تعصب توجه برای بدن و مواد غذایی در اختلالات تغذیه ای: افزایش حواس پرتی، تشخیص شتاب و یا هر دو؟

Abstract Previous research suggests that eating disorder patients show an attentional bias for body- and food-related information. However, so far little is known about the mechanisms that underlie the attentional favoring of this particular information in eating disorder patients. In the present study, we used both a body and a food visual search task to study speeded detection and increased distraction in eating disorder patients (n=67) and healthy controls (n=60). Compared with controls, eating disorder patients showed evidence of speeded detection of body-related information, and increased distraction by food information. These results suggest that the mechanism underlying the biased attentional allocation of eating disorder patients varies, and is dependent upon the type of information they are presented with.

Introduction It has been only two decades since researchers have begun to acknowledge the relevance of cognitive biases in the etiology and maintenance of eating disorders (EDs) (for reviews see Faunce, 2002; Lee & Shafran, 2004; Williamson, White, York-Crowe, & Stewart, 2004). Cognitive models point out that eating pathology arises from maladaptive knowledge structures (e.g., schemas) that are involved in the allocation of attention, in memory and in the interpretation of incoming information (Hargreaves & Tiggemann, 2002; Williamson et al., 2004). Activation of these knowledge structures causes disorder-relevant information to be processed in a biased manner, resulting in a range of cognitive biases in attention, judgment and memory (Williamson et al., 2004). The focus of the current study is on one of these biases: the exact nature of the attentional bias in EDs. An attentional bias refers to the tendency to selectively attend to disorder-relevant stimuli (e.g., Mathews & MacLeod, 2005; Williamson et al., 2004). According to cognitive models, individuals suffering from EDs are more likely to give priority to cues pertaining to body and food-related information than to neutral cues, in comparison to healthy people. Indeed, the great majority of studies employing the emotional Stroop paradigm (e.g., Williams, Mathews, & MacLeod, 1996) found that ED patients show increased interference when naming the color of disorder-relevant word stimuli as compared with neutral stimuli (for reviews see: Dobson & Dozois, 2004; Lee & Shafran, 2004). Although these interference effects have generally been interpreted as direct evidence for an attentional bias, alternative explanations have been put forward (Jansen, Nederkoorn, & Mulkens, 2005; Lee & Shafran, 2004; Macleod, 2005). For example, De Ruiter and Brosschot (1994) have argued that attempts to cognitively avoid the processing of disorder-relevant word stimuli might also result in increased interference scores. Given this uncertainty about the meaning of increased interference scores, no firm conclusions can be drawn about the existence of an attentional bias in ED patients on the basis of results in the emotional Stroop paradigm. A better alternative for studying attentional biases is the dot-probe paradigm (Macleod, Mathews, & Tata, 1986), because it allows for the differentiation between attention directed towards or away from a particular type of information. To date, two studies have investigated attentional processes in ED patients using the dot-probe paradigm. Rieger et al. (1998) demonstrated that ED patients showed a tendency to direct their attention towards words denoting a large physique and away from words denoting a thin physique. More recently, using a pictorial version of the dot-probe paradigm, Shafran, Lee, Cooper, Palmer, and Fairburn (2007) found robust attentional bias effects for eating and weight-related stimuli in ED patients in comparison to controls, but less consistent effects for shape-related stimuli. Although it is possible to use the dot-probe paradigm to distinguish engagement and disengagement subcomponents of attention, either by manipulating the presentation duration (Mogg, Bradley, Miles, & Dixon, 2004) or by including neutral trials (Koster, Crombez, Verschuere, & De Houwer, 2004; Salemink, van den Hout, & Kindt, 2007), the dot-probe paradigm has not been used in this way in EDs research. It thus remains an unresolved issue as to how the attentional bias in ED patients can be understood in terms of facilitated attention to or slowed withdrawal from the disorder-relevant information. Though possibly reflecting a somewhat different distinction, in this study we use a paradigm that is able to distinguish two subcomponents of attention: speeded detection (i.e., increased orienting towards relevant stimuli) and distraction (i.e., increased distraction by relevant stimuli). Like Rinck, Reinecke, Ellwart, Heuer, and Becker (2005) we use the odd-one-out variant of the visual search paradigm (Hansen & Hansen, 1988). Rinck et al. (2005) studied the nature of attentional bias in spider-fearful individuals. Participants were presented with matrices of 20 pictures and they were instructed to indicate whether the matrix consisted of 20 animal pictures of the same category or whether it included one animal picture from a different category. Results indicated that both speeded detection of threatening target pictures (i.e., faster detection of a spider picture among 19 neutral pictures than a neutral picture among 19 neutral pictures from another category) and increased distraction by threatening distractors (i.e., slower detection of a neutral image among 19 spider pictures than a neutral image among 19 neutral images from another category) were involved in the attentional processing of spider-fearful participants. To sum up, given both the controversy concerning the interpretation of increased interference scores in the emotional Stroop research and the lack of knowledge about which mechanism underlies an attentional bias in EDs, it is of interest to investigate more precisely the attentional bias for body and food stimuli in ED patients. Inspired by the visual search methodology as adopted by Rinck et al. (2005), we designed a body- and a food-related version of the odd-one-out visual search task to study both speeded detection and increased distraction. Speeded detection of disorder-relevant concepts (i.e., body- or food-related words) is studied by comparing response latencies to detect a disorder-relevant target word vs. a neutral target word among neutral distractor words from another category. Increased distraction is studied by comparing response latencies to detect a neutral target word among disorder-relevant vs. neutral distractor words from another category. It is hypothesized that ED patients show evidence of speeded detection of and increased distraction by body- and food-related information, in comparison to controls.

Results Data reduction and target-absent trials The main analyses were done on the target-present trials. For both tasks, errors (i.e., misses; food: 8.60%; body: 8.93% of the target-present trials) and responses faster than 200 ms and slower than 20.000 ms (food: 0.03%; body: 0.06% of the target-present trials) were discarded. Furthermore, response latencies higher than three SD above the overall mean of the remaining response latencies were excluded (body: 1.0%; food: 1.0% of the target-present trials). None of the response latencies was lower than three SD below the mean. False alarm rates to the target-absent trials in the body visual search task were low (body: 4.96%; country: 3.92%; music: 5.20%). A 3 (stimulus category: body vs. music vs. country)×2 (group: ED vs. controls) repeated measures ANOVA, revealed that the stimulus category×group interaction was not significant, F (2, 121)=0.35, ns, indicating no significant differences between ED and controls on the false alarm rates for body (ED: 5.45%; controls: 4.40%), country (ED: 3.94%; controls: 3.90%) and music (ED: 5.9%; controls: 4.41%) target-absent trials. Main effects of stimulus category, F (2, 122)=.35, ns, and of group, F (1, 123)=.61, ns, were also non-significant. False alarm rates to the target-absent trials in the food visual search task were also low (HCF: 2.48%; LCF: 3.68%; colors: 2.32%; names: 4.48%). A 4 (stimulus category: HCF vs. LCF vs. colors vs. names)×2 (group: ED vs. controls) repeated measures ANOVA, revealed a significant main effect of stimulus category, F (3, 121)=4.53, p<.01, but not of group, F (1, 123)=0.05, ns. The stimulus category×group interaction was not significant, F (3, 120)=0.42, ns, indicating no significant differences between ED and controls on the false alarm rates for HCF (ED: 2.73%; controls: 2.20%), LCF (ED: 3.48%; controls: 3.90%), colors (ED: 2.42%; controls: 2.20%) and names (ED: 3.94%; controls: 4.92%) target-absent trials. Hypothesis 1. ED patients show evidence of speeded detection of body-related information in comparison to controls. Results were analyzed in a 2 (group: ED vs. controls)×2 (target type: body vs. neutral) repeated measures ANOVA. Consistent with our hypothesis, a significant group×target type interaction, F (1, 123)=11.98, p<.01, ηp2=.09, was found, qualifying main effects of target type, F (1, 123)=30.95, p<.001, ηp2=.20, and group, F (1, 123)=15.03, p<.001, ηp2=.11. See Fig. 1a for means and SEs. Additional independent samples t-tests indicated that ED patients were significantly slower than controls at detecting body-related target words, t (123)=2.69, p<.01, d=.62, and neutral target words, t (123)=4.62, p<.001, d=.83. Even though ED patients were slower than controls on both types of trials, the significant group×target type interaction shows that ED patients, relative to controls, did show a benefit in the speed with which they detected a body-related target word as compared with a neutral target word. In other words, the difference in detection speed between ED patients and controls was less pronounced for body target words than for neutral target words, proving the relative benefit for body-related words in ED. Hypothesis 2. ED patients show evidence of increased distraction by body-related information in comparison to controls. (a) Mean response latencies for trials in which participants searched for one ... Fig. 1. (a) Mean response latencies for trials in which participants searched for one body target word among 19 neutral words (target type=body), and for trials in which participants searched for one neutral target word of one category among 19 neutral words of another category (target type=neutral). Results are presented separately for ED patients and healthy controls. Error bars represent one standard error. This graph shows that ED patients show evidence of speeded detection of body-related information, compared with controls. (b) Mean response latencies for trials in which participants searched for one neutral target word among 19 body-related distractor words (distractor type=body), and for trials in which participants searched for one neutral target word of one category among 19 neutral distractor words of another category (distractor type=neutral). Results are presented separately for ED patients and healthy controls. Error bars represent one standard error. This graph shows that ED patients are not more distracted by body-related information, compared with controls. Figure options Results were analyzed in a 2 (group: ED vs. controls)×2 (distractor type: body vs. neutral) repeated measures ANOVA. Of main interest to our hypothesis, a significant group×distractor type interaction, F (1, 123)=15.81, p<.001, ηp2=.11, was found, qualifying main effects of distractor type, F (1, 123)=60.17, p<.001, ηp2=.33, and of group, F (1, 123)=13.51, p<.001, ηp2=.10. See Fig. 1b for means and SEs. Additional independent samples t-tests indicated that ED patients were significantly slower than controls with neutral distractors, t (123)=4.62, p<.001, d=.83, and with body-related distractors, t (123)=2.31, p<.05, d=.42. Even though ED patients were slower than controls on both types of trials, the significant group×distractor type interaction shows that, relative to controls, they did show a benefit for the body-related distractors as compared with the neutral distractors. In contrast to our hypothesis, the difference in distraction between ED patients and controls was less pronounced for body distractor words than for neutral distractor words. Hypothesis 3. ED patients show evidence of speeded detection of high-caloric food information in comparison to controls. Results were analyzed in a 2 (group: ED vs. controls)×3 (target type: high-caloric food vs. low-caloric food vs. neutral) repeated measures ANOVA, which yielded a significant main effect of group, F (1, 123)=5.71, p<.05, ηp2=.05, and a marginal significant effect of target type, F (1.9, 231.24)=2.64, p=.07, ηp2=.02. However, the group×target type interaction, F (1.9, 231.24)=0.77, ns, was not significant. See Fig. 2a for means and SEs. Taken together, no evidence was found for speeded detection of high-caloric food words in ED. Hypothesis 4. ED patients show evidence of increased distraction by high-caloric food information in comparison to controls. (a) Mean response latencies for trials in which participants searched for one ... Fig. 2. (a) Mean response latencies for trials in which participants searched for one high-caloric food target word among 19 neutral words (target type=high-caloric food), for one low-caloric food target word among 19 neutral words (target type=low-caloric food), and for trials in which participants searched for one neutral target word of one category among 19 neutral words of another category (target type=neutral). Results are presented separately for ED patients and healthy controls. Error bars represent one standard error. This graph shows that ED patients are not faster at detecting high-caloric food information, compared with controls. (b) Mean response latencies for trials in which participants searched for one neutral target word among 19 high-caloric food distractor words (distractor type=high-caloric food), for one neutral target word among 19 low-caloric food distractor words (distractor type=low-caloric food) and for trials in which participants searched for one neutral target word of one category among 19 neutral distractor words of another category (distractor type=neutral). Results are presented separately for ED patients and healthy controls. Error bars represent one standard error. This graph shows that ED patients are more distracted by high-caloric food information, compared with controls. Error bar represents one standard error. Figure options Results were analyzed in a 2 (group: ED vs. controls)×3 (distractor type: high-caloric food vs. low-caloric food vs. neutral) repeated measures ANOVA. Of main interest to our hypothesis, a marginally significant group×distractor type interaction was found, F (2.0, 243.50 1)=2.91, p=.06, ηp2=.02, qualifying main effects of distractor type, F (2.0, 243.50 1)=2919, p<.001, ηp2=.19, and of group, F (1, 123)=7.27, p<.001, ηp2=.06. See Fig. 2b for means and SEs. Additional independent samples t-tests indicated that ED patients were significantly slower than controls at detecting neutral target words among high-caloric food distractors, t (123)=3.25, p<.01, d=.64, and among neutral distractors of another category, t (123)=2.20, p<.05, d=.40. In contrast, both groups were equally slow at detecting a neutral target word among low-caloric food distractors, t (123)=1.75, p=.08. To test whether the difference between ED patients and controls was larger for high-caloric foods than for neutral distractors, an additional 2 (group: ED vs. controls)×2 (distractor type: high-caloric food, vs. neutral) repeated measures ANOVA was conducted. Importantly, we found a significant group×distractor type interaction, F (1, 123)=4.94, p=.03, ηp2=.04, qualifying main effects of distractor type, F (1, 123)=30.51, p<.001, ηp2=.20, and of group, F (1, 123)=9.08, p<.001, ηp2=.07. The group×distractor type interaction shows that the difference between ED patients and controls was more pronounced for high-caloric food distractors than for neutral distractors. These results provide clear evidence for increased distraction by high-caloric food words in ED patients.