تعصب حافظه در بی اشتهایی عصبی: شواهدی از فراموشی هدایت شده

The aim of this study was to examine memory bias for disorder-relevant information in anorexia nervosa by using the directed forgetting paradigm. Normal controls and patients with anorexia nervosa were given a list consisting of neutral and disorder-relevant words, which they were either asked to remember (R) or forget (F). Memory performance was measured by a free recall and a Yes/No recognition task for all items. There was a directed forgetting effect for both groups; however, the magnitude of the effect (difference between R and F words) was smaller for the patient group due to higher recall of F items. Further analyses showed that this was true only for disorder-relevant but not for neutral items. Our findings support the existence of a strong memory bias for disorder-relevant information in patients with anorexia nervosa, who had difficulty in avoiding the processing of information that they were asked to forget.

Cognitive behavioral theories of eating disorders assign a significant role to cognitive processes in the maintenance of the pathology (Vitousek & Hollon, 1990; Williamson, White, York-Crowe, & Stewart, 2004). These theories argue that patients with eating disorders have a self-schema that focuses on different domains (e.g., size, shape, weight, food, etc.). These schema guide information processing and lead attention to weight-relevant stimuli (either internal or external) in the environment, resulting in a bias favoring disorder-related information (Williamson, 1996; Williamson, Muller, Reas, & Thaw, 1999; Williamson et al., 2004). One prediction of cognitive–behavioral models is that such disorder-relevant information would be well attended to and encoded because patients with eating disorders cannot avoid processing such information (i.e., cannot disengage from the stimuli). Since there already is an existing schema, acquisition of and memory for that information would be facilitated. More recent transdiagnostic theoretical approaches (Fairburn, Cooper, & Shafran, 2003) argue for the continuity between eating disorders that are claimed to be categorically different (anorexia nervosa (AN), bulimia nervosa (BN), and EDNOS). This model suggests that there is a core psychopathology common to all these eating disorders and that common mechanisms, including cognitive biases, play a role in the maintenance (Shafran, Lee, Cooper, Palmer, & Fairburn, 2007). 1.1. Attention bias in eating disorders There are several studies addressing attention biases in eating-related psychopathology, mostly using the emotional Stroop task. This line of research consistently showed that patients with AN and those with BN (e.g., Ben-Tovim & Walker, 1991; Ben-Tovim, Walker, Fok, & Yap, 1989; Cooper, Anastadiades, & Fairburn, 1992; Davidson & Wright, 2002; Green, Corr, & De Silva, 1999) as well as analogue groups such as dieters or high-restraint individuals (Ferraro, Andres, Stromberg, & Kristjanson, 2003; Francis, Stewart, & Hounsell, 1997) have positive attentional biases toward disorder-related material. Similar findings have been reported with the dot-probe paradigm (Reiger et al., 1998; Shafran et al., 2007). In their review of information processing in eating disorders, Lee and Shafran (2004) observed that (a) an overwhelming majority of studies with patients and analogue groups found a bias for disorder-relevant information, (b) such biases tended to be more consistent for patients than for analogue groups, and (c) larger effects were observed in patients with AN than those with BN. They also noted a relative scarcity of research on memory biases in patients with eating disorders. They were able to locate more than 25 published articles on attention biases, but only six on memory biases, three of which were carried out with analogue samples. 1.2. Memory bias in eating disorders In the first study on this issue, King, Polivy, and Herman (1991) looked at whether obese and eating disordered individuals showed a memory bias for weight- and food-related information about a person described in an essay. They found that both groups recalled more weight- and food-related information than other items, implying the existence of a memory bias. They also reported a moderate correlation between restraint scores and the frequency of mention of disorder-relevant information in a different task in which the participants were asked to report several types of information (e.g., five types of things they spent most time thinking about). However, the results of this study should be interpreted with some caution because the comparison involved only 6 eating disordered patients (vs 24 individuals with obesity), and also because data from normal control (NC) participants were not analyzed in comparison to these patient groups. Sebastian, Williamson, and Blouin (1996) compared free recall memory performance for fat-related and fat-unrelated information of patients with eating disorders, individuals with weight preoccupation and NC participants. They found that eating disordered patients but not the other two groups recalled more fat-related than fat-unrelated words. Hermans, Pieters, and Eelen (1998) compared AN patients with nondieting NCs on both explicit and implicit memory tasks and found a bias favoring anorexia-related words in the explicit memory task but not in the implicit memory task. Patients with AN recalled more anorexia-related words than positive, negative and neutral words whereas there was no difference in the recall performance of the nondieting controls. Furthermore, there was no difference between the groups in the implicit memory task, leading the authors to conclude that memory bias in AN patients may be limited to explicit memory. It has to be noted that these findings are based on a relatively small sample of patients and control participants (12 in each group). Research with analogue groups also found a memory bias for diagnosis-related information. For instance, Israeli and Stewart (2001) compared high- and low-restraint participants on memory for forbidden (high fat) food words and animal words by using an incidental learning paradigm. Although high-restraint participants did not remember more forbidden food words than the low-restraint participants, they remembered more forbidden food words than animal words, providing partial evidence for memory bias. Other studies reported a memory bias in individuals with body dysphoria. In one of these studies, individuals with body dysphoria recalled more fatness words compared to normals (Baker, Williamson, & Sylve, 1995) and in another (Watkins, Martin, Muller, & Day, 1995) high-dysphoria individuals recalled more body-related items than low-dysphoria individuals. The main purpose of the present study was to provide further data regarding memory bias in AN patients in comparison to NC participants. We employed the directed forgetting procedure, which, to our knowledge, had not been used with individuals diagnosed with eating disorders. It has, however, been used frequently in studies with patients diagnosed with anxiety disorders and posttraumatic stress disorder (e.g., McNally, Metzger, Lasko, Clancy, & Pitman, 1998; Tolin, Hamlin, & Foa, 2002; Wilhelm, McNally, Baer, & Florin, 1996). Directed forgetting provides an especially good test of memory bias because in this procedure individuals are asked to specifically forget some of the information they are presented. There are two methods of directed forgetting: item method and list method. We used the item method, where individual words were followed by an instruction either to remember or to forget that word. There is wide agreement that the directed forgetting in this method is due to selective rehearsal; that is, participants remember more R words than F words because R words receive more rehearsal than F words, which should not be rehearsed once the participants are informed that the word is not going to be on the memory test (Johnson, 1994; MacLeod, 1998). Therefore, item method directed forgetting might provide clues as to how AN patients compared to NC participants process information that needs to be ignored after brief exposure. Therefore, one novel aspect of the present study was the use of the directed forgetting procedure. Moreover, studies on cognitive biases mostly looked at bias for negative information. In their review Lee and Shafran (2004) identified only one study which used positive shape and weight items along with negative ones. We presented our participants with a study list of words containing neutral words along with positive and negative words that were disorder-relevant (e.g., shape, weight, food). First, we expected both the AN and the NC groups to show a standard directed forgetting effect (i.e., remember more R words than F words). Second, in line with the cognitive–behavioral models of eating disorders, we expected patients with AN to show a diminished directed forgetting effect, mainly because they would remember more F words compared to NCs, and that this would occur only for disorder-related words but not for neutral ones. Third, on the basis of earlier work with normal participants (Basden, Basden, & Gargano, 1993; MacLeod, 1999), we expected that these findings would hold true in the recognition task as well.