0% congruent block also included 48 neutral word trials; the 25% congruent block included 24 neutral word trials and 24 congruent trials; and the 50% congruent block included 48 congruent trials. Note that the neutral trials were present to replace the congruent trials without altering the relative frequency of incongruent and control trials (see Melara & Algom, 2003); responses to these trials, and indeed to congruent trials, were not analysed. The blocks were presented in a random order, and trials were presented in a random order within each block. Trials The experimenter then told the participant that they would complete the experimental phase a second time, but at the other volume. The second experimental phase consisted of the same blocks as in the first phase, in a newly-randomised order. After the Stroop task participants completed the Ospan WM test (Tuholski & Engle, 2002), as described by Turner and Engle (1989). The Ospan involves remembering lists of words while solving arithmetic problems. The task was controlled by the experimenter, and participants responded on pre-printed sheets. Having completed the Ospan, participants were given a full debriefing. Overall accuracy was very high (M = 0.95). Eight participants made more than 10% errors; these participants were considered as outliers, and removed from further analyses. Including these cases does not change the overall pattern of results. The Ospan is scored by summing the length of all completely correct lists. Scores ranged between two and 32 (M = 14.96, SD = 7.78), and fit the commonly- found distribution for the task (Kane & Engle, 2003). The sample was median split to produce a high span group with scores of 15 and above (N = 26, M = 21.85, SD = 4.66) and a low span group with scores of 14 and below (N = 29, M = 8.79, SD = The skin conductance data were examined to gauge the effectiveness of the stress manipulation. Data from two participants were lost due to experimenter error. For the 53 remaining participants, data were averaged for each of the six blocks of Stroop trials: these correspond to the chronological order of the blocks. This was done to test whether participants’ arousal levels varied with time. These data were subjected to a 2 (Volume) x 3 (Block) x 2 (WM Group) ANOVA. There was a significant interaction between Volume and Block (F(2,102) = 7.41, MSE = 0.37, p < 0.001, η 2 p = 0.13), such that the arousal difference between Volume conditions decreased with time (First block, 84dBC: M = 7.39µS, SD = 4.87µS; 65dBC: M = 7.02µS, SD = 4.79µS. Second block, 84dBC: M = 7.12µS, SD = 4.94µS; 65dBC: M = 6.75µS, SD = 5.09µS. Third block, 84dBC: M = 6.94µS, SD = 5.01µS; 65dBC: M = 7.12µS, SD = 5.13µS). This shows that the stress manipulation was effective, although its effectiveness disappeared in the final block. This does not affect the Stroop results as these effects are distributed equally across the three Correct response times were subjected to a participant- and condition-specific outlier trimming procedure, using a non-recursive moving criterion (Van Selst & Jolicoeur, 1994). Two interference scores were then calculated, one for response time (incongruent – control, see Figure 1) and one for accuracy rate (control – incongruent, ). Note that interference scores could not be calculated with congruent or neutral trials, as these were not present in every Proportion-Congruent condition: there were no congruent trials in 0% congruent condition, and no neutral trials in 50% congruent condition). RT interference was subjected to a 2 (Volume) x 3 (Proportion-Congruent) x 2 (WM Group) ANOVA. There was a significant three-way interaction (F(2,106) = 4.17, MSE = 5150, p < 0.05, η 2 p = 0.07). Specifically, among high spans the Proportion-Congruent effect was only significant at 65dBC (F(2,50) = 6.01, MSE = p = 0.19) and not at 84dBC (F(2,50) = 2.00, MSE = 3111, p = 0.15, p = 0.07); among low spans, the Proportion-Congruent effect was significant at 84dBC (F(2,56) = 9.53, MSE = 5563, p < 0.001, η 2 (F(2,56) = 0.85, MSE = 6441, p = 0.85, η 2 There was also a Proportion-Congruent main effect (F(2,106) = 12.18, MSE = p = 0.19) as expected, but no other effects reached significance (all Fs < 1.5, all ps > 0.2). To test against a possible motivational account of stress’s effects on Stroop interference, the RT analysis was repeated with mean correct RTs to control trials (i.e., not interference scores) as the DV. There was no effect of Volume (F(1,53) = 1.83, MSE = 6217, p = 0.18, η 2 p = 0.03, see Table 1), and there were no interactions between Volume, WM Group, and Proportion-Congruent (all Fs < 0.8, ps > 0.4). Accuracy interference scores were subjected to the same analyses. There was a tendency for Volume to interact with WM Group (F(1,53) = 3.60, MSE = 0.003, p = p = 0.06), so that the louder noise tended to reduce interference for high spans (84dBC: M = 0.005, SD = 0.032; 65dBC: M = 0.020, SD = 0.040) and increase interference for low spans (84dBC: M = 0.011, SD = 0.044; 65dBC: M = 0.002, SD = 0.041). However, these simple effects did not approach significance (both Fs < 2.6, both ps > 0.12). There was also a Proportion-Congruent main effect (F(2,106) = 5.24, MSE = 0.004, p < 0.01, η 2 p = 0.09), so that interference was highest in the 50% congruent condition (M = 0.025), and much lower in the 25% (M = 0.005) and 0% (M = -0.002) conditions. No other effects reached significance (all Fs < 0.6, all ps > 0.5). As predicted, stress effects on Stroop interference were jointly moderated by Proportion-Congruent and WM Group. This interaction (Figure 1) supports the prediction made above, that proportion-congruent effects will be less apparent in stressed high spans. This is evidence that these participants are indeed less aware of the irrelevant word information, and therefore seems to support Easterbrook’s (1959) hypothesis. This reduced proportion-congruent effect also meant that high spans tended to show less Stroop interference under stress, although this lower-order interaction did not reach significance in its own right, perhaps because it is subsumed within the significant three-way interaction. The present work suggests the intriguing possibility that the detectability of stress’s reduction of Stroop interference depends upon the composition of the task itself. For example, high spans did not show reduced interference under stress in the 0% congruent condition, perhaps because they were already strongly attending to the target colour (Kane & Engle, 2003). Future work will need to examine this possibility more closely. Note that these effects cannot be easily attributed to an increase in participants’ motivation. Participants did not show a significant decrease in RTs to control trials in the 84dBC condition, which would be expected if the loud noise motivated them to respond quickly. Furthermore, if the noise motivated them to respond as quickly as possible, they should have been tempted to pay more attention to the irrelevant word information in the 50% congruent condition, as this would have enabled them to respond quickly and efficiently to the large number of congruent trials presented in these blocks. However, the 84 dBC noise reduced high spans’ interference in this condition, suggesting that they were paying less attention to the word information. Clearly, this is more consistent with Easterbrook’s hypothesis than it is with a strategic or motivational account. This novel finding seems to contradict many recent theories on how stress and anxiety affect cognition. In the future, it will be important to further study this problem by manipulating the salience of the irrelevant word information using other methods (e.g., the size of the irrelevant word: see e.g. Melara & Algom, 2003). Oddly, high spans appear to show more interference than low spans in the 50% condition, when not under stress2. This would contradict the notion that WM correlates with goal maintenance. Previous work has found that WM span differences in Stroop interference only appear when there are many congruent trials, though high spans should show less interference than low spans (e.g. Kane & Engle, 2003). However, Kane and Engle warned their participants about the congruent trials, and instructed them to remain focussed on the colour information; in the present experiment, such a warning was not possible. Non-stressed high spans may therefore have switched to an easier word-reading strategy in the 50% congruent condition; this would have impeded their responses to incongruent trials, leading to increased interference (see Lindsay & Jacoby, 1994). One weakness of the present experiment is that it did not directly assess participants’ mood with, for example, an anxiety questionnaire. We feel that this is an acceptable omission given our physiological validation of our stress manipulation, together with the fact that loud noise manipulations have a long history in stress research. Noise has also been previously shown to increase arousal (Geen & To summarise, stress reduced proportion-congruent effects on Stroop interference, among participants with high working memory span. This provides further evidence that stress reduces the attention paid to irrelevant information (Easterbrook, 1959), and suggests stress’s reduction of Stroop interference is not likely to be strategic in nature. These effects remain a problem for theories stating that anxious states decrease attentional control (e.g. Eysenck et al., 2007; Mathews & Mackintosh, 1998). Eysenck et al. argue that Easterbrook-like effects in other paradigms (e.g. Weltman, Smith, & Ergstrom, 1971) can be explained in terms of decreases in attentional control: stressed individuals’ attention is drawn to the most salient stimulus, but in these experiments the most salient stimulus is actually the target, not the distracter. The present experiment suggests that such accounts must be expanded to cover the Stroop task, where the distracter word is usually regarded as much more salient than the target (Melara & Algom, 2003). Future work should attempt to reconcile these two apparently contradictory theoretical positions. Baddeley, A. D. (1992). Working memory. Science, 255, 556-559. Blais, C., Robidoux, S., Risko, E. F., & Besner, D. (2007). Item-specific adaptation and the conflict-monitoring hypothesis: a computational model. Psychological Callaway, E. (1959). The influence of amobarbital (amylobarbitone) and methamphetamine on the focus of attention. Journal of Mental Science, 105, Chajut, E., & Algom, D. (2003). Selective attention improves under stress: implications for theories of social cognition. Journal of Personality and Social De Fockert, J. W., Rees, G., Frith, C. D., & Lavie, N. (2001). The role of working memory in visual selective attention. Science, 291, 1803-1806. Easterbrook, J. A. (1959). The effect of emotion on cue utilization and the organization of behavior. Psychological Review, 66, 183-201. Eysenck, M. W., Derakshan, N., Santos, R., & Calvo, M. G. (2007). Anxiety and cognitive performance: attentional control theory. Emotion, 7, 336-353. Frankish, C. (2002). Lexicus (Version 3.0.1): University of Bristol. Geen, R. G., & McCown, E. J. (1984). Effects of noise and attack on aggression and physiological arousal. Motivation and Emotion, 8, 231-241. Huguet, P., Dumas, F., & Monteil, J. M. (2004). Competing for a desired reward in the Stroop task: when attentional control is unconscious but effective versus conscious but ineffective. Canadian Journal of Experimental Psychology, 58, Kane, M. J., Conway, A. R. A., Hambrick, D. Z., & Engle, R. W. (2007). Variation in working memory capacity as variation in executive attention and control. In A. R. A. Conway, C. Jarrold, M. J. Kane, A. Miyake & J. N. Towse (Eds.), Variation in working memory (pp. 21-48). New York: Oxford University Kane, M. J., & Engle, R. W. (2003). Working-memory capacity and the control of attention: the contributions of goal neglect, response competition, and task set to Stroop interference. Journal of Experimental Psychology: General, 132, 47- Kenemans, J. L., Wieleman, J. S. T., Zeegers, M., & Verbaten, M. N. (1999). Caffeine and Stroop interference. Pharmacology Biochemistry and Behaviour, 63, 589- Lavie, N., Hirst, A., de Fockert, J. W., & Viding, E. (2004). Load theory of selective attention and cognitive control. Journal of Experimental Psychology: General, Lindsay, D. S., & Jacoby, L. L. (1994). Stroop process dissociations: the relationship between facilitation and interference. Journal of Experimental Psychology: Human Perception and Performance, 20, 219-234. Mathews, A., & Mackintosh, B. (1998). A cognitive model of selective processing in anxiety. Cognitive Therapy and Research, 22, 539-560. McFall, S. R., Jamieson, J. P., & Harkins, S. G. (2009). Testing the mere effort account of the evaluation-performance relationship. Journal of Personality and Social Psychology, 96, 135-154. Melara, R. D., & Algom, D. (2003). Driven by information: a tectonic theory of Stroop effects. Psychological Review, 110, 422-471. O'Malley, J. J., & Gallas, J. (1977). Noise and attention span. Perceptual and Motor O'Malley, J. J., & Poplawsky, A. (1971). Noise-induced arousal and breadth of attention. Perceptual and Motor Skills, 33, 887-890. Pallak, M. S., Pittman, T. S., Heller, J. F., & Munson, P. (1975). The effect of arousal on Stroop color-word task performance. Bulletin of the Psychonomic Society, Sharma, D., Booth, R., Brown, R., & Huguet, P. (2009). Social Facilitation in the Stroop Task is Caused by Distracter Inhibition, Manuscript submitted for Stroop, J. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 653-662. Tuholski, S., & Engle, R. W. (2002). OSPAN E-prime version. Retrieved 17th March, 2005, from http://psychology.gatech.edu/renglelab Turner, M. L., & Engle, R. W. (1989). Is working memory capacity task dependent? Journal of Memory and Language, 28, 127-154. Van Selst, M., & Jolicoeur, P. (1994). A solution to the effect of sample size on outlier elimination. Quarterly Journal of Experimental Psychology, 47A, 631- Weltman, G., Smith, J. E., & Ergstrom, G. H. (1971). Perceptual narrowing during simulated pressure-chamber exposure. Human Factors, 13, 99-107. The Stroop data were re-analysed with the upper and lower tertiles as the WM groups. The same pattern of results were found as reported in Figure 1. Furthermore, when Proportion-Congruent effects were calculated by subtracting 0% condition interference scores from 50% condition interference scores, it was found that Proportion-Congruent effects in the 84dBC condition tended to correlate negatively with Ospan score (r(53) = -0.224, p = 0.099); Proportion-Congruent effects in the 65dBC condition did not correlate with Ospan scores (r(53) = 0.123, p = 0.26). These findings support our conclusions from our median split analyses. Although this tendency did not approach significance for RT interference, t(52) = 0.70, p = 0.49, or accuracy interference, t(52) = 1.30, p = 0.20. This research was carried out while Rob Booth was a doctoral student at the University of Kent, supervised by Dinkar Sharma. Table 1: Mean correct response times (RTs) and accuracy rates (Acc.) from the Stroop task, by Working Memory Span Group (WM) and Proportion-Congruent (P-C). Figure 1: Stroop response time interference scores (incongruent RT – control RT) for high and low working memory span groups, under stressful (84dBC) and non-

Source: http://www2.isikun.edu.tr/personel/rob.booth/Research/Booth%20&%20Sharma%202009%20motivation%20emotion.pdf