Emotional Stroop Dilution: The boundary conditions of attentional capture by threat words

Acta Psychologica 159 (2015) 108–115

Contents lists available at ScienceDirect

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

Emotional Stroop Dilution: The boundary conditions of attentional capture by threat words Michael G. Reynolds a,⁎, Robin M. Langerak b a b

Trent University, Canada Carleton University, Canada

a r t i c l e

i n f o

Article history: Received 14 October 2014 Received in revised form 7 May 2015 Accepted 11 May 2015 Available online 18 June 2015 Keywords: Emotional Stroop Attention Threat Cognitive slowing Automaticity Visual word recognition

a b s t r a c t It is widely believed that threatening stimuli in our environment capture attention. Much of the core evidence for attentional capture by threatening stimuli comes from the Emotional Stroop task. Yet recent evidence suggests that the Emotional Stroop task does not measure attentional capture (e.g., Algom et al., 2004). The present paper assesses whether threat words can capture attention using a modified Stroop Dilution procedure (e.g., Kahneman & Chajczyk, 1983), where attentional capture by a threat word is inferred from a reduction in color-word interference for threat words compared to non-threat words (emotional Stroop Dilution). The outcome of the present experiments indicates that threat words can capture attention, but only when task demands do not require that a word be attended. It is suggested that threat words produce (1) cognitive slowing, and influence two processes of selective attention (2) attentional capture and (3) the ability to filter irrelevant dimensions of an attended stimulus. © 2015 Elsevier B.V. All rights reserved.

1. Introduction In order to function adaptively, people need to selectively attend to information that is relevant to their ongoing activities and goals (James, 1890). Given the survival value of evaluating affectively laden stimuli, it is not surprising that there is a large body of work consistent with the claim that negatively valenced information captures our attention (e.g., Reynolds, Eastwood, Partanen, Frischen, & Smilek, 2009). One task that has been used to assess whether negatively valenced stimuli capture attention is the Emotional Stroop task (e.g., McKenna & Sharma, 1995, 2004; Williams, Watts, MacLeod, & Mathews, 1997). In the Emotional Stroop task, participants are asked to identify the inkcolor of threat words like “die” and non-threat words like “pie” (see Yiend, 2010; Williams, Mathews, & MacLeod, 1996 for a review). Usually, the different types of words are presented in separate blocks (see Algom, Chajut, & Lev, 2004; McKenna & Sharma, 2004). Under such circumstances, participants are slower to name the ink-color of threat compared to non-threat words (the Emotional Stroop effect). The received view is that the Emotional Stroop effect arises because threat words automatically capture attention away from the ink-color, thereby increasing the time required to respond on these trials (e.g., Williams et al., 1996).

⁎ Corresponding author at: Department of Psychology, Trent University, 1600 West Bank Drive, Peterborough, Ontario K7P 1G8, Canada. E-mail address: [email protected] (M.G. Reynolds).

http://dx.doi.org/10.1016/j.actpsy.2015.05.008 0001-6918/© 2015 Elsevier B.V. All rights reserved.

Research with the Emotional Stroop task has long demonstrated that the magnitude of the Emotional Stroop effect is affected by depression and anxiety, and has played a formative role in the development of cognitive models of the corresponding psychopathologies (Williams et al., 1997). The Emotional Stroop task continues to be widely used to study attentional biases for threat as a function of early childhood trauma (Wingenfeld et al., 2009), attachment (Atkinson et al., 2009), borderline personality disorder and posttraumatic stress disorder (Cisler et al., 2011), among others. It has also been used to identify the neural sources associated with emotion (Mitterschiffthaler et al., 2008). 1.1. Problems with the Emotional Stroop task Recently, a number of researchers have expressed concern that the Emotional Stroop task does not measure attentional biases for threatening information (Algom et al., 2004; Frings & Wühr, 2012; Larsen, Mercer, & Balota, 2006; McKenna & Sharma, 2004). In a seminal paper, Algom et al. (2004) assessed whether the Emotional Stroop effect demonstrates five diagnostic tests of attention, namely whether the effect was (1) sensitive to irrelevant variation, (2) asymmetric for words and colors, (3) dependent upon the salience of the relevant and irrelevant dimensions, (4) observed when neutral and threat words were presented in a mixed list context, and (5) whether the emotion words affected color naming, but not word reading. Given that the Emotional Stroop effect did not pass any of these diagnostic tests, Algom et al. concluded that it does not arise from an attentional bias for threatening information. Instead, they proposed that the Emotional Stroop task

M.G. Reynolds, R.M. Langerak / Acta Psychologica 159 (2015) 108–115

measures generic cognitive slowing (see also Öhman et al., 2001; McKenna & Sharma, 2004; Chajut, Mama, Levy, & Algom, 2010; and Frings & Wühr, 2012 for additional non attentional capture based accounts). According to the cognitive slowing account, the threat word is processed by a special threat detection system that interrupts all ongoing cognitive processes when a threat is detected. This interruption slows performance in the presence of a threatening stimulus. Unlike the standard interpretation of the Emotional Stroop effect, the cognitive slowing account has nothing to do with the relationship between the ink-color and the word. Therefore, performance decrements do not arise because of a failure to selectively attend to the ink-color. The conclusion that performance decrements in the Emotional Stroop task do not arise from an automatic attentional bias for threat has important implications for widely held cognitive models of psychopathologies, as well as their diagnostic counterparts, where attentional biases are attributed to individual difference factors such as anxiety and depression (e.g., Derryberry & Reed, 2002; MacLeod, Andrew, & Tata, 1986). This is particularly problematic for models where converging evidence from other methods has not been demonstrated. 1.2. An “Emotional” Stroop task Although several studies have provided evidence that threat words elicit cognitive slowing (Algom et al., 2004; Frings et al., 2010; McKenna & Sharma, 2004), whether threat words also capture selective attention is still an open question (e.g., Frings & Wühr, 2012; Mama, Behn-Haim, & Algom, 2012). One way to assess whether threat words capture attention would be to modify the Emotional Stroop task so that it passes the five diagnostic tests proposed by Algom et al. (2004). Here, we take this approach by modifying the traditional Emotional Stroop task so that it will be more like the original Stroop task. In the original Stroop task, participants were asked to name an ink-color while ignoring a concurrently presented color word. The stimuli in the Stroop task have a congruency relationship such that the color word and ink-color can be congruent (“blue” in blue ink), incongruent (“blue” in red ink) or neutral (“car” in blue ink). The Stroop effect (slower responses when naming the inkcolor on incongruent trials compared to congruent trials) is a hallmark failure of selective attention (Algom et al., 2004; MacLeod, 1991) and is sensitive to the five diagnostics of attention used by Algom et al. (2004) to examine the Emotional Stroop effect. Therefore, adding a congruency component to the Emotional Stroop task makes it possible to assess whether threat words can automatically capture attention. Recently, Chajut, Schupak and Algom (2010) added a congruency component to the Emotional Stroop task to assess whether there is an attentional bias for threat words. In Chajut et al.'s experiments, two stimuli were presented on each trial: a color word and a distractor word (threat vs. non-threat; see Cho, Lien, & Proctor, 2006; Kahneman & Henik, 1981; Roberts & Besner, 2005 for the use of this procedure without threat words). The distractor word was always the colored item and the color word could either be congruent or incongruent with the target color (see Fig. 1, panel A). Participants were instructed ignore the meaning of the words while naming the ink-color of the distractor word. Previous research using this procedure (without emotional words) has indicated that when the color carrying stimulus is a

109

word-like stimulus, the impact of the color word is reduced (Roberts & Besner, 2005). This effect is often called Stroop Dilution (Kahneman & Chajczyk, 1983) and is argued to arise because the word-like stimuli at fixation utilize attention. Consistent with automatic attentional capture by the threat words, the Stroop effect in Chajut, Schupak et al.'s (2010) study was smaller in the presence of a threat word than in the presence of a non-threat word (emotional Stroop Dilution). 2. Experiment 1 Although Chajut, Schupak et al.'s (2010) experiments are consistent with automatic attentional capture by threat words, the type of distractor was blocked such that in one block the distractor was always a threat word and in another block the distractor was always a nonthreat word. Blocking the emotional valence of the distractor word raises the possibility that differences in performance across the threat and non-threat conditions were due to top-down differences in attentional set, rather than differences in attentional capture (Algom et al., 2004; Francolini & Egeth, 1980). In order to assess whether the emotional Stroop Dilution reported by Chajut, Schupak et al. (2010) is due to attentional capture, the first experiment assesses whether emotional Stroop Dilution is still observed when the threat and non-threat distractors are randomly intermixed in a single block of trials. As noted by Algom et al. (2004), one of the hallmarks of automatic attentional capture is that its effects are still observed under mixed list conditions. Indeed, the conclusion that the Emotional Stroop effect does not arise from the capture of selective attention away from the color and towards the word is predicated, in part, on the observation that the Emotional Stroop effect is eliminated in a mixed list context (Algom et al., 2004; Frings et al., 2010; Holle, Neely, & Heimberg, 1998; Richards, French, Johnson, Naparstek, & Williams, 1992; McKenna & Sharma, 2004). Therefore, a failure to observe emotional Stroop Dilution when the stimuli are randomly intermixed would suggest that the emotional Stroop Dilution reported by Chajut, Schupak et al. (2010) is not due to attentional capture by a threat word. 2.1. Method 2.1.1. Participants Eighteen students from Trent University participated in the present study for credit in an undergraduate psychology course. All students reported normal or corrected to normal vision and normal color perception. 2.1.2. Stimuli The ink-colors were the standard E-Prime colors for red, yellow blue and green (Schneider, Eschman, & Zuccolotto, 2002). The word stimuli consisted of two sets of character strings, color words and distractor words. The color words consisted of the words RED, YELLOW, BLUE and GREEN and neutral nonwords (e.g., #&@). The nonwords were derived from non-alphanumeric characters from the top of the keyboard and matched to the color-words on length (see Roberts & Besner, 2005). The color words were used in conjunction with the ink-colors

Fig. 1. Examples of the displays used in Experiment 1 (panel A), Experiment 2 (panel B) and Experiment 3 (panel C).

110

M.G. Reynolds, R.M. Langerak / Acta Psychologica 159 (2015) 108–115

to create the congruent, incongruent and neutral conditions. The distractor stimuli consisted of 96 threat words and 96 non-threat words. Recent evidence suggests that many psycholinguistic variables known to affect how quickly words are read are confounded with emotionality in most emotional Stroop studies (Larsen et al., 2006). Under such conditions, the emotional Stroop effect may not measure attentional capture by threatening information. Instead, it may only indicate ease of lexical processing due to other factors (e.g., word frequency). Therefore, the threat and non-threat words were matched on onset phoneme and equated on letter length, printed word frequency (Kucera & Francis, 1967), Hyperspace Analogue to Language (HAL) frequency norms (Lund & Burgess, 1996; see also Balota et al., 2007), log HAL frequency, orthographic neighborhood density (N), and naming response time (RT; see Table 1) using the database provided by Larsen et al. (2006). 2.1.3. Displays The displays consisted of two character strings (a color word and a distractor word). In each display, the distractor word served as the color carrier. The strings were presented in upper case using an 18 pt Courier New font. Strings subtended approximately .5° horizontally and up to 3° vertically. The strings were centered horizontally and vertically at fixation with one above the other with a .2° visual angle separation. The position of the color word and the distractor word (threat or non-threat) was randomized across trials (see Fig. 1, Panel A). 2.1.4. Apparatus The experiment was run on a Dell Vostro 420 Desktop computer with a core 2 quad processor, an ATI Radion 4350 video card, and the Windows XP operating system with service pack 2. Stimuli were presented on a Dell E207 LCD screen. Data collection and stimulus presentation were controlled using E-Prime 2.0 software and a PST Response Box, microphone, and voice-key assembly (Schneider et al., 2002). 2.1.5. Procedure The experiment consisted of two blocks of trials. The practice block consisted of 12 trials and the experimental block consisted of 192 trials. In both blocks, the ratio of congruent, incongruent and neutral trials was 1:1:1 and the ratio of threat and non-threat trials was also 1:1. This resulted in 32 observations per condition per subject. Each color served as the target 8 times per condition. All conditions were randomly intermixed. The randomization procedure was performed without replacement until all conditions were sampled twice. The assignment of a distractor word to the congruent, incongruent or neutral Stroop conditions was counterbalanced across subjects. The assignment of a distractor to a specific target color was random. Each trial began with a white fixation marker (‘+’) on a black background in the center of the screen for 500 ms. Subjects were instructed to fixate at this location. The fixation marker was replaced by a blank screen for 500 ms, followed by the target display until a vocal response was made. The researcher then coded the identity of the verbal response made by the subject or classified it as a mistrial (e.g., voice key failure) via button press. Mistrials only consisted of correct trials where a hardware failure affected the interpretability of the RT data. Table 1 Characteristics of the stimuli used in Experiments 1–3.

Number of stimuli Letter length Kucera & Francis WF HAL word frequency Log HAL word frequency Number of orthographic neighbors Mean naming RT

Nonthreat

Threat

96 4.9 47.9 21,733.4 9.1 4.3 617.0

96 5.0 48.6 22,964.5 9.3 4.3 612.8

2.2. Results The present study contains two random factors — subjects and items. The data were therefore analyzed once with subjects as a random factor (Fs) and once with items as a random factor (Fi). Analyzing the data in this way makes it possible to assess whether an effect is present for subjects and for items (Forster & Dickson, 1976). The subject data were analyzed using a repeated-measures analysis of variance (ANOVA) with congruency (congruent vs. incongruent) and distractor (threat vs. non-threat) as factors. The item data were analyzed using a repeated-measures ANOVA with congruency as a repeated factor and distractor (threat vs. non-threat) as an independent factor. In order to reduce the impact of subject variability on item estimates, the RT data for each subject was z-scored before calculating item medians.1 2.2.1. Response time Median RTs were analyzed. Trials on which a voice-key error occurred (6.5%) or an error was made (2.4%) were excluded from the reaction time (RT) data. The means of median RTs can be seen in Table 2. Overall, a robust Stroop effect was observed, with responses taking approximately 33 ms longer on incongruent trials than on congruent trials, Fs(1,17) = 14.258, p = .002, MSE = 1342.634, η2 = .46; Fi(1, 190) = 16.314, p b .001, MSE = .096, η2 = .079. There was no main effect of distractor word type (Fs b 1). Critically, if the presence of a threat word reduces the extent to which the color word is attended (i.e., by capturing attention), then an interaction should be observed, whereby the Stroop effect is smaller in the threat than in the nonthreat condition. Inconsistent with threat words capturing attention, no interaction was observed (Fs b 1). The Stroop effect in the threat condition (30 ms) was not reliably different from the non-threat condition (36 ms). The power to detect a 40 ms effect as was reported by Chajut, Schupak et al. (2010) two-tailed at p b .05 was .91. 2.2.2. Percent error There was some evidence for a Stroop effect in the error rates with more errors on incongruent trials than on congruent trials. However, this difference did not reach the criterion for significance, Fs(1, 17) = 3.537, p = .077 MSE = 4.639, η2 = .172; Fi(1, 190) = 1.535, p = .217, MSE = 52.963, η2 b .012. There was no evidence for a main effect of threat (Fs b 1), nor was there any evidence for an interaction between these factors (Fs (1, 17) = 2.056, p = .170 MSE = 11.151, η2 = .108; Fi(1, 190) = 2.356, p = .126, MSE = 52.963, η2 = .012). Therefore, the error data do not compromise the interpretation of the RT data. 2.2.3. Neutral condition In order to assess whether there was any impact of the threat word in the absence of a color-word, we compared performance for the threat and non-threat conditions on neutral trials. As can be seen in Table 2, there was no evidence of an effect of threat on RT (ts b 1.2). There was some evidence for an effect of threat on errors, but it did not reach criterion for subjects (ts[17] = 1.844, p = .083), or items (ti[191] b 1). Indeed, the pattern in the error data is opposite to what would be expected if threat words were impeding performance. The data therefore replicate the observation that there is no difference in color naming performance for threat and non-threat words when they are randomly intermixed (Algom et al., 2004; Frings et al., 2010; Richards et al., 1992; McKenna & Sharma, 2004). 1 There is substantial debate about how to address the issue of two random factors in psycholinguistic research (Baayen, 2008; Barr, Levy, Scheepers, & Tily, 2013; Raaijmakers, Schrijnemakers, & Gremmen, 1999). To satisfy possible concerns about the generality of the present findings we therefore reanalyzed the data as suggested by Baayen (2008); Barr et al. (2013). For these data analyses, the raw correct trial level data were used instead of aggregated data at the subject and item level. The models included congruency (congruent vs. incongruent), distractor (threat vs. neutral) as crossed factors and trial number as a covariate. For both types of analyses the results of the experiments were unchanged.

M.G. Reynolds, R.M. Langerak / Acta Psychologica 159 (2015) 108–115 Table 2 Mean of median RT (ms) and percentage error in Experiment 1 as a function of distractor type (threat or non-threat word) and congruency (congruent, incongruent, or neutral word). Response time

Incongruent Congruent Difference Neutral

Percentage error

Nonthreat

Threat

Nonthreat

Threat

669 633 36 652

672 642 30 651

3.8 2.3 1.5 2.3

2.6 2.8 −0.2 1.2

111

task relevant object would provide a better opportunity for threat words to capture attention away from the target color. The goal of Experiment 2, therefore, was to assess whether threat words automatically capture attention when the distractor word is spatially separated from the target color. The color words therefore served as the color carrier and the threat and non-threat distractor words were presented in white in Experiment 2. If threat words automatically capture attention and evidence of attentional capture is better observed when the threat and non-threat words are spatially separated from the color carrier, then emotional Stroop Dilution should be observed. 3.1. Method

2.3. Discussion Experiment 1 examined whether the emotional Stroop Dilution reported by Chajut, Schupak et al. (2010) was due to automatic attentional capture by the threat words. One of the diagnostic criteria used by Algom et al. (2004) was that if the effect of threat words was due to automatic attentional capture then it should be observed under mixed-list conditions. Therefore, whether the emotional Stroop Dilution reported by Chajut, Schupak et al. (2010) was due to attentional capture was tested by assessing whether emotional Stroop Dilution would be observed when all conditions were randomly intermixed. As can be seen in Table 2, emotional Stroop Dilution was not observed under mixed-list conditions in Experiment 1. The absence of emotional Stroop Dilution when the items are randomly intermixed suggests that the effect observed by Chajut, Schupak et al. (2010) does not reflect automatic attentional capture away from the color and towards the threat word. This conclusion is further supported by the absence of a traditional emotional Stroop effect in the present experiment when the color-word stimulus was a neutral string of non-alphanumeric symbols from the top of the keyboard (see Algom et al., 2004). The present data suggest that the Stroop Dilution observed by Chajut, Schupak et al. (2010) is better attributed to top-down modulation of attentional set, rather than differences in attentional capture (Algom et al., 2004; Francolini & Egeth, 1979). Evidence suggests that when threat and non-threat words are presented in separate blocks, task irrelevant information in an attended object (e.g., the meaning of the word) is less likely to be ignored when it is threatening than when it is non-threatening (e.g., Frings & Wühr, 2012; Mama et al., 2012). Here, we hypothesize that the Stroop Dilution observed by Chajut, Schupak et al. (2010) arose because blocking the threat and nonthreat words led to a top-down modulation of how irrelevant information in the color carrier was filtered. According to this account, the filter is weaker in the presence of threatening stimuli. This creates a difference in the efficiency of the filter for threat and non-threat words when they are presented in separate blocks. However, in a mixed list context, the presence of threat words changes the filter for both the threat and non-threat words, resulting in difficulty filtering both types of items. This account predicts that the Stroop Effect in Experiment 1 (where both items are likely to make it through the filter) reported here should be smaller than the non-threat condition reported by Chajut, Schupak et al. (2010), and this is what is observed (33 ms in Experiment 1 vs. 57 ms in Chajut et al., Experiment 1).

3. Experiment 2 No evidence of automatic attentional capture was observed in Experiment 1. Instead, the outcome of Experiment 1 suggested that threat words can influence top down control of an attentional filter that determines whether unattended properties of an attended object affect performance. One explanation for the absence of automatic attentional capture by the threat words in Experiment 1 is that it is difficult to observe the effects of attentional capture when the threat word is part of the attended object. If so, then spatially separating the distractor stimulus from the

3.1.1. Participants Eighteen students from Trent University participated in the present study for credit in an undergraduate psychology course. All students reported normal or corrected to normal vision and normal color perception. 3.1.2. Stimuli The stimuli consisted of the character strings from Experiment 1. 3.1.3. Displays The displays were identical to Experiment 1, except that the Stroop stimuli (color words and neutral character strings) were colored and the distractors (threat and non-threat words) were presented in white (see Fig. 1, Panel B). 3.1.4. Apparatus Identical to Experiment 1. 3.1.5. Procedure Identical to Experiment 1. 3.2. Results The item “tutor” was excluded from the present analysis for yielding a Stroop effect that was 3.5 standard deviations away from the mean in its cell. The data were analyzed in the same manner as Experiment 1. 3.2.1. Response time Median RTs were analyzed. Trials on which a voice-key failure occurred (3.6%) or an error was made (4.1%) were excluded from the reaction time (RT) data. The means of median RTs can be seen in Table 3. Overall, a robust Stroop effect was observed with responses taking approximately 147 ms longer on incongruent trials than on congruent trials, Fs(1,17) = 96.397, p b .001, MSE = 4062.478, η2 = .850; Fi(1, 189) = 4447.471, p b .001, MSE = .062, η2 = .703. The observation of a larger Stroop effect in Experiment 2 compared to Experiment 1 is consistent with the color carrier (the color word in Experiment 2 and the distractor word in Experiment 1) being attended. Once again there was no main effect of distractor type (Fs b 1). Inconsistent with threat words capturing attention, there was no evidence of an interaction (Fs b 1). That is, the Stroop effect in the threat condition (145 ms) was not reliably different than in the non-threat condition (150 ms). The

Table 3 Mean of median RT (ms) and percentage error in Experiment 2 as a function of distractor type (threat or non-threat word) and congruency (congruent, incongruent, or neutral word). Response time

Incongruent Congruent Difference Neutral

Percentage error

Nonthreat

Threat

Nonthreat

Threat

753 603 150 623

747 602 145 630

10.1 1.0 9.1 2.3

8.3 1.0 7.3 1.7

112

M.G. Reynolds, R.M. Langerak / Acta Psychologica 159 (2015) 108–115

power to detect a 40 ms effect as was reported by Chajut, Schupak et al. (2010) two-tailed at p b .05 was .99. 3.2.2. Percent error A robust Stroop effect was observed, where more errors were made on incongruent trials than on congruent trials, Fs (1, 17) = 31.925, p b .001, MSE = 37.507, η2 = .653; Fi(1, 188) = 74.141, p b .001, MSE = 96.238, η2 b .282. There was no main effect of distractor word type (Fs (1, 17) = 2.220, p = .154, MSE = 6.096, η2 = .116; Fi(1, 191) b 1), nor was there any evidence for an interaction (Fs b 1). Therefore, the error data do not compromise the interpretation of the RT data. 3.2.3. Neutral condition As can be seen in Table 4, there was no evidence of an effect of threat on RT, (ts b 1) or on errors (ts b 1). 3.3. Discussion Experiment 2 assessed whether evidence of automatic attentional capture would be observed when the threat word was spatially separated from the target color. To examine this possibility, the color word served as the color carrier in Experiment 2 with the distractor word being presented either slightly above or slightly below the color word. As in Experiment 1, there was no evidence for emotional Stroop Dilution. This outcome is inconsistent with threat words automatically capturing attention (e.g., Algom et al., 2004; McKenna & Sharma, 2004; Frings & Wühr, 2012) and suggests that the failure to observe evidence of attentional capture in Experiment 1 was not due to the task demands requiring that the threat words be attended. 4. Experiment 3 One characteristic often ascribed to automatic processes is that they do not require processing capacity (see Reynolds & Besner, 2006 for a more elaborate discussion). A plausible explanation for why automatic attentional capture was not observed in Experiment 2 is that threat words have the same attentional demands as other words. Evidence from visual word recognition suggests that the process of recognizing non-threat words is capacity limited (McCann, Remington, & Van Selst, 2000; Reynolds & Besner, 2006; Roberts & Besner, 2005). Consistent with this conclusion, Cho et al. (2006) reported that when a color-word is the color carrier (as in Experiment 2), the presence of an additional (nonthreat) word does not affect the magnitude of the Stroop effect. The observation that the Stroop effect did not differ for threat words and nonthreat words in Experiment 2 suggests that the threat and non-threat distractor words both require access to the limited capacity mechanism. The purpose of Experiment 3, was to assess whether evidence of automatic attentional capture was not observed in Experiment 2 because threat words, like non-threat words, are processed using a limited capacity processing mechanism. If the distractor word and the color word both require access to the same capacity limited mechanism then this suggests that only one of the words will be processed to completion on any given trial (Cho et al., 2006; Roberts & Besner, 2005) or

Table 4 Mean of median RT (ms) and percentage error in Experiment 3 as a function of distractor type (threat or non-threat word) and congruency (congruent, incongruent, or neutral word). Response time

Incongruent Congruent Difference Neutral

Percentage error

Nonthreat

Threat

Nonthreat

Threat

592 531 61 562

579 544 35 563

5.6 2.8 2.8 2.9

5.6 3.0 2.6 3.3

processed with sufficient priority so as to influence performance. This was problematic in Experiments 1 and 2 because a word always served as the color carrier and was therefore more likely to be processed on each trial. To avoid this problem, Kahneman and Chajczyk's (1983) Stroop Dilution procedure was used. In this procedure, three stimuli are presented on each trial: a rectangular color patch, a color word, and a distractor word (for studies that have used this procedure without emotional stimuli see Brown, Roos-Gilbert, & Carr, 1995; Brown, Gore, & Carr, 2002; Cho, Choi, & Proctor, 2012; Mitterer, La Heij, & Van der Heijden, 2003; Yee & Hunt, 1991). Critically, the color patch is always presented at fixation so the location of the task relevant stimulus is known with absolute certainty. In addition, the locations of the color and distractor words are randomly assigned on a trial-by-trial basis so that on half of the trials the distractor word appears directly above the color patch with the color word directly below the color patch and vice versa on the other half of the trials (Cho et al., 2012; Kahneman & Chajczyk, 1983; Yee & Hunt, 1991). All of the conditions are randomly intermixed in a single block of trials and the participants' task is to name aloud the color of the color patch. Work by Cho et al. (2012) suggests that in this context, the magnitude of the Stroop effect is determined by the likelihood that the distractor word will capture attention. If threat words capture attention more often than the non-threat words, then the Stroop effect will be smaller when the distractor is a threat word compared to a non-threat word. Evidence of attention capture under these conditions would be consistent with threat words the same limited capacity processing mechanism as non-threat words. 4.1. Method 4.1.1. Participants Eighteen students from Trent University participated in the present study for credit in an undergraduate psychology course. All students reported normal or corrected to normal vision and normal color perception. 4.1.2. Stimuli Same as Experiments 1 and 2. 4.1.3. Displays A rectangular color patch was centered at fixation and subtended approximately .5° visual angle vertically and 3° horizontally. The color patch was flanked (above and below) by two character strings. The distance between the outer edge of the color patch and the nearest edge of the letter strings was .7°. The color words once again subtended .5° vertically and 3° horizontally. The position of the Stroop string (congruent, incongruent or neutral identity) and the distractor word (threat or nonthreat) was randomized across trials (see Fig. 1, Panel C). 4.1.4. Apparatus Same as Experiments 1 and 2. 4.1.5. Procedure Same as Experiments 1 and 2. 4.2. Results Two items “blood” and “reject” were excluded form the present analyses because they yielded Stroop effects that were greater than 3.5 standard deviations away from the mean in their cell. The data were analyzed in the same manner as Experiments 1 and 2. 4.2.1. Response time Median RTs were analyzed. Trials on which a voice-key failure occurred (2.5%) or an error was made (4.2%) were excluded from the RT data. The mean of median RTs can be seen in Table 4. Overall, a robust Stroop effect was observed with responses taking approximately 49 ms longer on incongruent trials than on congruent trials, Fs(1,

M.G. Reynolds, R.M. Langerak / Acta Psychologica 159 (2015) 108–115

17) = 90.991, p b .001, MSE = 460.276, η2 = .843; Fi(1, 188) = 79.094, p b .001, MSE = .132, η2 = .296. The magnitude of this Stroop effect is comparable to previous work using the Stroop Dilution procedure with vocal responses (e.g., Kahneman & Chajczyk, 1983). There was no main effect of distractor word type (Fs b 1). Critically, if the presence of a threat word reduces the extent to which the color word is attended (e.g., by capturing attention), then the Stroop effect should be smaller in the threat compared to the non-threat condition. Consistent with threat words capturing attention, an interaction was observed, whereby the Stroop effect in the threat condition was 26 ms smaller than the Stroop effect in the non-threat condition, Fs(1, 17) = 6.535, p = .020, MSE = 433.259, η2 = .278; Fi(1, 188) = 6.984, p = .009, MSE = .132, η2 = .036. 4.2.2. Percent error A robust Stroop effect was observed with more errors on incongruent trials than on congruent trials, Fs (1, 17) = 13.424, p = .002, MSE = 9.710, η2 = .441; Fi(1, 188) = 11.444, p = .001, MSE = 53.530, η2 = .057. No other effects approached significance (Fs b 1). Therefore, the error data do not compromise the interpretation of the RT data. 4.2.3. Neutral condition There was no difference between the threat and non-threat conditions on the neutral Stroop trials in either RT (ts b 1) or percent error (ts b 1.2) replicating Algom et al. (2004). 4.3. Discussion Experiment 3 assessed whether threat words would capture attention when the task demands do not require attending to a word (Cho et al., 2012; Kahneman & Chajczyk, 1983; Yee & Hunt, 1991). If threat words can capture attention and use the same processing resources as non-threat words, they should capture attention when the task relevant stimulus (here the color patch) does not require the same resources. Consistent with threat words capturing attention, emotional Stroop Dilution was observed, where the Stroop effect was smaller when the distractor was a threat word compared to when it was a non-threat word. Indeed, not only was the Stroop effect smaller, but both the facilitation and interference effects were reduced in the presence of threat words compared to non-threat words, consistent with what would be expected from threat words capturing attention away from the color words. Evidence of attentional capture in the present experiment where the colored stimulus was a rectangle, but not in Experiment 2 where the color carrier was a word provides converging evidence that that absence of Stroop Dilution in Experiment 2 was due to threat words having the same processing demands as non-threat words and therefore not being processed automatically. Therefore, although the present experiment provides support for the claim that threat words can capture attention, the outcome of Experiments 2 and 3 suggests that attentional capture, in this instance, cannot be considered automatic (Algom et al., 2004; Frings & Wühr, 2012; McKenna & Sharma, 2004). Given that the attentional capture observed in Experiment 3 is not automatic, it is unclear how the threat words capture attention. Here, we hypothesize that threat words gained priority access to the limited capacity system more often than the neutral words because they were processed more quickly than the neutral words. However, given that the stimuli were matched on overall naming time using data from the Lexicon project (Balota et al., 2007), speed of processing must have been influenced by local contextual factors. One, factor that likely influenced how quickly the words were processed was semantic priming (e.g., Neely, 1977). Semantic priming refers to faster performance for a target word (e.g., DOCTOR) following a semantically related word (e.g., NURSE), than a non-related word (e.g., TABLE). Evidence suggests that semantic priming can influence early visual processing of words when the proportion of semantically related trials is high (.50), but

113

not when it is low (.25; Stolz & Neely, 1995). In the present experiments, the threat words were semantically related, whereas the nonthreat words were not semantically related. This created a relatedness proportion of .50. It is therefore possible that semantic priming could have lead to an early processing advantage for the threat words relative to the non-threat words. This processing advantage could have increased the likelihood that the threat words would access the limited capacity system compared to the non-threat words.

5. General discussion The present findings add to recent evidence that the consequences of threatening stimuli are complex. Presently, we believe that there is compelling evidence for at least three separately identifiable cognitive processes being affected by threatening stimuli in the Emotional Stroop Task: (1) cognitive slowing (Algom et al., 2004; McKenna & Sharma, 2004), and two processes involved in selective attention, namely (2) filtering (as indexed by Garner Interference; Mama et al., 2012), and (3) attentional capture. This suggests that variations of the traditional Emotional Stroop task actually need to be conceptualized as separate tasks that can be used to examine these different processes. These processes are discussed below. One process indexed using the traditional Emotional Stroop task is cognitive slowing. Cognitive slowing is thought to arise from a special purpose mechanism that interrupts ongoing processing in the presence of threat information (Algom et al., 2004; McKenna & Sharma, 2004; Öhman et al., 2001). Evidence for cognitive slowing is typically observed when the stimuli consist of only a color and a distractor word, with threat and non-threat conditions blocked. However, the observation that performance in this context may also be affected by differences in how irrelevant information is filtered (see below) suggests that the best approach for observing the unique effects of cognitive slowing is to examine performance in a mixed list context. In a mixed list context, cognitive slowing from a threat stimulus is observed as slowed responding on subsequent neutral trials (Frings et al., 2010; McKenna & Sharma, 2004).2 A second process affected by threat stimuli in the Emotional Stroop task is filtering (Mama et al., 2012). Evidence suggests that when threat and non-threat words are presented in separate blocks, task irrelevant information in an attended object (e.g., the meaning of the word) can be more difficult to ignore when it is threatening than when it is nonthreatening (e.g., Frings & Wühr, 2012; Mama et al., 2012). Difficulty in filtering irrelevant information is often assessed by examining Garner Interference, which refers to a slowdown in performance in the face of variability on an irrelevant stimulus dimension. Mama et al. (2012) reported a series of Garner Interference experiments in which they found an Emotional Stroop effect when the assignment of color to words was random so that irrelevant variation in the words needed to be filtered but did not find an Emotional Stroop effect when the 2 We conducted the analyses suggested by Frings et al. (2010) to assess whether there were fast and/or slow effects of the threat words in the present experiments. A fast effect is one that arises from the valence of the current trial and is calculated by comparing threat and non-threat conditions on trial n when the distractor stimulus on trial n-1 was nonthreat. A slow effect is one that arises from the valence of a previous trial and is calculated by comparing performance for non-threat stimuli on trial n when preceded on trial n-1 either by a non-threat stimulus or a threat stimulus. Frings et al. (2010) reported both a fast and a slow effect when the stimuli were randomly intermixed. There was no evidence of these phenomena in Experiment 1 (Fs b 1.09). Analyses of Experiments 2 and 3 yielded similar results. In Experiment 2 there was no evidence of either a fast (F(1, 17) = 1.127, MSE = 1764.996, p = .303, eta = .062) or slow effect (F b 1). Similarly in Experiment 3 there was no evidence of either a fast (F(1, 17) = 2.350, MSE = 925.570, p = .144, eta = .121) or a slow effect (F(1, 17) = 3.201, MSE = 811.570, p = .091, eta = .158). At present it is unclear why we were unable to replicate Frings et al.'s observation of fast and slow effects given the vast number of differences between our experimental methodologies. However, it should be noted that Chajut, Schupak et al. (2010) also failed to find evidence of cognitive slowing (a slow effect) in their study, which also used an emotional Stroop Dilution procedure.

114

M.G. Reynolds, R.M. Langerak / Acta Psychologica 159 (2015) 108–115

assignment of color to words was systematic thereby reducing filtering demands. Here, we hypothesize that difficulty filtering threat words compared to non threat words (as opposed to attentional capture) can explain the emotional Stroop Dilution reported by Chajut, Schupak et al. (2010) when the distractor words were the color carriers and the threat and non-threat words were presented in separate blocks. Top-down control of the attentional filter also explains why Stroop Dilution was not observed in Experiment 1. Any change to the filter arising from the presence of threat words would have been applied to both the threat and non-threat distractors. Finally, the third process is attentional capture, which we distinguish from filtering by conceptualizing attentional capture as a movement of selective attention towards a threat object. This process of selective attention is demonstrated in Experiment 3. Evidence for attentional capture in our emotional Stroop task was demonstrated using a modified version of Kahneman and Chajczyk's (1983) Stroop Dilution procedure in which the target color is presented separately from both the color words and the distractor words (threat and non-threat words). Critically, the location of the target color is known (e.g., always presented at fixation in Experiment 3) and the locations of the color and distractor words are random (above/below fixation). This provides a context in which attentional capture by the threat stimulus is possible. By our account, if the location of either word was attended, then there would no opportunity for attentional capture to occur due to the capacity limited nature of the word recognition system. Furthermore, randomly intermixing the threat and non-threat stimuli in a single block of trials is essential in order to avoid other factors such as filtering and cognitive slowing from influencing task performance. Here, we hypothesize that the threat words may have gained priority access to the limited capacity processor because the threat words, but not the nonthreat were semantically related. This may have lead to an early processing advantage for the threat words, which increased the likelihood that they would gain access to the limited capacity bottleneck before the color-words.

populations. However, assessing whether the findings from Experiments 1 and 2 generalize to special populations could address the processing requirements of threat words. For instance, the absence of attentional capture by threat words in Experiment 2 when the color word was attended suggests that threat words are subject to the same processing limitations as non-threat words in a non-clinical population. Therefore, if emotional Stroop Dilution were observed under these conditions for a specialized population it would provide evidence that in this population, threat words bypass the limited capacity system that is required by non-threat words.

5.1. Attentional requirements of threatening words

References

It has been hypothesized that threatening stimuli are processed automatically by a special threat detection system (e.g., Öhman et al., 2001). Although the results of Experiment 3 are consistent with threat words capturing attention, the results of Experiment 2 suggest that this process is not automatic and not due to a special threat detection mechanism (see also van den Hout, de Jong, & Kindt, 2000). As such it is unlikely that capture is due to automatic amygdala activation as suggested by Öhman et al., 2001. Instead, the observation that threat words captured attention can be accounted for by appealing to processes known to affect visual word recognition, in particular that they require access to a limited capacity mechanism (see Harris & Pashler, 2004; Janczyk, Augst, & Kunde, 2014). Whether this extends to other stimuli (e.g., pictures) remains to be seen. For instance, evidence suggests that negative faces are able to capture attention even when presented among neutral face distractors (Reynolds et al., 2009). 5.2. Special populations The most widespread use of the Emotional Stroop task has been with specialized populations (e.g., those with post-traumatic stress disorder, Williams et al., 1996). In contrast, the present experiments were conducted with a random sample of undergraduate students. Some critical differences between these populations may be how strongly threat stimuli affect performance and also the extent to which threatening stimuli can bypass processing limitations (Williams et al., 1996). Evidence that specialized populations show more emotional Stroop Dilution than controls using the methods from Experiment 3 would be evidence for greater attentional capture by threat words for these

5.3. Conclusion Findings from the traditional Emotional Stroop task have been important in the construction of cognitive models of psychopathology. In particular, the Emotional Stroop effect has been used to support the claim that threatening information captures attention. Recent evidence that the Emotional Stroop task arises from cognitive slowing (Algom et al., 2004; McKenna & Sharma, 2004; Frings et al., 2010) and filtering (Mama et al., 2012) is problematic for these models. However, the outcome of Experiment 3 suggests that threat words can capture attention, but only when task demands do not require attending to a stimulus that requires the same processing resources (Experiments 1 & 2). The present findings also suggest that subtle modifications to the Emotional Stroop task can be used to isolate three cognitive processes involving threat words: (1) cognitive slowing (Algom et al., 2004; Frings et al., 2010; McKenna & Sharma, 2004), as well as two selective attention processes, (2) attentional capture (Experiment 3) and (3) difficulty filtering (Mama et al., 2012). An important goal for future research will be to examine the extent to which these processes are interconnected. Acknowledgments This work was supported by grants 341586 to MGR from the Natural Sciences and Engineering Research Council of Canada.

Algom, D., Chajut, E., & Lev, S. (2004). A rational look at the emotional Stroop phenomenon: A generic slowdown, not a Stroop effect. Journal of Experimental Psychology, 133, 323–338. http://dx.doi.org/10.1037/0096-3445.133.3.323. Atkinson, L., Leung, E., Goldberg, S., Benoit, D., Poulton, L., Myhal, N., et al. (2009). Attachment and selective attention: Disorganization and emotional Stroop reaction time. Development and Psychopathology, 21, 99–126. http://dx.doi.org/10. 1017/S0954579409000078. Baayen, R.H. (2008). Analyzing linguistic data: A practical introduction to statistics using R. Cambridge: Cambridge University Press. Balota, D.A., Yap, M.J., Cortese, M.J., Hutchison, K.A., Kessler, B., Loftis, B., et al. (2007). The English Lexicon Project. Behavior Research Methods, 39, 445–459. Barr, D.J., Levy, R., Scheepers, C., & Tily, H.J. (2013). Random effects structure for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language, 68, 255–278. Brown, T.L., Gore, C.L., & Carr, T.H. (2002). Visual attention and word recognition in Stroop color-naming: Is word recognition “automatic”? Journal of Experimental Psychology: General, 131, 220–240. http://dx.doi.org/10.1037//0096-3445.131.2.22. Brown, T.L., Roos-Gilbert, L., & Carr, T.H. (1995). Automaticity and word perception: Evidence from Stroop and Stroop dilution effects. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 1395–1411. Chajut, E., Schupak, A., & Algom, D. (2010). Emotional Dilution of the Stroop effect: A new tool for assessing attention under emotion. Emotion, 10, 944–948. http://dx.doi.org/ 10.1037/a0020172. Chajut, E., Mama, Y., Levy, L., & Algom, D. (2010). Avoiding the approach trap: A response bias theory of the emotional Stroop effect. Journal of Experimental Psychology: Learning, Memory, and Cognition, 36, 1567–1572. Cho, Y.S., Choi, J.M., & Proctor, R.W. (2012). Likelihood of attending to the color word modulates Stroop interference. Attention, Perception, & Psychophysics, 74, 416–429. http://dx. doi.org/10.3758/s13414-011-0250-3. Cho, Y.S., Lien, M., & Proctor, R.W. (2006). Stroop dilution depends on the nature of the color carrier but not on its location. Journal of Experimental Psychology: Human Perception and Performance, 32, 826–839. http://dx.doi.org/10.1037/0096-1523.32.4. 826. Cisler, J.M., Wolitzky-Taylor, K.B., Adams, T.G., Babson, K.A., Badour, C.L., & Willems, J.L. (2011). The emotional Stroop task and posttraumatic stress disorder: A metaanalysis. Clinical Psychology Review, 31, 817–828. http://dx.doi.org/10.1016/j.cpr. 2011.03.007.

M.G. Reynolds, R.M. Langerak / Acta Psychologica 159 (2015) 108–115 Derryberry, D., & Reed, M.A. (2002). Anxiety-related attentional biases and their regulation by attentional control. Journal of Abnormal Psychology, 111, 225–235. http://dx. doi.org/10.1037//0021-843X.111.2.225. Francolini, C. M., & Egeth, H. (1980). On the automaticity of “automatic” activation: Evidence of selective seeing. Perception & Psychophysics, 27, 331–342. Frings, C., Englert, J., Wentura, D., & Bermeitinger, C. (2010). Decomposing the emotional Stroo effect. Quarterly Journal of Experimental Psychology, 63, 42–49. Frings, C., & Wühr, P. (2012). Don't be afraid of irrelevant words: The emotional Stroop effect is confide to attended words. Cognition & Emotion, 26, 1056–1068. Forster, K. I., & Dickinson, R. G. (1976). More on the language-as-fixed-effect fallacy: Monte Carlo estimates of error rates for F 1, F 2, F′, and min F′. Journal of Verbal Learning and Verbal Behavior, 15, 135–142. Harris, C.R., & Pashler, H. (2004). Attention and the processing of emotional words and names. Psychological Science, 15, 171–178. Holle, C., Neely, J.H., & Heimberg, R.G. (1998). The effects of blocked versus random presentation and semantic relatedness of stimulus words on response to a modified Stroop task among social phobics. Cognitive Therapy and Research, 21, 681–697. James, W. (1890). The Principles of Psychology, Vol. 1, New York: Holt. Janczyk, M., Augst, S., & Kunde, W. (2014). The locus of the emotional Stroop effect: A study with the PRP paradigm. Acta Psychologica, 151, 8–15. Kahneman, D., & Chajczyk, D. (1983). Tests of the automaticity of reading: Dilution of Stroop effects by color-irrelevant stimuli. Journal of Experimental Psychology: Human Perception and Performance, 9, 497–509. Kahneman, D., & Henik, A. (1981). Perceptual organization and attention. In M. Kubovy, & J.R. Pomerantz (Eds.), Perceptual Organization. Hillsdale, N.J.: Erlbaum. Kucera, H., & Francis, W. (1967). Computational analysis of present day American English. Providence, RI: Brown University Press. Larsen, R.J., Mercer, K.A., & Balota, D.A. (2006). Lexical characteristics of words used in emotional Stroop experiments. Emotion, 6, 62–72. http://dx.doi.org/10.1037/15283542.6.1.62. Lund, K., & Burgess, C. (1996). Producing high-dimensional semantic spaces from lexical co-occurrences. Behavior Research Methods, Instruments, & Computers, 28, 203–208. MacLeod, C. (1991). Half a century of research on the Stroop effect: An integrative review. Psychological Bulletin, 109, 163–203. MacLeod, C., Andrew, M., & Tata, P. (1986). Attentional bias in emotional disorders. Journal of Abnormal Psychology, 95, 15–20. Mama, Y., Behn-Haim, M.S., & Algom, D. (2012). When emotion does and does not impair performance: A Garner theory of the emotional Stroop effect. Cognition and Emotion, 12, 1–14. http://dx.doi.org/10.1080/02699931.2012.726212. McCann, R.S., Remington, R.W., & Van Selst, M. (2000). A dual-task investigation of automaticity in visual word processing. Journal of Experimental Psychology: Human Perception and Performance, 26, 1352 (1270). McKenna, F.P., & Sharma, D. (1995). Intrusive cognitions: An investigation of the emotional Stroop task. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 1595–1607. McKenna, F.P., & Sharma, D. (2004). Reversing the emotional Stroop effect reveals that it is not what it seems: The role of fast and slow components. Journal of Experimental Psychology: Learning, Memory, and Cognition, 30, 382–392. http://dx.doi.org/10. 1037/0278-7393.30.2.382.

115

Mitterer, H., La Heij, W., & Van der Heijden, A.H.C. (2003). Stroop dilution but not wordprocessing dilution: Evidence for attention capture. Psychological Research, 67, 30–42. http://dx.doi.org/10.1007/s00426-002-0108-3. Mitterschiffthaler, M.T., Williams, S.C.R., Walsh, N.D., Cleare, A.J., Donaldson, C., Scott, J., et al. (2008). Neural basis of the emotional Stroop interference effect in major depression. Psychological Medicine, 38, 247–256. http://dx.doi.org/10.1017/ S0033291707001523. Neely, J. H. (1977). Semantic priming and retrieval from lexical memory: Roles of inhibitionless spreading activation and limited-capacity attention. Journal of Experimental Psychology: General, 106, 226–254. Öhman, A., Flykt, A., & Esteves, F. (2001). Emotion drives attention: Detecting the snake in the grass. Journal of Experimental Psychology: General, 130, 466–478. http://dx.doi.org/ 10.1037/AXJ96-3445.130.3.466. Raaijmakers, J. G. W., Schrijnemakers, J. M. C., & Gremmen, F. (1999). How to deal with “the language-as-fixed-effect fallacy”: Common misconceptions and alternative solutions. Journal of Memory and Language, 41, 416–426. Reynolds, M., & Besner, D. (2006). Reading aloud is not automatic: Processing capacity is required to generate a phonological code from print. Journal of Experimental Psychology: Human Perception and Performance, 32, 1303–1323. http://dx.doi.org/10. 1037/0096-1523.32.6.1303. Reynolds, M.G., Eastwood, J.D., Partanen, M., Frischen, A., & Smilek, D. (2009). Monitoring eve movements while searching for affective faces. Visual Cognition, 17, 318–333. Richards, A., French, C.C., Johnson, W., Naparstek, J., & Williams, J. (1992). Effects of mood manipulations and anxiety on performance of an emotional Stroop task. British Journal of Psychology, 83, 479–491. Roberts, M., & Besner, D. (2005). Stroop dilution revisited: Evidence for domain-specific, limited capacity processing. Journal of Experimental Psychology: Human Perception and Performance, 31, 3–13. http://dx.doi.org/10.1037/0096-1523.31.1.3. Schneider, W., Eschman, A., & Zuccolotto, A. (2002). E-Prime 2.0 user's guide. Pittsburgh: Psychology Software Tools Inc. Stolz, J. A., & Neely, J. H. (1995). When target degradation does and does not enhance semantic context effects in word recognition. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21(3), 596–611. van den Hout, M.A., de Jong, P., & Kindt, M. (2000). Masked fear words produce increased SCRs: An anomaly for Öhman's theory of pre-attentive processing in anxiety. Psychophysiology, 37, 283–288. Williams, J.M.G., Mathews, A., & MacLeod, C. (1996). The emotional Stroop task and psychopathology. Psychological Bulletin, 120, 3–24. Williams, J.M.G., Watts, F.N., MacLeod, C., & Mathews, A. (1997). Cognitive psychology and emotional disorders (2nd Edition ). Toronto: John Wiley & Sons. Wingenfeld, K., Mensebach, C., Rullkoetter, N., Schlosser, N., Schaffrath, C., Woermann, F.G., et al. (2009). Attentional bias to personally relevant words in borderline personality disorder is strongly related to comorbid posttraumatic stress disorder. Journal of Personality Disorders, 23, 141–155. http://dx.doi.org/10.1016/j.psyneuen.2008.10.024. Yee, P.L., & Hunt, E. (1991). Individual differences in Stroop dilution: Tests of the attention-capture hypothesis. Journal of Experimental Psychology: Human Perception and Performance, 17, 715–725. Yiend, J. (2010). The effects of emotion on attention: A review of attentional processing of emotional information. Cognition and Emotion, 24, 3–47.