Acta Psychologica 151 (2014) 214–221

Contents lists available at ScienceDirect

Acta Psychologica journal homepage: www.elsevier.com/ locate/actpsy

Stress improves selective attention towards emotionally neutral left ear stimuli Robert Hoskin ⁎, M.D. Hunter, P.W.R. Woodruff Sheffield Cognition and Neuroimaging Lab (SCANLAB), Academic Clinical Psychiatry, Department of Neuroscience, Faculty of Medicine, Dentistry & Health, University of Sheffield, Longley Centre, Norwood Grange Drive, Sheffield S5 7JT, UK

a r t i c l e

i n f o

Article history: Received 1 February 2014 Received in revised form 25 May 2014 Accepted 18 June 2014 Available online 31 July 2014 PsycINFO classification: 2320 2326 2360 Keywords: Stress Anxiety Auditory oddball Gender Selective attention Auditory perception

a b s t r a c t Research concerning the impact of psychological stress on visual selective attention has produced mixed results. The current paper describes two experiments which utilise a novel auditory oddball paradigm to test the impact of psychological stress on auditory selective attention. Participants had to report the location of emotionallyneutral auditory stimuli, while ignoring task-irrelevant changes in their content. The results of the first experiment, in which speech stimuli were presented, suggested that stress improves the ability to selectively attend to left, but not right ear stimuli. When this experiment was repeated using tonal stimuli the same result was evident, but only for female participants. Females were also found to experience greater levels of distraction in general across the two experiments. These findings support the goal-shielding theory which suggests that stress improves selective attention by reducing the attentional resources available to process task-irrelevant information. The study also demonstrates, for the first time, that this goal-shielding effect extends to auditory perception. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Psychological stress can be defined as the response of an organism to threats (stressors) in the environment (Lazarus, 1993). An open question within behavioural science relates to how psychological stress alters the ability to selectively attend to task-relevant aspects of the sensory scene. While periods of stress may be expected to alter the response to threatening stimuli, it is not clear whether (and how) stress affects selective attention when the information being perceived is not associated with any emotional significance. It has been proposed that stress reduces the level of attentional resources available for perception, and that in response to this change, the processing of distracting, task-irrelevant information is sacrificed in order to preserve goal-relevant processing (Chajut & Algom, 2003). This ‘goal-shielding’ effect (Plessow, Fischer, Kirschbaum, & Goschke, 2011) produces an improvement in selective attention under stress because relatively less processing is available to be dedicated to task-irrelevant information (Chajut & Algom, 2003). The majority of evidence in support of the goal-shielding theory comes from visual tasks, such as those assessing the impact of stress on the Stroop effect (Booth & Sharma, 2009; ⁎ Corresponding author. Tel.: +44 7805 238 919; fax: +44 114 222 6250. E-mail address: [email protected] (R. Hoskin).

http://dx.doi.org/10.1016/j.actpsy.2014.06.010 0001-6918/© 2014 Elsevier B.V. All rights reserved.

Chajut & Algom, 2003; Hu, Bauer, Padmala, & Pessoa, 2012), Simon task (Plessow et al., 2011) or flanker task (Caparos & Linnell, 2012). An alternative view of how stress might affect attention is offered by attentional control theory (ACT: Eysenck, Derakshan, Santos, & Calvo, 2007). ACT proposes that aversive emotional states serve to reduce the attentional control required to inhibit distracting information. In contrast to goal-shielding theory, one would predict from ACT that the appearance of a stressor would increase the susceptibility to distraction during selective attention tasks. Existing evidence tends to support the predictions of ACT as regards the processing of threatening distracters (Eysenck et al., 2007). However there are only a small number of (visual) selective attention studies which demonstrate an increase in distraction by emotionally neutral stimuli when stressors are present (Moser, Becker, & Moran, 2012) in contrast to those which show the opposing, goal-shielding effect (e.g. Booth & Sharma, 2009). Although there is a growing literature on the effect of stress on visual selective attention, there are very few studies concerning the impact of stress on auditory selective attention. A recent study found electrophysiological evidence which suggested that selective attention towards emotionally neutral auditory stimuli is disrupted during stress, in concert with the predictions of ACT (Elling et al., 2011). However as no behavioural metrics of task performance were recorded during this study it is not certain whether the electrophysiological changes found actually

R. Hoskin et al. / Acta Psychologica 151 (2014) 214–221

relate to alterations in attention. The current study utilises a novel passive oddball paradigm to test the impact of psychological stress on selection attention towards emotionally neutral sounds. In the first experiment participants were required to report the side to which a monaural speech stimulus was presented. The content of the speech stimuli was arranged in an oddball pattern, with same word (the ‘standard’) being presented during the majority of trials, while a different word (the ‘deviant’) was presented only on occasional trials. As the semantic content of the speech stimuli was irrelevant to the task, the unexpected appearance of the deviant speech content serves to distract attention away from the task. Thus reaction times to deviant trials were expected to be longer than reaction times to standard trials, a characteristic known as the ‘oddball effect.’ The size of the oddball effect during passive oddball paradigms can be taken as a behavioural metric of distraction, as its size reflects the ability of the distracting information within the deviant stimuli to interfere with task processing. Since in the aforementioned paradigm the distracting information is contained within the task stimulus itself, the task requires selective auditory attention to resist this distraction. Psychological stress was manipulated during the task by interspersing either aversive or neutral auditory-visual stimuli between trials. Goal-shielding theory predicts that the oddball effect will reduce during the stress condition (improved selective attention), whereas ACT predicts that the oddball effect will increase during the stress condition (thus showing a reduced ability to inhibit distracting information). ACT proposes that aversive emotional states cause a loss of attentional control regardless of whether the aversive emotional state is provoked by the presence of a stressor (i.e. psychological stress) or by high trait anxiety (Eysenck et al., 2007, p. 336). Trait anxiety refers to an individual's general predisposition to perceive threats in the environment. Trait anxiety is therefore distinct from the psychological stress engendered by a particular stressor, although the two concepts are, of course, related. The majority of studies showing support for the predictions of ACT have utilised between-participant differences in trait anxiety, rather than variations in psychological stress (Eysenck et al., 2007). A self-report measure of trait anxiety was therefore included in the study to allow the impact of both psychological stress and trait anxiety on task performance to be assessed separately. In line with ACT it was predicted that individuals with high trait anxiety would show greater distraction (i.e. a larger oddball effect) than less anxious individuals. Finally as trait anxiety has been found to predict the effect of manipulations of psychological stress (e.g. Hoskin, Hunter, & Woodruff, 2014) trait anxiety was also regressed against the impact of the stress manipulation on the size of the oddball effect. It was predicted that any effect of the stress manipulation would be greater in individuals reporting high trait anxiety.

2. Method: Experiment 1 2.1. Participants Fifty-three participants were recruited for the experiment. Two participants withdrew due to discomfort with the stress manipulation and a further two participants were excluded due to poor performance and equipment malfunction respectively. This left 49 participants (27 female, mean age 27, σ = 7.79) whose data was analysed. Participants reporting

215

either a current psychiatric diagnosis or hearing difficulties were excluded from the study. All participants had normal or corrected-to-normal vision. The study received ethical approval from the University of Sheffield Medical School Research Ethics Committee.

2.2. Task design Participants were required to respond to a monaural speech stimulus by pressing the arrow button on a laptop that corresponded to the ear in which the sound had been played (Fig. 1). Each speech stimulus lasted 250 ms and participants had 800 ms from stimulus onset to make a response. The inter-trial interval was 450 ms, giving an overall trial length of 1250 ms. Responses made within 150 ms of stimulus onset were treated as errors. The speech stimuli comprised of 4 common one-syllable words, spoken in a neutral male voice (Supplementary Table 1). These words were presented in an oddball pattern, such that the majority of trials involved the same word (the standard) being presented, while the 3 remaining speech sounds acted as deviant stimuli by appearing only on occasional trials (Fig. 2). The arrangement of trials within the experiment was similar to that described in Hoskin, Hunter, and Woodruff (in press). Trials were presented in ‘blocks,’ with each of the four speech stimuli acting as the standard stimulus in 2 blocks. There were therefore 8 blocks arranged in this manner. The balanced use of each stimulus in both the standard and deviant positions ensured that any differences between the speech sounds did not systematically contribute to the oddball effect. Each block included 76 oddball trials alongside 6 presentations of audio-visual stimuli. Each audio-visual stimulus was presented for 2500 ms. Every block began with the presentation of one audio-visual stimulus with the remaining 5 stimuli arranged in a pseudo-random pattern within the block such that between 8 and 20 oddball trials separated each presentation of an audio-visual stimulus (Fig. 2). In half the blocks (herein referred to as ‘stress blocks’) the audio-visual stimulus involved the presentation of an aversive image alongside speech relevant to the content of the image. In the remaining (non-stress) blocks neutral images were presented alongside content-relevant speech. Participants were made aware of the valence of the images that would appear in the upcoming block, with the purpose of evoking psychological stress as regards the threat of the appearance of unpleasant content during the stress blocks only. Block presentation was arranged so that each pair of blocks that employed the same standard stimulus was presented in succession. Each of these pairs contained one stress block and one non-stress block. The arrangement of blocks in this manner therefore minimised the number of times the identity of the standard stimulus changed during the paradigm, while allowing stress and non-stress blocks to alternate. Across participants the position of the 4 pairs of blocks were randomised using a 4x4 Latin square formation to ensure that each pair of blocks using a particular standard stimuli appeared (across participants) in each position (first to fourth) an equal number of times. Half the participants started with a stress block, and the other half started with a non-stress block. Twelve deviant trials were presented within each block (probability of occurrence 16%) with each of the three different deviant stimuli being presented 4 times. Trial position was pseudo-randomised such that (1) each block started with at least 5 consecutive standard trials, (2) at least

Fig. 1. The speech oddball task. Participants had 800 ms to respond as to which side of the head a speech sound was presented.

216

R. Hoskin et al. / Acta Psychologica 151 (2014) 214–221

Fig. 2. Block design of the speech oddball task. Each block began with the presentation of a 2500 ms audio-visual stimulus. Similar audio-visual stimuli were presented throughout the block, sporadically interrupting the train of oddball trials. The oddball trials were arranged so that the majority involved the presentation of a particular speech sound (standard trial) but occasional (deviant) trials involved the presentation of a different sound. Each speech sound lasted for 250 ms, and each trial lasted for 1250 ms.

2 standard trials followed each audio-visual stimulus and each deviant trial, and (3) that the same deviant stimulus was not presented on successive deviant trials. In addition, the type of response required for each trial in comparison to the previous trial (e.g. either maintaining or switching from the previous response) was pseudo-randomised such that an equal number of ‘switch’ and ‘stay’ responses were required, both within each block, and within deviant and standard trials. The side of stimulation (left or right ear) was pseudo-randomised so that an equal number of each stimulus was presented within each block, and also within deviant and standard trials. In order to force participants to attend to the audio-visual stimuli, each block concluded with the presentation of three recognition tests. During each recognition test, two images were presented side-by-side and the participant was required to indicate which of the images had appeared (as part of an audio-visual stimulus) in the preceding block. Following the recognition tests, participants were asked to provide a rating on a seven point Likert scale as to how stressful they found it to complete the preceding block (1 = not at all, 7 = very stressful).

2.3. Stimulus and apparatus The images used for the audio-visual stimuli were selected largely from the International Affective Picture System (IAPS) using their normative arousal and valence ratings (Lang, Bradley, & Cuthbert, 2008), although 11 additional aversive images were sourced from the internet. Spoken sentences, describing in general terms the image content, were generated for each image. There was no difference in duration (max 2500 ms), number of syllables or number of words between the sentences used in each condition. A list of the aversive (valence rating b 3.15, arousal rating N 4.3) and emotionally-neutral (valence 4.5–5.6, arousal b 3.5) images and the associated sentences used during the study are contained in Supplementary Table 2. Each audio-visual stimulus was presented only once during the paradigm. All auditory stimuli were recorded in a neutral male voice using Audacity (http://audacity.sourceforge.net/) and were normalised to the same average RMS amplitude using a custom MATLAB script (http:// www.mathworks.com/products/matlab/). The experimental paradigm was written using the Presentation software (http://www.neurobs. com/) and presented via a laptop. Auditory stimuli were delivered at 70 dB through Sennheiser HD 265 headphones.

2.5. Skin conductance response Skin conductance response (SCR) data, sampled at 20 Hz, was collected from participants whilst they undertook the task. SCR metrics were calculated for the entire duration of each block (excluding the periods when the recognition tests and rating scale were being completed). This allowed arousal levels throughout the entire period in which the oddball task was being performed to be assessed. Two indices of physiological arousal were garnered from the SCR data; an area measure of the phasic SCR response, and a count of the number of individual skin conductance responses with an amplitude over 0.05μS. Further details of the SCR analysis are contained in the supplementary material. 2.6. Data analysis A 3-way repeated-measures ANOVA was performed using stress (stress, non-stress) trial_type (standard, deviant) and side of presentation (left, right) as the factors. Within this ANOVA the oddball effect (OE) is represented by the main effect of trial_type. Gender was included as a between-subject factor in light of previous research which has found gender differences in the influence of emotion on auditory attention (Garcia-Garcia, Dominguez-Borras, SanMiguel, & Escera, 2008). Standard trials that immediately followed the presentation of an audio-visual stimulus were excluded from the analysis because the interruption of the task instigated by the audio-visual stimuli disrupts the expectation as to the nature of the standard stimulus, thus also disrupting the basis for the OE. Standard trials that occurred immediately after deviant trials were also excluded because these trials involve the processing of a perceptual change (vs. the previous trial). This characteristic makes them different from other standard trials, and therefore dilutes the OE (Parmentier & Andres, 2010; Parmentier, Elsley, Andres, & Barcelo, 2011). The non-parametric Serlin–Harwell Aligned Rank Procedure (Serlin & Harwell, 2004) was used to assess whether trait anxiety predicted the size of the OE, or the impact of the stress manipulation upon it. Metrics for the OE (deviant trail reaction time – standard trial reaction time) and the impact of stress on the oddball effect (OE during stress blocks – OE during non-stress blocks) were calculated for each participant. Trait anxiety was then regressed against these values with the impact of age, gender and stress order (whether the participant received a stress or non-stress block first) controlled for. 3. Results: Experiment 1

2.4. Procedure 3.1. Stress manipulation After signing a consent form participants completed the Edinburgh Handedness Scale (Oldfield, 1971) and the Spielberger State-Trait Anxiety Inventory (STAI: Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983). Participants first performed a separate task, which is discussed elsewhere (Hoskin et al., in press). They then completed a practice block of the speech oddball task, before progressing to undertaking the 8 blocks of the task itself.

Wilcoxon signed-rank tests performed on the stress ratings collected after each block revealed that the stress blocks were rated as more stressful to complete than the non-stress blocks (z = 5.07, p b .0001, r = .51). Trait anxiety correlated with these stress ratings, with more anxious participants reporting higher ratings of stress (r = .38, p b .01). However the skin conductance response measures (successfully collected from 44

R. Hoskin et al. / Acta Psychologica 151 (2014) 214–221

participants) did not differ significantly between the stress and nonstress blocks (area: z = − .57, p N .1, count: z = − .53, p N .1). Error rates from the recognition tests were less than 2%, suggesting that participants were attending to the audio-visual stimuli on which the stress manipulation was reliant.

3.2. Task performance Task performance (% correct, averaged across participants) was 93.7%. The 3-way ANOVA performed on the reaction time data revealed a 3-way interaction between stress, trial_type and side of presentation (F(1,47) = 4.79, p b .05). This 3-way interaction was underpinned by three 2-way interactions. A stress*trial_type interaction (F(1,47) = 4.73, p b .05) was present, indicating a smaller oddball effect in the stress condition (49 ms vs 55 ms). A trial_type*side interaction was also present (F(1,47) = 13.9, p b .01, r = .48) with the oddball effect being larger for left-ear that right ear stimuli (59 ms vs 46 ms). Finally, there was also a trend towards a trial_type*gender interaction (F(1,47) = 3.77, p = .06, r = .27) with the oddball effect being larger for females (60 ms vs 45 ms). As the primary focus of the research concerned the interaction between stress and trial_type the 3-way interaction was analysed by running 2x2 ANOVAs for each side of presentation individually, thus allowing an identification as to how the interaction of stress and trial_type differed for stimuli in each ear (Fig. 3).

217

The ANOVA for the left ear revealed a main effect of trial_type (F(1,47) = 162.6, p b .001, r = .88) representing the OE, with trials involving the deviant stimulus evoking a longer reaction time than those involving the standard stimulus (476 ms vs 417 ms). There was also a stress*trial_type interaction (F(1,47) = 8.77, p = .005, r = .40) with the oddball effect being smaller in the stress condition (53 ms vs 65 ms). The equivalent analysis for the right ear also revealed a main effect of type (F(1,47) = 136.7, p b .001, r = .86). However there was no significant stress*trial_type interaction (F(1,47) = 0.01, p N .1) suggesting that the 3-way stress*trial_type*side interaction was due to the oddball effect being reduced under stress only for stimuli presented to the left ear. A gender*trial_type interaction was also present, with the oddball effect being larger in women (F(1,47) = 4.12, p b .05, r = .28). Finally there was a trend towards a main effect of stress (F(1,47) = 3.76, p b .1, r = .27) with reaction time tending to be longer in the stress condition. As the reaction times to the standard trials were found to differ significantly from the normal distribution (Shapiro–Wilk test), non-parametric statistics were used to assess whether the positive findings from the ANOVAs might be sensitive to the violations of the assumption of normality. Wilcoxon signed-rank tests1 revealed that the main effect of trial_type (i.e. the oddball effect) was indeed significant in both the left (z = 6.1, p b .001, r = .61) and right ears (z = 6.0, p b .001, r = .61). The oddball effect in the left ear was again found to be smaller in the stress compared to non-stress condition (z = 2.8, p b .005, r = .29). The main effect of stress found in the right ear during the ANOVA was however no longer significant when a Wilcoxon signed-rank test was performed (z = 1.5, p = .13). A Mann–Whitney test confirmed the presence of a significant gender*trial_type interaction in the right ear, with the oddball effect being significantly larger in women compared to men (z = 2.23, p b .05, r = .32). Trait anxiety did not predict the extent of the oddball effect (X2 = 0.21, p N .1), or the impact of stress upon it in either ear (left: X2 = 0.4, p N .1, right: X2 = 0.64, p N .1). 4. Discussion: Experiment 1

Fig. 3. Mean reaction times (ms) during the stress and non-stress conditions across standard and deviant trials. The analysis revealed that the increase in distraction during deviant trials was smaller in the stress condition than the non-stress condition, but only when the stimulus was delivered to left ear (A). No effect was found for the right ear (B). Errors bars represent the standard error of the mean.

Participants performed a novel auditory passive oddball task where the level of psychological stress was systematically manipulated within participants. The appearance of the deviant stimulus within the task induced an increase in reaction time, thus demonstrating the oddball effect. In the context of this task the size of the oddball effect reflects the relative difficulty in selectively attending to the task-relevant aspect of a sound. Of particular interest in the current study was the influence of both psychological stress and trait anxiety upon this oddball effect. Stress served to reduce the oddball effect for stimuli presented to the left ear. Trait anxiety did not however predict the size of the oddball effect, or the impact of the stress manipulation upon it. The findings of the current study appear to run counter to the predictions of attentional control theory (ACT: Eysenck et al., 2007). ACT proposes that negative emotional states, such as those provoked by psychological stress and high trait anxiety, reduce attentional control. ACT would therefore predict that the deviant stimuli in passive oddball tasks should cause greater distraction during stressful conditions. In contrast, during the current study stress was found to reduce the oddball effect, suggesting that the distracting information presented by the deviant stimuli was processed less in the stress condition. ACT also predicts that highly anxious individuals would show greater distraction (i.e. a larger oddball effect) than less anxious participants, regardless of emotional context. However levels of trait anxiety failed to predict the size of the oddball effect demonstrated by participants. In contrast to previous electrophysiological research (Elling et al., 2011) the current study therefore suggests that stress does not reduce attentional control 1 As no non-parametric equivalent of the 3-way repeated-measures ANOVA is readily available, 1-way non-parametric tests were used.

218

R. Hoskin et al. / Acta Psychologica 151 (2014) 214–221

over the processing of emotionally neutral auditory stimuli, and therefore does not disrupt selective auditory attention. The finding that the oddball effect was smaller in the stress condition lends support to the ‘goal-shielding’ theory of selective attention. This theory suggests that stress serves to reduce the attention paid to taskirrelevant information relative to that focused on task-relevant information (Chajut & Algom, 2003; Plessow et al., 2011). In particular the result is analogous with findings from the Stroop paradigm, where the cost to reaction time caused by the incongruent, task-irrelevant, semantic content of a word is reduced when the task is performed in stressful circumstances (Booth & Sharma, 2009; Chajut & Algom, 2003; Hu et al., 2012). The current study extends this result to a different type of distraction (based on stimulus probability rather than incongruence) and a different modality (auditory rather than visual). A particularly novel finding within the current study was that this goal-shielding effect was restricted to stimuli presented in the left ear (although the effect remained significant when responses to left and right ear stimuli were amalgamated). As the initial cortical processing of auditory information occurs in the hemisphere contralateral to input, this result suggests that the goal-shielding effect was only evident for stimulus processed largely in the right auditory cortex. As the processing of speech is generally thought to be lateralised to the left hemisphere (although see Hickok & Poeppel, 2007) it could be argued that the left ear 'goalshielding' effect found during the speech oddball task might be an artifact of the lateralised processing of the speech stimuli used in the task. For example it may be that the distracting speech content was processed more quickly for right ear/left hemisphere stimuli, thereby preventing stress from having a significant effect on such processing. Indeed the oddball effect was smaller for right ear stimuli, suggesting that the laterality of speech processing did indeed have some impact on task performance. To investigate whether the laterality of the goal-shielding effect evident during the speech oddball task might be due to the nature of the auditory stimulus used, it was decided to attempt to replicate the results using a slightly altered paradigm. To this end the speech oddball task was adapted so that tones differing in frequency were used as the oddball stimulus instead of speech. As frequency discrimination is not considered to be lateralised to the left hemisphere (Mathiak, Hertrich, Lutzenberger, & Ackermann, 2002) this ‘tone oddball’ task allows a test as to whether the laterality of the findings evident from the speech oddball task was due to the use of speech stimuli. If the left ear lateralisation of the goalshielding effect was an artifact of the laterality of speech processing, then it should not be evident when tonal stimuli are used instead. In contrast if the left-sided effect remains when tonal stimuli are used, this would suggest that the lateralisation of the goal-shielding effect has a more general cause. One alternative explanation for a left-sided goalshielding effect may be extrapolated from existing evidence which suggests that strong negative emotions such as stress are predominately processed in the right hemisphere (Demaree, Everhart, Youngstrom, & Harrison, 2005; Gainotti, 2012). If the stress induced goal-shielding effect were driven by a right-hemisphere mechanism this might explain why the effect was stronger for left-ear stimuli, since stress-triggered signals originating from the right hemisphere would presumably reach the right auditory cortex before the left auditory cortex. More generally performing a partial replication of the speech oddball task provides an opportunity to replicate the novel finding of a goalshielding effect relating to auditory selective attention. Implementing the tone oddball task also provides an opportunity to further investigate the unexpected gender*trial_type interaction found during the speech oddball task (larger oddball effect for women both overall, and for right ear stimuli on their own). Finally, despite participants reporting the stress blocks as being more stressful, the SCR data collected during the speech oddball task failed to support the prediction that the stress blocks would induce greater physiological arousal than the non-stress blocks. Performing an additional experiment using the same stress manipulation provides a further chance to assess the effectiveness of this stress manipulation.

5. Method: Experiment 2 5.1. Participants Forty-eight participants (29 female, mean age 25, σ = 8.4) took part in the experiment. The recruitment criteria was the same as Experiment 1; however, the sample was wholly independent of that used in Experiment 1. The study received ethical approval from the University of Sheffield Medical School Research Ethics Committee. 5.2. Task design The tone oddball task was similar to the speech oddball task with the exception that tonal rather than speech stimuli were used. Participants had to indicate in which ear a 250 ms tonal stimulus had been presented while ignoring task-irrelevant changes in the frequency of the tone (Fig. 4). Some other minor alterations were made to the task design: 1. Participants only completed 4 blocks of the task rather than 8 blocks. The length of each block was extended so that it involved 100 oddball trials, alongside 9 presentations of the audio-visual stimulus. Sixteen deviant trials were presented during each block, thus maintaining the frequency of deviant trials used during the speech oddball task (16%). 2. Only two different tones (with frequencies of 300 and 1200 Hz) were used during the task. Although this meant that only one stimulus acted as the deviant at any one time, piloting revealed that reducing the number of different stimuli acting as deviants did not alter the size of the oddball effect. All other aspects of the task design, such as the counterbalancing of stress vs. non-stress blocks, the positioning of the standard and deviant stimuli and the distribution of stay and switch responses were maintained from the design of the speech oddball task. 5.3. Stimuli and apparatus The tones used were created using the Audacity software and normalised using MatLab in a similar manner to that described for the speech oddball stimuli. All other stimuli were identical to those used before. 6. Results: Experiment 2 6.1. Stress manipulation The stress blocks were rated as more stressful than the non-stress blocks (z = −5.6, p b .001, r = .57). The correlation between trait anxiety and these stress ratings displayed a trend (r = .25, p = .09) towards the relationship found during Experiment 1 (more anxious participants rating the task as being more stressful). The error rates for the recognition tests were again very low (2%). SCR data (successfully recorded from 45 participants) did reveal a difference between the two conditions. The number of SCRs was significantly larger during the stress blocks (z = 2.10, p b .05, r = .23). The area SCR value was also greater during the stress blocks, although this difference did not reach significance (z = 1.49, p = .14). 6.2. Task performance Overall task performance (% correct) was 96%. The ANOVA of reaction times, with stress (stress, non-stress) trial_type (standard, deviant) and side (left, right) as factors, did not replicate the significant stress*trial_type*side interaction found during the speech oddball task (F(1,46) = 0.54, p N .1). However there was a four-way interaction with gender (F(1,46) = 4.03, p = .05, r = .28). A 3-

R. Hoskin et al. / Acta Psychologica 151 (2014) 214–221

219

Fig. 4. The tone oddball task. Participants had to respond as to which side a 250 ms tonal stimuli was delivered while ignoring task-irrelevant changes in the pitch of the tone which were introduced by delivering the tones in an oddball pattern. Each trial lasted 1250 ms (inclusive of the tone presentation).

way trial_type*side*gender interaction was also statistically significant (F(1,46) = 7.07, p b .05, r = .37). To assess the meaning of these interactions, separate ANOVAs were performed for each side of presentation. For stimuli presented to the left ear there was the expected main effect of trial_type (F(1,46) = 196.8, p b .001, r = .9) representing the oddball effect. The stress*trial_type*gender interaction was also significant (F(1,46) = 5.4, p b .05, r = .32). A post-hoc t-test revealed that the impact of stress on the oddball effect was significantly larger for females than for males (t(46) = 2.3, p b .05, r = .32). The oddball effect was smaller in the stress condition (reflecting the goal-shielding effect) for females (37 ms vs 53 ms) but not for males (43 ms vs 46 ms). Reaction times to left ear stimuli by gender are shown in Fig. 5.

For stimuli presented to the right ear there was no stress*trial_type*gender interaction (F(1,46) = 0.46, p N .1) suggesting that the 4-way interaction from the main ANOVA was due to the goalshielding effect for women being only present for left ear stimuli. Again there was a main effect of trial_type (F(1,46) = 123.8, p b .001, r = .85) representing the oddball effect. There was also a trial_type*gender interaction (F(1,46) = 4.88, p b .05, r = .31) with the oddball effect being larger in females than males (57 ms vs. 38 ms). This suggests that the trial_type*side*gender interaction from the main ANOVA was due to the oddball effect being larger for females but only for right ear stimuli. The data for right ear stimuli are presented in Fig. 6.

Fig. 5. Reaction times towards left ear stimuli for male (A) and female (B) participants. An interaction between stress and trial_type is only seen for female participants. Error bars represent the standard error of the mean.

Fig. 6. Reaction times towards right ear stimuli for male (A) and female (B) participants. The oddball effect (deviant – standard) was larger for females than for males. Error bars represent the standard error of the mean.

220

R. Hoskin et al. / Acta Psychologica 151 (2014) 214–221

As the reaction times during the tone oddball task differed significantly from the normal distribution (Shapiro–Wilk test), nonparametric statistics were used to assess whether the positive findings from the ANOVAs might be sensitive to the violations of the assumption of normality. A Mann–Whitney test revealed that the right ear oddball effect was larger for females than males (Z(46) = 2.4, p b .05). In contrast the left ear gender difference in the stress*trial_type interaction (i.e. the goal-shielding effect) only demonstrated a trend towards statistical significance (Z(46) = 1.7, p b .1). A Wilcoxon signed-rank test was used to test whether, for females only, the goal-shielding effect was evident for left ear stimuli. This analysis did reveal a significant goalshielding effect in the left ear for female participants (Z(28) = 2.4, p b .05, r = .3). Trait anxiety was not found to predict the size of the oddball effect (X2 = 1.25, p N .1), or the impact of stress upon it in either ear (left: X2 = 0.52, p N .1, right: X2 = 0.63, p N .1).

7. Discussion: Experiment 2 The results of the speech oddball task were not fully replicated during the tone task, as the 3-way stress*trial_type*side interaction was not found to be significant. However, in contrast to the speech oddball task, gender had an effect on this interaction. The reduced oddball effect for left ear stimuli was evident, but only for female participants. The trial_type*gender interaction for right ear stimuli that was evident during the speech task was also present during the tone task. The findings from the tone oddball task lend support to the conclusion (suggested by the results of the speech oddball task) that the goal-shielding effect applies to the auditory modality. Stress serves to reduce the distraction caused by emotionally neutral auditory stimuli during tasks requiring selective attention. Also mirroring the results of the speech task, the goal-shielding effect found during the tone oddball task was only evident for left ear stimuli. This suggests that the left ear lateralisation of the goal-shielding effect found during the speech oddball task was not an artifact of the properties of the auditory stimulus used. One possible alternative explanation for this left lateralisation may be that the goal-shielding effect of stress is driven by a right brain process. The proposition is supported by existing evidence that the right hemisphere is involved in the processing of strong negative emotions (Demaree et al., 2005; Gainotti, 2012) such as the psychological stress which triggers the goal-shielding effect. However behavioural responses do not provide any direct evidence of the neural location of cognitive functions, so this theory must be treated with caution. An alternative explanation might be that the left-sided laterality of the results is a consequence of the particular characteristics of the auditory oddball task used in this study. The conclusion that stress improves selective attention towards emotionally neutral auditory stimuli appears to contradict the predictions of attentional control theory (ACT: Eysenck et al., 2007). ACT suggests that psychological stress reduces attentional control, causing a greater vulnerability to distraction, the opposite effect to that predicted by goal-shielding theory. ACT also proposes that those with high trait anxiety demonstrate less attentional control (and therefore greater distractibility) than those with lower anxiety. The absence of a relationship between trait anxiety and the size of the oddball effect in either of the tasks described in this study therefore also contradicts ACT. Although the predictions of ACT were not supported during the current study, there is a sizeable amount of evidence in favour of ACT as regards the processing of threatening distracters (Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg, & van IJzendoorn, 2007; Eysenck et al., 2007). In contrast the goal-shielding effect only appears to be found during tasks (such as those performed during this study) where emotionally neutral stimuli are presented. This suggests that the effect of stress on attention can vary dramatically depending on the emotional valence of the stimuli being presented to the participant.

Unlike the speech oddball task, where the goal-shielding effect was exhibited regardless of gender, the reduced oddball effect under stress was only found to be statistically significant for female participants during the tone oddball task. Garcia-Garcia et al. (2008) found that females were more distracted by novel (non-speech) sounds during negative emotional conditions, although their task did not test selective attention (as the distracting sounds preceded the task stimuli). Along with the results of the tone oddball task, the findings of Garcia-Garcia et al. suggest that emotion may affect auditory attention more in females than in males. However it is not clear why there was no gender difference in the (statistically significant) effect of stress during the speech oddball task. Further research is therefore needed in order to elucidate gender differences in the impact of emotion on attention. Gender did not influence the difference in stress ratings given after the stress and non-stress conditions during the tone task (t(46) = 1.35, p N .1). There was also no effect of gender on the difference in either of the SCR measures between the stress and non-stress blocks (count: t(43) = 0.06, p N .1, area: t(43) = 1.39, p N 1) during the tone task. Hence it seems unlikely that a difference between males and females in the sensitivity to the stress manipulation would account for the gender difference in the impact of stress on attention found during the tone task. The effectiveness of the stress manipulation was confirmed by participant's self-report ratings of stress, which were greater after the stress than non-stress blocks during both experiments. Skin conductance response was also measured during each experiment. However this only revealed greater physiological arousal during the stress blocks in experiment 2. Other work in our laboratory (Hoskin et al., in press) has previously demonstrated that the method used to evoke a state of stress in the current study is successful in altering physiological arousal (measured via the skin conductance response). Alongside the results from the self-report ratings from the current study we therefore feel confident that the stress manipulation used in the current study was successful in altering the emotional state of the participant. In addition to the differing effects of the stress manipulation, there were also gender differences in the size of the oddball effect. During the speech oddball task females displayed a larger oddball effect than males regardless of the side of presentation. This effect was also evident when data for right ear stimuli were analysed in isolation. During the tone task female participants again exhibited a significantly larger oddball effect than males, but this effect was present only when stimuli presented to the right ear were considered. There is some existing evidence that females may be more vulnerable to distraction during selective attention tasks involving both visual (Stoet, 2010) and auditory-spatial (Zundorf, Karnath, & Lewald, 2011) distracters. The results of the current study suggest that females may also be more vulnerable to nonspatial auditory distraction. Interestingly gender differences have not regularly been found in the interference generated by the Stroop paradigm (Macleod, 1991). This suggests that while semantic content may be more distracting for women in the auditory modality (as illustrated by the results of the speech oddball task) this may not be the case in the visual modality. A further question that arises is why the greater distraction in females that existed regardless of the ear of presentation during the speech task was only present for right ear stimuli during the tone task? The greater selective attention to right-ear sounds in males (compared to females) might be related to males having greater right ear auditory-spatial acuity. For example Lewald (2004) found that males were better at auditory-spatial localisation when they were only able to use the right ear, but no different from females when only the left ear could be used. Adult females show less lateralised speech processing (i.e. less rightear dominance) than adult males (Hirnstein, Westerhausen, Korsnes, & Hugdahl, 2013) and may process semantic material to a greater extent than males (Judge & Taylor, 2012). These findings are consistent with the increased distraction in females to left-ear stimuli during the speech task. Nevertheless further research is needed to understand gender differences in auditory selective attention. Gender differences in cognitive

R. Hoskin et al. / Acta Psychologica 151 (2014) 214–221

functioning are of interest because they may allow more tailored (i.e. gender-specific) educational and neuropsychological interventions to be developed. 8. Conclusion Two auditory passive oddball tasks were conducted to identify the effect of psychological stress on auditory selective attention. Stress was manipulated by presenting either aversive or neutral audio-visual stimuli between trials of the auditory tasks. Subjective ratings and, during the second task, SCR data, suggested that this manipulation was successful in inducing psychological stress in participants. Selective attention generally improved during stressful conditions, but only towards stimuli presented to the left ear. These results largely support the presence of a goal-shielding effect of stress on selective attention. In contrast no evidence was found to support the prediction (arising from attentional control theory) that psychological stress or high trait anxiety causes a greater vulnerability to distraction by emotionally neutral stimuli. These findings add to the ongoing discussion concerning the impact of stress on attention, by demonstrating the existence of the goal-shielding effect in the auditory modality. Acknowledgements This research was funded by the Sheffield University Scholarship Scheme. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.actpsy.2014.06.010. References Bar-Haim, Y., Lamy, D., Pergamin, L., Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H. (2007). Threat-related attentional bias in anxious and nonanxious individuals: A meta-analytic study. Psychological Bulletin, 133(1), 1–24. http://dx.doi.org/10.1037/ 0033-2909.133.1.1. Booth, R., & Sharma, D. (2009). Stress reduces attention to irrelevant information: Evidence from the Stroop task. Motivation and Emotion, 33(4), 412–418. http://dx.doi. org/10.1007/s11031-009-9141-5. Caparos, S., & Linnell, K. J. (2012). Trait anxiety focuses spatial attention. Emotion, 12(1), 8–12. http://dx.doi.org/10.1037/A0026310. Chajut, E., & Algom, D. (2003). Selective attention improves under stress: Implications for theories of social cognition. Journal of Personality and Social Psychology, 85(2), 231–248. http://dx.doi.org/10.1037/0022-3514.85.2.231. Demaree, H. A., Everhart, D. E., Youngstrom, E. A., & Harrison, D. W. (2005). Brain lateralization of emotional processing: Historical roots and a future incorporating "dominance". Behavioral and Cognitive Neuroscience Reviews, 4(1), 3–20. http://dx.doi.org/ 10.1177/1534582305276837. Elling, L., Steinberg, C., Brockelmann, A. K., Dobel, C., Bolte, J., & Junghofer, M. (2011). Acute stress alters auditory selective attention in humans independent of HPA: A study of evoked potentials. PLos One, 6(4). http://dx.doi.org/10.1371/journal.pone. 0018009. Eysenck, M. W., Derakshan, N., Santos, R., & Calvo, M. G. (2007). Anxiety and cognitive performance: Attentional control theory. Emotion, 7(2), 336–353. http://dx.doi.org/ 10.1037/1528-3542.7.2.336.

221

Gainotti, G. (2012). Unconscious processing of emotions and the right hemisphere. Neuropsychologia, 50(2), 205–218. http://dx.doi.org/10.1016/j.neuropsychologia. 2011.12.005. Garcia-Garcia, M., Dominguez-Borras, J., SanMiguel, I., & Escera, C. (2008). Electrophysiological and behavioral evidence of gender differences in the modulation of distraction by the emotional context. Biological Psychology, 79(3), 307–316. http://dx.doi.org/10. 1016/j.biopsycho.2008.07.006. Hickok, G., & Poeppel, D. (2007). Opinion - The cortical organization of speech processing. Nature Reviews Neuroscience, 8(5), 393–402. http://dx.doi.org/10.1038/Nrn2113. Hirnstein, M., Westerhausen, R., Korsnes, M. S., & Hugdahl, K. (2013). Sex differences in language asymmetry are age-dependent and small: A large-scale, consonant-vowel dichotic listening study with behavioral and fMRI data. Cortex, 49(7), 1910–1921. http://dx.doi.org/10.1016/j.cortex.2012.08.002. Hoskin, R., Hunter, M. D., & Woodruff, P. W. R. (2014). The effect of psychological stress and expectation on auditory perception: A signal detection analysis. British Journal of Psychology. http://dx.doi.org/10.1111/bjop.12048. Hoskin, R., Hunter, M. D., & Woodruff, P. W. R. (2014). Neither State or Trait Anxiety alter the response to distracting emotionally neutral sounds. Experimental Psychology. http://dx.doi.org/10.1027/1618-3169/a000268 (in press). Hu, K. S., Bauer, A., Padmala, S., & Pessoa, L. (2012). Threat of bodily harm has opposing effects on cognition. Emotion, 12(1), 28–32. http://dx.doi.org/10.1037/A0024345. Judge, J., & Taylor, P. J. (2012). Gender differences on the semantic flanker task using transposed-letter target words. Quarterly Journal of Experimental Psychology (Hove), 65(10), 2008–2017. http://dx.doi.org/10.1080/17470218.2012.676654. Lang, P. J., Bradley, M. M., & Cuthbert, B. B. (2008). International affective picture system (IAPS): Affective ratings of pictures and instruction manual. Technical Report A-8.: University of Florida, Gainesville, FL. Lazarus, R. S. (1993). From psychological stress to the emotions—a history of changing outlooks. Annual Review of Psychology, 44, 1–21. http://dx.doi.org/10.1146/annurev. ps.44.020193.000245. Lewald, J. (2004). Gender-specific hemispheric asymmetry in auditory space perception. Brain Research. Cognitive Brain Research, 19(1), 92–99. http://dx.doi.org/10.1016/j. cogbrainres.2003.11.005. Macleod, C. M. (1991). Half a century of research on the Stroop effect—an integrative review. Psychological Bulletin, 109(2), 163–203. http://dx.doi.org/10.1037//0033-2909. 109.2.163. Mathiak, K., Hertrich, I., Lutzenberger, W., & Ackermann, H. (2002). Functional cerebral asymmetries of pitch processing during dichotic stimulus application: A wholehead magnetoencephalography study. Neuropsychologia, 40(6), 585–593. http://dx. doi.org/10.1016/S0028-3932(01)00159-2. Moser, J. S., Becker, M. W., & Moran, T. P. (2012). Enhanced attentional capture in trait anxiety. Emotion, 12(2), 213–216. http://dx.doi.org/10.1037/A0026156. Oldfield, R. C. (1971). Assessment and analysis of handedness: The Edinburgh inventory. Neuropsychologia, 9(1), 97–113. Parmentier, F. B. R., & Andres, P. (2010). The involuntary capture of attention by sound novelty and postnovelty distraction in young and older adults. Experimental Psychology, 57(1), 68–76. http://dx.doi.org/10.1027/1618-3169/A000009. Parmentier, F. B. R., Elsley, J. V., Andres, P., & Barcelo, F. (2011). Why are auditory novels distracting? Contrasting the roles of novelty, violation of expectation and stimulus change. Cognition, 119(3), 374–380. http://dx.doi.org/10.1016/j.cognition.2011.02.001. Plessow, F., Fischer, R., Kirschbaum, C., & Goschke, T. (2011). Inflexibly focused under stress: Acute psychosocial stress increases shielding of action goals at the expense of reduced cognitive flexibility with increasing time lag to the stressor. Journal of Cognitive Neuroscience, 23(11), 3218–3227. Serlin, R. C., & Harwell, M. R. (2004). More powerful tests of predictor subsets in regression analysis under nonnormality. Psychological Methods, 9(4), 492–509. http://dx. doi.org/10.1037/1082-989x.9.4.492. Spielberger, C. D., Gorsuch, R. L., Lushene, R., Vagg, P. R., & Jacobs, G. A. (1983). Manual for the State-Trait Anxiety Inventory (STAI). Palo Alto, California: Consulting Psychologists Press. Stoet, G. (2010). Sex differences in the processing of flankers. Quarterly Journal of Experimental Psychology (Hove), 63(4), 633–638. http://dx.doi.org/10.1080/17470210903464253. Zundorf, I. C., Karnath, H. O., & Lewald, J. (2011). Male advantage in sound localization at cocktail parties. Cortex, 47(6), 741–749. http://dx.doi.org/10.1016/j.cortex.2010.08. 002.