Contents lists available at SciVerse ScienceDirect
Body Image j our nal homepage: www. el sevi er . com/ l ocat e/ bodyi mage I cant stop looking at them: Interactive effects of body mass index and weight dissatisfaction on attention towards body shape photographs Xiao Gao a,b,,1 , Xiaojing Li a,b,1 , Xiaoying Yang c , Yang Wang a,b , Todd Jackson a,b , Hong Chen a,b a Faculty of Psychology, Southwest University, Chongqing, China b Key Laboratory of Cognition and Personality, Southwest University, Chongqing, China c School of Pre-school Education, Chongqing Second Normal College, Chongqing, China a r t i c l e i n f o Article history: Received 12 April 2012 Received in revised form 22 December 2012 Accepted 22 December 2012 Keywords: Attentional bias Disengagement Body dissatisfaction BMI Inhibition of return a b s t r a c t Although attentional biases toward body-related information contribute to the etiology and maintenance of body dissatisfaction (BD) and eating disorders (EDs), attentional disengagement in women with BD and EDs is not clear. The present study investigated the association between weight dissatisfaction and attentional disengagement from body-related pictures and the possible moderating effect of body mass index (BMI) on this relation. Two hundred and four undergraduate women engaged in an experiment using a pictorial spatial cueing paradigm including fat/thin bodies and neutral household photos. Partial correlations and simple slopes regression analyses were conducted with attentional disengagement index scores of each category of cues. Findings suggested that independent of BMI, weight dissatisfaction was directly associated with attentional disengagement from both fat and thin pictures. In addition, among women with low and medium BMIs, the more they were dissatised with their bodyweight, the more difculty they had disengaging their attention from fat body pictures. 2013 Elsevier Ltd. All rights reserved. There is an abundant literature on information processing pre- ferences related to body size, body shape and food cues, some of which suggest that attentional biases toward such cues contribute to the etiology and maintenance of body dissatisfaction and eat- ing disorders (EDs; Cash & Strachan, 2002; Lee & Shafran, 2004). According to Vitousek and Hollons (1990) cognitive theory of EDs, attentional biases toward weight-related information arise as a result of underlying maladaptive schemata associated with shape, weight and self. Individuals with maladaptive schemata differ from those without in several ways, including enhanced attention to and enhanced memory for schema-consistent information (e.g., fat stimuli) and resistance to schema-inconsistent information (e.g., thin stimuli). Two approaches have emerged within this liter- ature. The rst examines how body dissatised or ED samples process information about their own body (e.g., Blechert, Ansorge, & Tuschen-Cafer, 2010; Janelle, Hausenblas, Ellis, Coombes, & Duley, 2008; Jansen, Nederkoorn, & Mulkens, 2005; Roefs, Jansen, Moresi, Willems, van Grootel, & van der Borgh, 2008; Smeets, Jansen, & Roefs, 2011; Smeets, Tiggemann, et al., 2011), and the second focuses on how these samples process body-, weight- and
Corresponding author at: Faculty of Psychology, Southwest University, No. 1,
Tiansheng Road, BeiBei, Chongqing 400715, China. Tel.: +86 13627639919; fax: +86 23 68253629. E-mail address: gaoxiaox@swu.edu.cn (X. Gao). 1 They are co-rst authors. shape-related information in their general environment (e.g., Chen & Jackson, 2005; Chen & Jackson, 2006; Gao, Deng, et al., 2011; Glauert, Rhodes, Fink, & Grammer, 2010; Rieger, Schotte, Touyz, Bemont, Grifths, & Russell, 1998; Shafran, Lee, Cooper, Palmer, & Fairburn, 2007). In the rst approach, consistent evidence indicates that body dissatised and ED samples are highly self-critical when looking at their own bodies (Jansen et al., 2005; Roefs et al., 2008; Smeets, Jansen, et al., 2011; Smeets, Tiggemann, et al., 2011). Con- versely, research on the second approach has been less consistent. Therefore, the nature of attentional biases toward external body, weight, and shape stimuli are not well understood. Early studies often used modied Stroop color naming tasks to study attentional bias in ED samples, and most of the results indicated that ED samples were often slower to name the color of disorder-relevant stimuli than neutral cues (Davidson & Wright, 2002; Lee & Shafran, 2004; Perpina, Leonard, Treasure, Bond, & Banos, 1998; Sackville, Schotte, Touyz, Grifths, & Beaumont, 1998). However, there are conceptual ambiguities and method- ological issues when using modied Stroop color naming task; namely, it cannot distinguish different attentional components (Dobson & Dozois, 2004; Lee & Shafran, 2004). Attentional biases corresponding to disturbance-relevant stimuli (e.g., anx- iety related stimuli, depression related stimuli, or ED related stimuli) can include facilitated attention, disengagement difculty, or attentional avoidance (see Cisler & Koster, 2010). Facilitated attention refers to the relative ease or speed with which atten- tion is drawn to a disturbance-relevant stimulus (i.e., attentional 1740-1445/$ see front matter 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.bodyim.2012.12.005 192 X. Gao et al. / Body Image 10 (2013) 191199 orienting). Disengagement difculty refers to the degree to which a disturbance-relevant stimulus captures attention and impairs switching attention away from it to another stimulus. Attentional avoidance refers to the phenomenon in which attention is pref- erentially allocated towards locations opposite the location of the disturbance-relevant cue, thus indicatingavoidanceof this cue(e.g., Cisler & Koster, 2010; Gao, Wang, Jackson, Zhao, Liang, & Chen, 2011; Glauert et al., 2010; Koster, Verschuere, Crombez, & Van Damme, 2005). Recently, researchers haveturnedtotwo maintasks (the dot probe task and odd-one-out visual search task) to dis- tinguish different attentional components to clarify the nature of attentional biases in body dissatised and ED samples. The dot probe task was originally developed by MacLeod, Mathews, and Tata (1986). In this task, a pair of stimuli (e.g., a body-related stimulus and a neutral stimulus) are briey presented on a screen, one above the other or one beside the other, and are immediately followed by a dot (the probe) in the location of one of the stimuli. Participants are required to respond to the probe as quickly as possible. The reaction time to probe would be reduced if participants were attending to the region of the screen. This task is possible to distinguishengagement anddisengagement sub- components of attention, either by manipulating the presentation duration (Mogg, Bradley, Miles, & Dixon, 2004) or by including neutral trials (Koster, Crombez, Verschuere, & De Houwer, 2004). The odd-one-out visual search task was introduced by Hansen and Hansen (1988) and later modied by other researchers. In this task, a group of words/pictures stimuli are simultaneously presented on the screen, and they are often presented in the shape of a 22, 33 or 54 matrices. Participants are asked to indicate whether one of the stimuli is different fromothers on a predened dimen- sion by pressing the corresponding key. For example, the matrix might contain a single body-related target word (e.g., fat) among eight distractor or neutral words (e.g., box, bed, cot) in a 33 matrix. In this case, the participants would have to react to the body-related word. The odd-one-out task can be used to assess attentional engagement by evaluating response latencies in detec- ting disturbance-relevant targets (e.g., body-relatedstimuli) versus neutral targets (e.g., car-related stimuli) among other kind of neu- tral distractors (e.g., furniture-related stimuli). Disengagement can be evaluated by comparing response latencies in detecting a neu- tral target (e.g., car-related stimuli) among disturbance-relevant distractors (e.g., body-related stimuli) versus latencies for detec- ting a neutral target (e.g., car-related stimuli) fromanother neutral stimulus category (e.g., furniture-related stimuli; Rinck, Reinecke, Ellwart, Heuer, &Becker, 2005; Smeets, Jansen, et al., 2011; Smeets, Roefs, van Furth, & Jansen, 2008; Smeets, Tiggemann, et al., 2011). Using an odd-one-out task, Smeets et al. (2008) found that patients with EDs were faster detecting body words in arrays of neutral words thanindividuals without EDs. These ndings suggest attentional engagement towards body information was speeded in patients with EDs. However, because valences of body words were not specied, it was not clear whether the effect reected a bias towards negative, positive, or general body information. Dot probe studies suggest ED and body dissatised samples dis- played facilitated attention towards negative body cues (e.g., fat, obese, or tubby). Rieger et al. (1998) reported that patients with EDs were faster at responding to probes in the location of previously-presented words denoting a large physique and slower responding toprobes inthe locationof previously-presentedwords denoting thinness. Shafran et al. (2007) found that relative to thin body shape pictures, ED samples were faster at responding to probes in the same location as overweight or neutral body shape pictures. Recently, using the dot probe paradigm combined with eye-tracking technology, Gao, Wang, et al. (2011) reported that womenwithbody dissatisfactionshowed initial orienting, speeded detection, and initial maintenance biases towards fat body words (e.g., fat or plump) inadditiontoa speededdetection-avoidance patternof biases inrelationto thinbody words (e.g., slim or slen- der). Together, these ndings suggest that individuals with EDs or body dissatisfaction preferentially process negative body informa- tion and/or resist or avoid thin body information as proposed by Vitousek and Hollon (1990). To our knowledge, fewpublished studies have focused onatten- tional disengagement among womenwithbody dissatisfactionand EDs. The importance of inhibited disengagement from threat in individuals with anxiety has been highlighted in anxiety disorder studies, withmodels of anxiety-basedattentional biases suggesting that inhibited disengagement has a signicant role in the etiology and maintenance of anxiety disorders (e.g., Amir, Elias, Klumpp, & Przeworski, 2003; Fox, Russo, Bowles, & Dutton, 2001; Ouimet, Gawronski, & Dozois, 2009). Given that patterns of attentional bias toward food or body-related information among individuals with body dissatisfaction or EDs are similar to those linked to anxiety disorders (Lee & Shafran, 2004; Rieger et al., 1998), attentional disengagement difculties warrant serious consideration as a cog- nitive vulnerability to exacerbations in body dissatisfaction and ED risk. Recently, two odd-one-out studies assessed attentional bias in ED or body dissatised samples (Smeets et al., 2008; Smeets, Jansen, et al., 2011; Smeets, Tiggemann, et al., 2011). Neither of them showed any indication of attentional distraction by body- related stimuli. However, this paradigmmay not be the best option for investigating attentional disengagement. In the studies from Smeets et al. (2008), Smeets, Jansen, et al. (2011), and Smeets, Tiggemann, et al. (2011), the 54 stimuli matrix was presented on the screen until a response was made or for a maximumof 20s. The mean behavioral response latencies were around 4s, during which engagement, disengagement, and re-engagement of visual attention could occur several times. Toillustrate, Gao, Wang, et al. (2011) assessedeyemovements of women with and without body dissatisfaction within a dot probe task and found that women with body dissatisfaction had com- paratively longer initial xations on fat words and shorter initial xations onthinwords. Theauthors suggestedthat followingatten- tional capture by fat stimuli in very early stage of information processing (i.e., less than one second), women with body dissatis- factionhadmoredifculty, at least initially, disengagingfromthem. In contrast, shorter initial xations on thin words among this group may have reected rapid disengagement of overt attention from these stimuli (Garner, Mogg, & Bradley, 2006). Although tracking of eye movements may illuminate subtle biases in visual attention that arise soon after stimulus presentations, opinions differ when choosing appropriate eye movement indices for evaluating disen- gagement. For example, in a free-viewing task, Giel, Friederich, Teufel, Hautzinger, Enck, and Zipfel (2011) used total gaze duration to evaluate attentional engagement and disengagement and found that patients withanorexianervosademonstratedmoreattentional disengagement to food pictures than did controls, whereas Gao, Wang, et al. (2011) used rst xation duration, nding attentional disengagement difculty fromfat words and disengagement facil- ity from thin words. Moreover, the dot probe paradigm may not be ideal for evaluating disengagement, because the pairs of stimuli are simultaneously presented on the screen and attention could be easily implicitly captured by the unattended one, which could involve covert eye movements such as saccades (for a review, see Simons, 2000). Therefore, it remains an issue as to how the atten- tional disengagement obtained from the dot probe paradigm can be understood or explained. To investigate disengagement, researchers have used a mod- ied spatial cueing paradigm (Posner, 1980). The task is widely used because the direction and disengagement of attention can be observed clearly (e.g., Dai & Feng, 2009; Fox et al., 2001; Fox, X. Gao et al. / Body Image 10 (2013) 191199 193 Russo, &Dutton, 2002; Verkuil, Brosschot, Putman, &Thayer, 2009; Waters, Nitz, Craske, & Johnson, 2007; Yiend & Mathews, 2001). In this task, a visual cue is presented in one of the two locations followed by a target, either at the same location as the target (a valid trial) or at the opposite location (an invalid trial). Participants were required to respond to the target as quickly and accurately as possible. Response latencies of the valid trails are typically shorter than those of invalid trials when the time lag between the onset of the cue and the target is relatively short, which is called cue-target stimulus onset asynchrony (SOA). However, as the SOAgets longer, response latencies of the valid trials often turn longer than those of invalid trials. This phenomenon is called inhibition of return (IOR) and enables individuals to disengage attention away from a searched location to a new location. Modest inhibition of return effects may reect a reduced capacity to disengage attention from cued locations (Fox et al., 2002; Verkuil et al., 2009). Inhibition of return effects can be modulated by cues having emotional valence (Fox et al., 2002). Specically, participants took longer to respond to a target when the cue was angry or happy relative to neutral. Inhibition of return effects for disturbance-relevant stimuli may also be reduced by trait anxiety (Fox et al., 2002; Verkuil et al., 2009; Waters et al., 2007; Yiend & Mathews, 2001) and depression (Dai & Feng, 2009; Karparova, Kersting, & Suslow, 2005). Another issue salient to this research is the impact of body mass index (BMI) on the relation between body dissatisfaction and attentional bias. For example, Roefs et al. (2008) measured eye movements in 51 normal-weight (BMI range: 18.525) female stu- dents who were successively shown pictures of their own body and a control body for 30s each. Attention directed towards ones self-identied least attractive body part and the controls most attractive bodypart was associatedwithhigher BMI andlower level of self-rated attractiveness. Later, using pictorial dot probe task, Glauert et al. (2010) found undergraduate women were faster to discriminatethedirectionof acuethat appearedinthelocationpre- viously occupied by thin rather than fat body gures. This bias was weakest for women who had a higher BMI and elevated body dis- satisfaction. Signicant correlations between attentional bias and either BMI or body dissatisfaction were eliminated if either of these two variables was controlled. However, using 500ms and 150ms cue presenting time, Glauert et al.s (2010) study only assessed attentional engagement, while attentional disengagement can not bemeasuredintheir study, becauseneutral trials werenot included in their dot probe task (Koster et al., 2004). Furthermore, the sta- tistical methods used in their study only allowed for testing the covariate effects of BMI and body dissatisfaction on attentional bias to body-related cues, but other effects could not be analyzed. For example, Glauert et al. (2010) could not determine whether BMI moderated the relationship between body dissatisfaction and attentional bias, and more specically, whether the effects of body dissatisfaction on attentional bias differed between samples with high and low BMI. The present study added to the literature by addressing these questions. Finally, of additional interest for the present study is the atten- tional disengagement course over time. In view of Yiend and Mathews (2001) ndings, attentional disengagement fromthreats was greater for individuals high in anxiety than individuals low in anxiety at 500ms, while group differences in disengagement dis- appeared by 2000ms. Later, using 300ms SOA condition, Fox et al. (2002) found that emotionally-valenced facial expressions (angry or happy) produced delayed disengagement in people high in trait anxiety but not in people low in trait anxiety. However, the group differences diminished on SOA 960ms conditions, such that both groups showed delayed attentional disengagement from angry faces. Theauthors suggestedthat althoughanSOAof 960ms is fairly standard in the IOR literature, using a SOA of 960ms might miss differences in disengagement between groups. For example, if low anxious participants take about 300ms to disengage froman initial threat stimulus while high anxious participants take about 600ms to disengage from the same stimulus, this pattern would be con- sistent with the theoretical hypothesis but would not be reected in any IOR differences with an SOA of 960ms (Fox et al., 2002, p. 370). Therefore, in the current study, SOA was varied (760ms and 1160ms) to allow for a more thorough investigation of atten- tional disengagement over time. The rationale for these changes is as follows: SOA760ms is more powerful than SOA960ms in detec- ting attentional disengagement effects from body-related stimuli if the difference in attentional disengagement magnitude between high and low weight dissatisfaction groups is relatively low or the attentional disengagement difference between two groups occurs relativelyearlyduringstimulus processing. Then, if thedifferencein attentional disengagement magnitude between two groups is rel- atively large or the attentional disengagement difference between two groups occurs relatively late during stimulus processing, SOA 1160ms could detect it. Based on the preceding review, this study used a modied spa- tial cueing paradigmto investigate the association between weight dissatisfaction and attentional disengagement from body-related pictures and the potential moderating effect of BMI on this rela- tionship. Based on Vitousek and Hollons cognitive model (1990) and previous ndings (Gao, Wang, et al., 2011; Glauert et al., 2010), three hypotheses were generated: First, weight dissatisfac- tionwouldbepositivelyassociatedwithdifcultydisengagingfrom pictures denoting fat body shape/parts relative to neutral pictures: participants with greater weight dissatisfaction would be slower than those with less weight dissatisfaction to disengage attention from fat pictures. Second, weight dissatisfaction could be related to facilitated disengagement from pictures denoting thin body shape/parts relative to neutral pictures: participants with greater weight dissatisfaction would be faster than those with less weight dissatisfaction to disengage attention from thin pictures. Finally, it was expected that BMI would moderate associations between weight dissatisfaction and attentional disengagement from body shape-related information. To be exact, these effects of weight dissatisfaction on attentional disengagement would be stronger among participants with higher BMIs than those with lower BMIs. Method Participants The sample included 204 women drawn from undergraduate classes at Southwest University (SWU), Chongqing, China. The women ranged from 18 to 25 years of age (M=20.89, SD=1.44). One hundred and ninety six students were Han nationality, three were Uygur nationality, two were Zang nationality, two were Hui nationality, and one was Yi nationality. BMIs ranged from16.02 to 26.17 (M=19.99, SD=2.08). All were right-handed non-smokers, with no current or previous neurological or psychiatric illness, had normal or corrected-to-normal vision, and had normal color vision as assessed by several basic color tests. Materials Body image measure. The Negative Physical Self Scale (NPS; Chen, Jackson, & Huang, 2006) is a 38-itemmeasure of body image concerns developedinMainlandChinese samples. The measure has subscales that assess specic concerns related to Fatness, Stature, and Facial Appearance as well as a General Appearance Concern dimension. Eachitemis ratedona 5-point scale of agreement. Chen et al. (2006) found that the NPS has a stable factor structure among college women fromChina. 194 X. Gao et al. / Body Image 10 (2013) 191199 In this study, the 11-item Fatness Concern subscale (NPS-F) assessing thoughts, feelings, and behaviors related to fatness and overweight concerns was used to measure weight dissatisfaction. Sample items include When I weigh myself, I feel depressed, I have tried many ways to lose weight, and I think I amfat in oth- ers eyes. Each item was rated between 0 (not at all like me) and 4 (very much like me), with total scores reecting the average of summed responses. The NPS-F yields internally consistent scores (=.88), as well as stable scores over three weeks (r =.89) among female and male middle and high school students and undergrad- uates (Chen, 2006) and nine months (r =.70) among middle school and high school girls (Chen & Jackson, 2007). It has satisfactory convergent and predictive validity (Chen et al., 2006; Chen, Gao, & Jackson, 2007; Chen &Jackson, 2008; Jackson &Chen, 2008) among samples of adolescents andyoung adults. Inthe current sample, the NPS-F had an alpha of .90, its scores ranged from 0.09 to 3.18 on a converted 04 scale (M=1.31, SD=0.79), and its correlation with BMI was r =.45, p<.001. Apparatus and picture stimuli. Stimuli were presented on a 19-in., 85-Hz monitor connected to a Pentium IV 3.2-GHz computer. Participants sat 70cmfromthe monitor screen. The pho- tographic stimuli consisted of 30 fat body pictures, 30 thin body pictures, and 30 neutral household pictures adapted from recent work (Gao, 2010; Gao, Wang, Chen, Wang, &Zhang, 2012). Initially, these images were downloaded from free sites on the Internet. Then, they were cropped and sized into uniformsize. The two sets of body shape pictures featuredthe whole body excluding the head. Fat versus thin pictures of different body parts such as arms, waists, thighs or stomachs were matched for gestures. Also, sets of body shape pictures and household pictures were closely matched for physical properties such as color and visual complexity and pre- tested for valence and arousal by female undergraduates. There was no signicant difference among the three kinds of pictures on valence, F(2, 106) =2.54, p=.087, or arousal, F(2, 106) =1.82, p=.170. Examples of stimuli are presented in Appendix A. The Modied Spatial Cueing Paradigm The modied spatial cueing paradigm procedure was pro- grammed with E-Prime v1.2 software (Schneider, Eschman, & Zuccolotto, 2002). The main trial events in the current modied spatial cueing paradigmare illustrated in Fig. 1. At the beginning of each trial, the xation point and two empty boxes were displayed for 8001000ms. The two boxes were 5cm high and 3cm wide and were displayed 2.25cm to the left and the right of the cen- tral xation point. Shortly thereafter, a body or a household cue was presented for 300ms in either the left or right box (Verkuil et al., 2009). The cue pictures were 3.75cm high and 2.4cm wide. Then the cue was blanked out, and 200ms later the central xation point was presented in bold type for either 100ms (SOA=760ms) or 500ms (SOA=1160ms). The initial xation point was then pre- sented for 160ms. Following this, the target (O or C) appeared in the lower half of the same box as the cues (valid trials) or in the box opposite to the cues (invalid trials) until a response was made or for a 5s maximum. Participants were required to identify the target as quickly and accurately as possible, with left index nger pressing the key C if the target was a C and right index nger pressing the key O if the target was an O. During the task, they were instructed to keep the left index nger on key C and the right index nger on key O. The diameter of the target was about 0.8cm. Each participant completed a total of 24 practice trials and 360 experimental trials. Similar to the procedure used by Fox et al. (2002) andVerkuil et al. (2009), 50%(180trials) of the experimental trials werevalid, and50%trials wereinvalid. Theformal experiment phase consisted of four blocks with 90 trials in each block, and the trials were blocked by SOA condition. Two blocks were SOA 760ms condition and the other two were SOA 1160ms condition. Each block consisted of fat body cues-valid, fat body cues-invalid, thinbody cues-valid, thinbody cues-invalid, neutral cues-validand neutral cues-invalid trials, and each category of trials appeared 15 times inevery block. That is, fat body, thinbody andhouseholdcues appeared 60 times each on valid trials and 60 times each on invalid trials, half of which were on SOA 760ms condition and the other half were on SOA 1160ms condition. Every picture was presented four times during the whole experiment, and each was presented twice on invalid and valid trials on SOA 760ms and 1160ms condi- tion, respectively. The order of presentation of the 90 trials within each block was randomized, and the sequence of four blocks was counterbalanced across participants. Procedure Following ethics approval from SWU, participants were recruitedvia on-campus advertisements. Subsequently, 204under- graduate women engaged in the initial and follow-up phases of the study. They were told the experiment focused on determining spatial locations of stimuli and would last for about 60 minutes. After reading a general overview of the research and signing an informed consent sheet, participants completed the individually- administered self-report measures noted above as well as color vision tests. Participants reported their height and weight, and this information was used to calculate their BMI. There was a 1-month delay between the initial and follow-up phases of the study to attempt to mask the true purpose of the study and reduce demand characteristics during the experiment (Gao, Wang, et al., 2011). Participants were asked to consume their regular meals. To avoid the inuence of caffeine on the participants attention and cognitive performance, they were also required to refrain from drinking caffeinated beverages for 12 hours before the experiment. Individual testing sessions were scheduled between 9:0011:00amand 2:305:30pmin a quiet and soundproof room. Following their session, women were debriefed and paid 20 Yuan as compensation for their time. Data Preparation Trials with errors and response time exceeding two standard deviations of each individuals mean reaction time were excluded fromanalyses (Ratcliff, 1993). The number of errors ranged from0 to 16 times (M=10.27, SD=4.68). Design and Statistical Analyses The statistical analyses followedEllenbogenandSchwartzmans (2009) and Verkuil et al.s (2009) studies. A 322 within-group design assessed effects of Cue (Fatness, Thinness, and Neutral), SOA (760ms vs. 1160ms) and Validity (valid trials vs. invalid trials) on probe reaction time, measured in milliseconds. To inves- tigate whether weight dissatisfaction, BMI, and their interaction were associated with delayed disengagement from body cues, but not from neutral cues, weight dissatisfaction, BMI and their interaction were used as continuous between-subjects variables while Cue, SOA, and Validity were within-subjects variables in a repeated measures analysis of variance (ANOVA). To minimize the number of post hoc tests conducted, only main effects and interactions relevant to the studys hypotheses were examined. Similar to EllenbogenandSchwartzmans (2009) andVerkuil et al.s (2009) studies, when a signicant interaction emerged, disengage- ment index scores for two categories of body-related cues were calculated to further examine the signicant interactions using X. Gao et al. / Body Image 10 (2013) 191199 195 Fig. 1. The modied spatial cueing paradigm(one complete invalid cue condition of thin body cues). partial correlations and simple slopes regression analyses (Koster et al., 2005; Verkuil et al., 2009; Waters et al., 2007). For all within-subjects analyses, p-values were corrected for deviations according to the GreenhouseGeisser correction (Greenhouse & Geisser, 1959). A p-level of .05 was used for determining statistical signicance. Results Inhibition of Return Effects After excluding trials with errors and response times exceed- ing two standard deviations of each individuals mean reaction time, reaction times on the different trials were analyzed. Descrip- tive statistic results of reaction time data are presented in Fig. 2. A 3 (Cue) 2 (SOA) 2 (Validity) repeated measures ANOVA on reaction time data revealed a signicant main effect of Validity, F(1, 203) =14.42, p<.001, with longer reaction times on valid tri- als (M=576.10ms, SD=98.62) than on invalid trials (M=565.42ms, SD=99.53). The main effect of SOA was also signicant, F(1, 203) =6.40, p=.012. Participants had shorter reaction times for the SOA 760ms condition (M=567.49ms, SD=100.09) than the SOA 1160ms condition (M=574.03ms, SD=95.81). These effects were qualied by a signicant SOAValidity interaction, F(1, 203) =19.28, p<.001. Simple effects analyses indicated that the inhibitionof returneffects onthe760ms SOAconditionweresignif- icant (p<.001), while no inhibition of return effects were observed on the 1160ms SOA condition. Other main effects or interactions were not signicant. Effects of Weight Dissatisfaction, BMI, and Their Interaction on Attentional Disengagement A 3 (Cue) 2 (SOA) 2 (Validity) within-subjects repeated measures ANOVA was conducted with BMI, weight dissatis- faction, and their interaction as continuous between-subjects variables. Signicant interactions emerged for Cue BMI, F(2, 400) =3.27, p=.040, Cue BMIWeight Dissatisfaction interaction, F(2, 400) =3.64, p=.028, Cue Validity, F(2, 400) =4.72, p=.010, Cue ValidityWeight Dissatisfaction, F(2, 400) =3.35, p=.036; Cue ValidityBMI, F(2, 400) =4.28, p=.015, Cue ValidityBMIWeight Dissatisfaction interac- tion, F(2, 400) =3.21, p=.042, as well as marginally signicant interactions between SOACue Validity, F(2, 400) =2.94, p=.055, and SOACue ValidityBMI, F(2, 400) =2.90, p=.057. In addition, relevant to the studys predictions, the four-way Cue SOAValidityBMIWeight Dissatisfaction interaction was signicant, F(2, 400) =3.48, p=.032. To further analyze the 4-way interaction, the cue validity factor and valid trials were dropped fromfurther analyses (Ellenbogen & Schwartzman, 2009). Datafor invalidtrials wereconvertedintodis- engagement index scores by subtracting reaction times for invalid trials with neutral picture cues fromthe reaction times for invalid trials with fat or thin picture cues, respectively. A positive score indicates longer reaction time followed by an invalid fat or thin pic- ture cue than a neutral cue, which is consistent with the delayed disengagement from fat or thin-related stimuli. A negative score indicates shorter reaction time followed by an invalid fat or thin picture cue than a neutral cue, which suggests the rapid disengage- ment from fat or thin-related stimuli. As indicated by Ellenbogen and Schwartzman (2009), a further advantage of using index scores is that they control for overall group differences in reaction times. A repeated measures ANOVA on disengagement index scores with Cue type as the within-subject variable and BMI, weight dissatisfaction, and their interaction as continuous between- subjects variables were assessed regarding each SOA condition. To further test the rst and second hypotheses that weight dis- satisfaction would be associated with attentional disengagement from body-related cues, partial correlations between weight dissatisfaction and disengagement index scores of fat and thin picture cues were conducted while controlling for BMI and the BMIweight dissatisfaction interaction when the main effect of weight dissatisfaction or the Cue Weight Dissatisfaction inter- action was signicant. Then, to test the nal hypothesis that BMI would moderate the relationship between weight dissatisfaction and attentional disengagement, partial correlations between the BMIweight dissatisfaction interaction and index scores of each cue type were conducted, while controlling for BMI and weight dissatisfaction, when the main effect of BMIweight dissatisfaction interaction or the Cue BMIWeight Dissatisfaction interaction 196 X. Gao et al. / Body Image 10 (2013) 191199 0.00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 Fat Thin Neutral Fat Thin Neutral Valid trials Invalid trials SOA760ms SOA1160ms M e a n
r e a c t i o n
t i m e
( m s ) Fig. 2. Mean reaction time on each condition within the modied spatial cueing task. was also signicant. If partial correlations betweenthe BMIweight dissatisfaction interaction and index scores of each cue type were signicant, simple slopes analyses were conducted to explore how BMI moderated the relation between weight dissatisfaction and attentional disengagement. SOA760ms condition. For indexscores, therepeatedmeasures ANOVA indicated signicant main effects of weight dissatisfaction, F(1, 200) =36.31, p<.001, andthe BMIweight dissatisfactioninter- action, F(1, 200) =8.82, p=.003. Other main effects or interactions were not signicant. Since the main effect of weight dissatisfaction was signi- cant, partial correlations betweenthefat/thindisengagement index scores and weight dissatisfaction were conducted respectively while controlling for BMI and the BMIweight dissatisfaction inter- action, which was aimed to test Hypotheses 1 and Hypothesis 2. Signicant associations between weight dissatisfaction and fat disengagement index scores, r =.28, p<.001, as well as weight dis- satisfaction and thin disengagement index scores, r =.28, p<.001, were observed. These ndings showed independent effects of weight dissatisfaction on attentional disengagement fromboth fat and thin pictures that the more women were dissatised with their body weight, the more difcult it was for them to disengage their attention fromboth fat and thin body pictures. Therefore, the nd- ings on the SOA 760ms condition were consistent with Hypothesis 1, but contrary to Hypothesis 2. To test Hypothesis 3, the signicant main effect of BMIweight dissatisfaction interaction found in the repeated measures ANOVA was further examined by inspecting partial correlations between disengagement index scores and the BMIweight dissatisfaction interaction while controlling for BMI and weight dissatisfac- tion. Partial correlation analyses revealed signicant associations between the BMIweight dissatisfaction interaction and fat disen- gagement index scores, r =.16, p=.025, as well as a marginally signicant correlation between BMIweight dissatisfaction inter- action and thin disengagement index scores, r =.13, p=.066. Next, simple slopes regression analyses on the fat and thin disengage- ment indexscores wereconductedrespectively, usingstandardized variables. The simple slopes regression analyses on the fat dis- engagement index scores conrmed that coefcients for weight dissatisfaction(=.25, p=.001) andthe BMIweight dissatisfaction interaction (=.18, p=.007) were signicant. Signicance tests on the separate regression slopes showed that the slopes of the low (=.43, p<.001) and middle (=.25, p=.001) BMI line were signicant (see Fig. 3). The simple slopes regression analyses on the thin disengagement index scores showed that only the main -1.2 -0.6 0 0.6 1.2 BMI: -1 SD BMI: Mean BMI: +1 SD NPS-F: -1 SD NPS-F: +1 SD (Z scores) D i s e n g a g e m e n t
i n d e x
s c o r e s
f o r
f a t
c u e s Fig. 3. Regressionlines onSOA760ms for the relationship betweenweight dissatis- faction and the disengagement index scores for fat picture cues, moderated by body mass index (BMI). The y-axis was presented in Z scores of disengagement index scores for fat picture cues. Mean=6.44ms, SD=86.39ms; NPS-F =Negative Physical Self-Fatness Concern subscale. effect of weight dissatisfaction (=.30, p<.001) was signicant. Because the main effect of the BMIweight dissatisfaction interac- tiononthindisengagement indexscores was not signicant, further analysis was not conducted. Therefore, Hypothesis 3 was partly supportedonthe SOA760ms conditioninthat BMI only moderated the relationship between weight dissatisfaction and attentional disengagement from fat body pictures but not thin body pictures, and the effects of weight dissatisfaction on attentional disengage- ment from fat body pictures were only signicant for individuals at lowor mediumBMI levels. SOA 1160ms condition. A signicant Cue Weight Dissatis- faction interaction, F(1, 200) =4.61, p=.033, Cue BMI interaction, F(1, 200) =4.77, p=.031 and Cue BMIWeight Dissatisfaction interaction, F(1, 200) =4.47, p=.036, were obtained. To further analyze the Cue Weight Dissatisfaction interaction, partial correlations betweenweight dissatisfactionandfat/thindis- engagement index scores were conducted respectively, controlling for BMI and the BMIweight dissatisfaction interaction. However, no signicant correlation was obtained. The signicant Cue BMIWeight Dissatisfaction interaction was further examined by conducting partial correlations between disengagement index scores and the BMIweight dissatisfaction interaction while controlling for BMI and weight dissatisfaction. X. Gao et al. / Body Image 10 (2013) 191199 197 A marginally signicant correlation between thin disengagement index scores and the BMIweight dissatisfaction interaction was obtained, r =.12, p=.090. Other correlations were not signicant. The simple slopes regression analyses on the thin disengagement index scores showed that no signicant main effect of weight dissatisfaction or the BMIweight dissatisfaction interaction was observed. Therefore, none of the three hypotheses was supported for the SOA 1160ms condition. Discussion This is therst experimental studytoexamineattentional disen- gagement fromimages of fat and thin body shapes in a non-clinical sample that varied in degree of weight dissatisfaction. First, results demonstrated that independent of BMI, weight dissatisfaction was associated with delayed attentional disengagement from both fat and thin body pictures only within the SOA 760ms condition. Second, BMI only moderated the relation between weight dissat- isfaction and attentional disengagement from fat body pictures within the SOA 760ms condition. Specically, weight dissatisfac- tion was associated with delayed attentional disengagement from fat body pictures only among individuals at a lowor mediumlevel of BMI. The current results suggest a positive relationship between weight dissatisfactionanddelayedattentional disengagement from general body information. Attentional disengagement difculties among women with elevations in weight dissatisfaction were con- sistent with eye movement research indicating that women high in weight dissatisfaction had longer initial xations on fat body words (Gao, Wang, et al., 2011) and fat body images (Gao et al., 2012) than did controls as well as ndings indicating anxiety is associated with difculty disengaging attention from spatial loca- tions of threat cues (Fox et al., 2001; Yiend & Mathews, 2001). Fat body pictures may serve as ego threats to women who feel dis- contented with their own body weight and shape, and assimilation effects might be induced between fat body parts described in the pictures and their own bodies. Perhaps such processing involves self-reection or rumination on perceived unattractive parts of their own bodies. Assessments of salient variables such as anxi- ety, downwardsocial comparison, andself-reectioninducedbyfat body images are needed in future work to test this interpretation. Contrary to predictions, weight dissatisfaction was also inde- pendently associated with delayed attentional disengagement from thin body pictures within the SOA 760ms condition. These results suggest womenwithhighlevels of weight dissatisfaction, in particular, had difculty disengaging from pictures denoting thin body shape relative to images of neutral household images. The mechanismunderlying this effect might differ fromthe disengage- ment bias related to fat body images. Women with high weight dissatisfaction may engage in an automatic upward appearance social comparison process whereby they compare their bodies to thin bodies (Posavac, Posavac, & Posavac, 1998; Schutz, Paxton, & Wertheim, 2002). Women high in weight dissatisfaction also focus attention more heavily on the attractive body parts of the compar- ison targets (Jansen et al., 2005; Roefs et al., 2008). As indicated by Friederich et al. (2007), anxiety ratings in response to thin body images among women average in BMI and who do not have an eat- ing disorder are positively correlated with brain activation related to anxiety processing, fear processing, emotional regulation, self- reection, self-judgment, and cognitive conict monitoring. On this basis, similar brain mechanisms may underlie inhibited dis- engagement from thin body images in the current study among women with elevated weight dissatisfaction. The reduced capacity to disengage from thinness cues may aggravate body dissatisfac- tionandcontribute toother negative consequences suchas anxiety, depressed mood, and even increases in disordered eating behav- iors. Regarding the moderating inuence of BMI on association between weight dissatisfaction and attentional disengagement, effects were observed in the SOA760ms condition only for fat body pictures. Notably, weight dissatisfaction was positively associated withsloweddisengagement fromfat body pictures only whenBMIs were at a low or mediumlevel. Conversely, no signicant associa- tionbetweenweight dissatisfactionandattentional disengagement from fat body pictures was observed when BMI was high. Previ- ously, Roefs et al. (2008) found a positive association between BMI and the amount of attention focusing on ones own perceived most unattractive body part and anothers perceived most attractive body part. In Glauert et al.s study (2010), a signicant correla- tion between attentional engagement and body dissatisfaction was eliminated when BMI was controlled. Together, these studies sug- gest that BMI has an effect on attention towards body information. The nonsignicant ndings of women high in BMI in the current study may result from a ceiling effect. Specically, women high in BMI might show inhibited attentional disengagement from fat body images, independent of self-reported weight dissatisfaction. Because societyadvocates thinbodies as ideal andstigmatizes large bodies, women high in BMI would likely internalize societal ideals and have more negative implicit attitudes toward their own body weight and shape, regardless of the extent they explicitly report weight dissatisfaction. Hence, women with a higher BMI may be more sensitive to body-relevant stimuli in a similar manner to women with negative body image. With respect to the time course of attentional disengagement, weight dissatisfaction was associated with slowed disengagement from both fat and thin body pictures, but only in the SOA 760ms condition. In other words, women high in weight dissatisfaction were less able to disengage from body images during a shorter interval between the cue and the target but could shift attention and focus on locations opposite to body cues as quickly as their healthier counterparts did during a relatively longer interval. Dis- engagement difcultyamongwomen highinweight dissatisfaction during the shorter intervals may be based on automatic schema related processing (Vitousek & Hollon, 1990), while later disen- gagement from fat and thin body images may help them escape awareness of their perceived unattractive bodies or from the dis- comfort or anxiety induced by the images. The current ndings contrast with those provided by Glauert et al. (2010), who found that undergraduate women showed an attentional bias toward thin bodies, which was weakest for women who had a higher BMI and elevated body dissatisfaction. However, methodological differences betweenthe twostudies may have con- tributed to the discrepant ndings. Specically, while a spatial cueing paradigmwas utilized to assess attentional disengagement in the present study, a modied dot probe task was used by Glauert et al. (2010) to investigate attentional engagement. Attentional disengagement and engagement are two subcomponents of atten- tion that do not always complement one another. Second, images of actual human bodies were stimuli in this study compared to computer-generated nude female bodies having emaciated BMIs of 11.7 and 15.1 in that study. Skeletally-thin nude bodies may arouse negative emotion for many women and override effects of weight dissatisfaction that may be observed in more realistic depictions of thinness. Findings from the present study were relatively novel and require replication in nonclinical high-risk samples (e.g., females with body dissatisfaction) and clinical ED samples. Nonetheless, this study has potential implications for research questions perti- nent to intervention. For example, since weight dissatisfaction was related to delayed attentional disengagement from body stimuli, will interventions that reduce absorption in body-related cues 198 X. Gao et al. / Body Image 10 (2013) 191199 Fig. A1. reduce body dissatisfaction and, possibly, promote body esteem among college women? Furthermore, will degree of changes in dis- engagement inhibition following treatment serve as a measure of successful outcome and/or a marker for post-treatment prognosis and relapse risk? Although this research provided potential insights regarding links between weight dissatisfaction, BMI, and attentional disen- gagement related to body-related environmental cues, its main limitations warrant mention to guide future work. First, because generalizations cannot be made to clinical ED samples, future research should test the hypothesis that disengagement difcul- ties are even more entrenched among women with EDs. On a related note, ndings do not necessarily generalize to groups that are demographically dissimilar vis--vis gender, age, and culture. Second, reaction time is a relatively crude index of attentional disengagement that fails to elucidate visual attention during stim- ulus presentations. In order to measure disengagement with more precision, use of a spatial cueing paradigm combined with eye movement indices such as saccade latency could be adopted in future research (Hooge & Frens, 2000). Furthermore, participants weight dissatisfactioncouldhave beenalteredduring the relatively long time lag between the assessment of weight dissatisfaction and experimental task, due to body image state uctuations (Melnyk, Cash, & Janda, 2004). In order to reduce body image variability, future studies should include a shorter time lag. In conclusion, this study is the rst that uses a pictorial spa- tial cueing paradigm which presents fat and thin body image cues in order to study whether an association exists between weight dissatisfactionanddelayedattentional disengagement from these images in a nonclinical sample. Specically, undergradu- ate women who experienced high levels of dissatisfaction with their body weight were slower to disengage from both fat and thin body images in a relatively early processing stage. As well, BMI moderated the association between weight dissatisfaction and attentional disengagement. Specically, whenparticipants hadlow or medium levels of BMI, the more they were dissatised with their body weight, the more difcult it was for them to disen- gage their attention from fat body pictures. Overall, difculties in disengaging frombody-related environmental stimuli may be one cognitive bias that co-occurs with weight dissatisfaction among young women. Acknowledgements This research was supported by The Chinese National Natural Science Foundation (31100758) grant to Xiao Gao. We thank two anonymous reviewers who extensively commented on an earlier version of this article. Special gratitude goes to Dr. Tracy L. Tylka for her enthusiasmand thoughtful comments to improve this article. Appendix A. See Fig. A1. References Amir, N., Elias, J., & Klumpp, H. (2003). Attentional bias to threat in social phobia: Facilitated processing of threat or difculty disengaging attention fromthreat? Behaviour Research and Therapy, 41, 13251335. Blechert, J., Ansorge, J., & Tuschen-Cafer, B. (2010). A body-related dot-probe task reveals distinct attentional patterns for bulimia nervosa and anorexia nervosa. Journal of Abnormal Psychology, 119, 575585. Cash, T. F., & Strachan, M. D. (2002). Cognitive behavioral approaches to changing body image. In T. F. Cash & T. Pruzinsky (Eds.), Body image: A handbook of theory, research, and clinical practice (pp. 478486). NY: Guilford Press. Chen, H. (2006). The adolescent physical self: Theoryandresearch. Beijing, China: Xihua Press. Chen, H., Gao, X., & Jackson, T. (2007). Predictive models for understanding body dissatisfaction among young males and females in China. Behaviour Research and Therapy, 45, 13451356. Chen, H., & Jackson, T. (2005). Are cognitive biases associated with body image concerns similar between cultures? Body Image, 2, 177186. Chen, H., &Jackson, T. (2006). Differential processingof self-referencedversus other- referenced body information among American and Chinese young adults with body image concerns. Eating Behaviors, 7, 152160. Chen, H., & Jackson, T. (2007). Stability of body image concerns among Chinese ado- lescents: Nine-month testretest reliabilities of negative physical self scales. Perceptual and Motor Skills, 105, 677680. Chen, H., & Jackson, T. (2008). Prevalence and sociodemographic correlates of eat- ing disorder endorsements among adolescents and young adults from China. European Eating Disorders Review, 16, 375385. Chen, H., Jackson, T., & Huang, X. (2006). Initial development and validation of the Negative Physical Self Scale among Chinese adolescents and young adults. Body Image, 3, 401412. X. Gao et al. / Body Image 10 (2013) 191199 199 Cisler, J. M., & Koster, E. W. (2010). Mechanisms of attentional biases towards threat in anxiety disorders: An integrative review. Clinical Psychology Review, 30, 203216. Dai, Q., &Feng, Z. (2009). Decient inhibition of return for emotional faces in depres- sion. Progress in Neuro-Psychopharmacology &Biological Psychiatry, 33, 921932. Davidson, E. J., &Wright, P. (2002). Selective processing of shape and weight-related words in bulimia nervosa: Use of a computerised Stroop test. Eating Behaviors, 3, 261273. Dobson, K. S., & Dozois, D. A. (2004). Attentional biases in eating disorders: A meta-analytic review of Stroop performance. Clinical Psychological Review, 23, 10011022. Ellenbogen, M. A., & Schwartzman, A. E. (2009). Selective attention and avoidance on a pictorial cueing task during stress in clinically anxious and depressed par- ticipants. Behaviour Research and Therapy, 47, 128138. Friederich, H. C., Uher, R., & Brooks, S. (2007). I m not as slim as that girl: Neural bases of body shape self-comparisonto media images. NeuroImage, 37, 674681. Fox, E., Russo, R., & Bowles, R. (2001). Do threatening stimuli draw or hold visual attentioninsubclinical anxiety? Journal of Experimental Psychology: General, 130, 681700. Fox, E., Russo, R., &Dutton, K. (2002). Attentional bias for threat: Evidencefor delayed disengagement fromemotional faces. Cognition and Emotion, 16, 355379. Gao, X. (2010). Attentional bias among females with weight-related negative physi- cal self schema: Theoretical model and processing rules (Unpublished doctorial dissertation). Southwest University. Gao, X., Deng, X., & Chen, N. J. (2011). Attentional biases among body-dissatised young women: An ERP study with rapid serial visual presentation. International Journal of Psychophysiology, 2, 133142. Gao, X., Wang, Q. C., & Chen, H. (2012). Time course of attentional bias components toward body-shape related pictures among women with fat negative physical self: An eye movement study. Acta Psychologica Sinica, 44, 498510. Gao, X., Wang, Q. C., & Jackson, T. (2011). Biases in orienting and maintenance of attention among weight dissatised women: An eye-movement study. Behaviour Research and Therapy, 49, 252259. Garner, M., Mogg, K., & Bradley, B. P. (2006). Orienting and maintenance of gaze to facial expressions insocial anxiety. Journal of Abnormal Psychology, 115, 760770. Giel, K. E., Friederich, H., & Teufel, M. (2011). Attentional processing of food pic- tures in individuals with anorexia nervosa An eye-tracking study. Biological Psychiatry, 69, 661667. Glauert, R., Rhodes, G., & Fink, B. (2010). Body dissatisfaction and attentional bias to thin bodies. International Journal of Eating Disorders, 43, 4249. Greenhouse, W. W., & Geisser, S. (1959). On methods in the analysis of prole data. Psychometrika, 24, 95112. Hansen, C. H., & Hansen, R. D. (1988). Finding the face in the crowd: An anger superiority effect. Journal of Personality and Social Psychology, 54, 917924. Hooge, I. T. C., & Frens, M. A. (2000). Inhibition of saccade return (ISR): Spatio- temporal properties of saccade programming. Vision Research, 40, 34153426. Jackson, T., & Chen, H. (2008). Predicting changes in eating disorder symptoms among Chinese adolescents: A 9-month prospective study. Journal of Psycho- somatic Research, 64, 8795. Janelle, C. M., Hausenblas, H. A., & Ellis, R. (2008). The time course of attentional allocationwhilewomen highandlowinbodydissatisfactionviewself andmodel physiques. Psychology and Health, 24, 351366. Jansen, A., Nederkoorn, C., & Mulkens, S. (2005). Selective visual attention for ugly and beautiful body parts in eating disorders. Behaviour Research and Therapy, 43, 183196. Karparova, S. P., Kersting, A., & Suslow, T. (2005). Disengagement of attention fromfacial emotion in unipolar depression. Psychiatry Clinical Neuroscience, 59, 723729. Koster, E. H. W., Crombez, G., & Verschuere, B. (2004). Selective attention to threat in the dot probe paradigm: Differentiating vigilance and difculty to disengage? Behaviour Research and Therapy, 42, 11831192. Koster, E. H. W., Verschuere, B., & Crombez, G. (2005). Time-course of attention for threatening pictures in high and low trait anxiety. Behaviour Research and Therapy, 43, 10871098. Lee, M., & Shafran, R. (2004). Information processing biases in eating disorders. Clinical Psychology Review, 24, 215238. MacLeod, C., Mathews, A., & Tata, P. (1986). Attentional bias in emotional disorders. Journal of Abnormal Psychology, 95, 1520. Melnyk, S. E., Cash, T. F., & Janda, L. H. (2004). Body image ups and downs: Pre- diction of intra-individual level and variability of womens daily body image experiences. Body Image, 1, 225235. Mogg, K., Bradley, B. P., & Miles, F. (2004). Time course of attentional bias for threat scenes: Testing the vigilance-avoidance hypothesis. Cognition and Emotion, 18, 689700. Ouimet, A. J., Gawronski, B., &Dozois, D. A. (2009). Cognitive vulnerability toanxiety: A reviewand an integrative model. Clinical Psychology Review, 29, 459470. Perpina, C., Leonard, T., & Treasure, J. (1998). Selective processing of food and body- related information and autonomic arousal in patients with eating disorders. Spanish Journal of Psychology, 1, 310. Posavac, H. D., Posavac, S. S., & Posavac, E. J. (1998). Exposure to media images of female attractiveness and concern with body weight among young women. Sex Roles, 38, 187201. Posner, M. I. (1980). Orienting of attention. Quarterly Journal of Experimental Psychol- ogy, 32, 325. Ratcliff, R. (1993). Methods for dealing with reaction time outliers. Psychological Bulletin, 114, 510532. Rieger, E., Schotte, D. E., &Touyz, S. W. (1998). Attentional biases in eating disorders: A visual probe detection procedure. International Journal of Eating Disorders, 23, 199205. Rinck, M., Reinecke, A., & Ellwart, T. (2005). Speeded detection and increased dis- traction in fear of spiders: Evidence from eye movements. Journal of Abnormal Psychology, 114, 235248. Roefs, A., Jansen, A., & Moresi, S. (2008). Looking good, BMI, attractiveness bias and visual attention. Appetite, 51, 552555. Sackville, T., Schotte, D. E., & Touyz, S. W. (1998). Conscious and preconscious processing of food, body weight and shape, and emotion-related words in women with anorexia nervosa. International Journal of Eating Disorders, 23, 7782. Schneider, W., Eschman, A., & Zuccolotto, A. (2002). E-Prime users guide. Pittsburgh, PA: Psychology Software Tools Inc. Schutz, H. K., Paxton, S. J., &Wertheim, E. H. (2002). Investigationof bodycomparison among adolescent girls. Journal of Applied Social Psychology, 32, 19061927. Shafran, R., Lee, M., & Cooper, Z. (2007). Attentional bias in eating disorders. Inter- national Journal of Eating Disorders, 40, 369380. Simons, D. J. (2000). Attentional capture and inattentional blindness. Trends in Cog- nitive Sciences, 4, 147155. Smeets, E., Jansen, A., & Roefs, A. (2011). Bias for the (un)attractive self: On the role of attention in causing body (dis)satisfaction. Health Psychology, 30, 360367. Smeets, E., Roefs, A., & van Furth, E. (2008). Attentional bias for body and food in eating disorders: Increased distraction, speeded detection, or both? Behaviour Research and Therapy, 46, 229238. Smeets, E., Tiggemann, M., &Kemps, E. (2011). Body checking induces an attentional bias for body-related cues. International Journal of Eating Disorders, 44, 5057. Verkuil, B., Brosschot, J. F., &Putman, P. (2009). Interacting effects of worry and anx- iety on attentional disengagement fromthreat. Behaviour Research and Therapy, 47, 146152. Vitousek, K. B., & Hollon, S. D. (1990). The investigation of schematic content and processing in eating disorders. Cognitive Therapy and Research, 14, 191214. Waters, A. M., Nitz, A. B., &Craske, M. G. (2007). The effects of anxiety upon attention allocation to affective stimuli. Behaviour Research and Therapy, 45, 763774. Yiend, J., &Mathews, A. (2001). Anxiety and attention to threatening pictures. Quar- terly Journal of Experimental Psychology, 54A, 665681.