Attentional biases to body shape images in adolescents with anorexia nervosa: An exploratory eye-tracking study

Highlights

• Visual scanning behavior (VSB) in anorexia nervosa (AN) differs from controls.

• Biases in VSB appear when body shape images compete with social images.

• AN exhibits a bias toward thin body shapes over social images while controls do not.

• AN exhibits a bias toward fat body shapes over social images while controls do not.

• AN has a hierarchy of bias in VBS first to thin then fat, and finally social images.

Abstract

Body image distortion (BID) plays an important role in the etiology and maintenance of anorexia nervosa (AN). Previous studies of BID in AN showed small biases in visual scanning behavior (VSB) towards images of body shapes. The aim of this study is to investigate biases in VSB when body shape images compete with images with a different theme (social interactions) for subjects׳ attention. When images of thin body shapes (TBS) were presented alongside images of social interactions, AN patients (n=13) spent significantly more time looking at TBSs rather than at social interactions, but controls (n=20) did not. When images of fat body shapes (FBS) were presented alongside images of social interactions, AN patients spent significantly more time looking at FBSs rather than at social interactions, but controls did not. When images of TBSs, FBSs and social interactions were presented alongside each other, AN patients demonstrated a hierarchy in their attention allocation, choosing to spend the most viewing time on TBS images, followed by FBS images and then images with social interactions. Under the three experimental conditions, AN patients demonstrated large biases in their visual scanning behavior (VSB). Biases in VSB may provide physiologically objective measures that characterize patients with AN.

Introduction

Anorexia nervosa (AN) is a severe and potentially chronic disorder with high mortality (Harris and Barraclough, 1998, Hannerz et al., 2001, Arcelus et al., 2011, Rosling et al., 2011), and onset that is typically in adolescence (Hoek and van Hoeken, 2003). Body image distortion is pathognomonic of AN (American Psychiatric Association, 2000, Zhu et al., 2012) where patients over-estimate both their own and other women׳s body sizes (Tovee et al., 2000, George et al., 2011). Behavioral expressions of body-size over-estimation are dietary restraint, repeated checking of shape and/or weight (body checking) and the avoidance of seeing shape and/or weight when patients start to gain weight (body avoidance) (Shafran et al., 2004). While research is limited, evidence suggests that body image disturbance in AN is based on cognitive evaluative dissatisfaction (Epstein et al., 2001, Skrzypek et al., 2001, Gao et al., 2013). The cognitive behavioral theory of AN suggests that the psychopathological process accounting for the persistence and severity of AN may be a maladaptive cognitive schemata where individuals judge themselves largely in terms of their eating habits, body shape/weight and their ability to control those factors (Williamson, 1996, Jung and Lennon, 2003, Luck et al., 2005), whereby being underweight is associated with positive attributes (Ahern et al., 2008). These beliefs about body weight and shape lead to behavior such as repeated body checking that provides AN patients with a sense of control (Shafran et al., 2004), and to biases in emotion processing of body shape stimuli that may be associated with their strong fears of gaining weight and their relentless pursuit of thinness (Williamson et al., 1999, Zhu et al., 2012).

Historically, selective attention (attention bias) in AN to body shape images has largely been inferred through the use of the Stroop (Long et al., 1994, Overduin et al., 1995, Sackville et al., 1998, Lee and Shafran, 2004) and the Dot-probe tests (Shafran et al., 2007, Lee and Shafran, 2008). Meta-analyses of the Stroop interference test for body shape related words suggested that patients with AN demonstrate the modest interference for shape related words (Dobson and Dozois, 2004) in general, as well as for body shape words with negative overtones (large body physique) (Johansson et al., 2005). Studies with the dot-probe paradigm (Shafran et al., 2007) suggest that patients with AN have the modest attention bias to negative body shape related words and pictures. However, both paradigms provide an indirect measure of attention bias and have significant limitations. These limitations include the possibility that non-attention processes are responsible for the observed response bias and an inability to provide a full picture of the manner in which the word or picture was observed or looked at by the participant (Lee and Shafran, 2004).

A more direct and sensitive method to assess attention biases is to study visual scanning behavior (VSB). Unlike the Stroop and Dot-probe tests which only measure behavioral end products of cognitive processes, VSB provides a continuous measure of attention throughout the visual scanning process (Hermans et al., 1999). The method has been utilized to study attention bias in eating disorders (ED) with regard to body image and body size estimations. George et al. (2011) found that all female participants tend to fixate on the abdominal region. However, women with AN had a wider fixation pattern when compared to controls that incorporates collar bones, thighs, and the hip prominences. Jansen et al. (2005) found that women with ED symptoms demonstrated decreased attention to their own ‘beautiful’ body parts, and greater attention to their own ‘ugly’ body parts , while healthy controls focused more on their own ‘beautiful’ body parts and less on their own ‘ugly’ body parts. When viewing other bodies the pattern was reversed: high symptom participants allocated more attention to the beautiful parts of other bodies, and controls focused more on the ugly parts of the other bodies (Jansen et al., 2005). Blechert et al. (2009) found that bulimic patients, when considering other women׳s bodies, were more attentive than controls to images of thin body shapes and less attentive than controls to images of fat body shapes. This focus on the positives of other bodies was also supported by a study by Janelle et al. (2003). Women with greater body image disturbance avoided looking at the abdomen and thigh of endomorphic women (‘ugly’ parts of other bodies) when compared to controls and the investigators hypothesized a cognitive avoidance pattern. However, in a second, later study, Janelle et al. (2009) reported that when specifically considering the time course of attention allocation, women with high body image disturbance tended to focus on areas typical of body dissatisfaction (e.g., abdomen) during the latter stages of the viewing process whether they were looking at themselves or other women. These results fall in line with the results of Gao et al., (2013) who demonstrated that women with low and medium body mass indexes had more difficulty disengaging their attention from fat body images when they were more dissatisfied with their body weight. Finally, in a study that compared visual scanning patterns of adolescents with AN and controls, little difference was found between the two groups on pictures of underweight, normal-weight and overweight women apart from a greater focus on unclothed areas of the body by AN subjects (Horndasch et al., 2012).

To date, the studies of VSB have focused on the evaluation of the focus and pattern of attentional biases to somatic regions of the body when subjects look at images of themselves and/or controls. In these studies, the differences in VSB between AN patients and healthy controls are small and within groups variations are large, leading to significant overlap between the gaze patterns of the groups (Horndasch et al., 2012, von Wietersheim et al., 2012).

Given that the current literature rests on studies where visual stimuli have only images of one theme: body shapes, participants have never been provided with opportunities to shift their attention away from body shape images to images of a competing theme. We hypothesize, that the lack of opportunities to shift attention between images from competing themes reduces the observed differences in VSB between AN patients and healthy controls. To test this hypothesis, we explored the use of a new paradigm that we have used successfully to measure biases in VSB of patients with major depression disorder (Eizenman et al., 2003). In the new paradigm, participants are presented with visual stimuli that included images from different themes/categories and participants have adequate time to scan and re-scan the presented images. In such a paradigm, participants have more choice in the type of images they choose to focus on, and biases in VSB due to both early and late cognitive processing can be measured (Eizenman et al., 2003). Given that healthy controls tend to dwell on images with themes of positive social interactions (Eizenman et al., 2003), while AN patients may have a diminished response to pleasant pictures (Davies et al., 2011), and may derive less pleasure from social interactions (social anhedonia) (Tchanturia et al., 2012), we expect that by using images with social interaction alongside images of body shapes the differences in VSB between AN patients and healthy controls will increase.

Since frequent examination of specific body parts is a prominent clinical feature of AN and patients with AN find images with thin body shapes (TBS) and fat body shapes (FBS) more attention worthy than controls (Norris et al., 2006), we hypothesize the following:

• When presented with visual stimuli (slides) that contain images of thin body shapes (TBS) and positive social interactions, adolescent AN patients will spend more time viewing TBS images when compared with controls.

• When presented with slides that contain images of fat body shapes (FBS) along with images of positive social interactions, adolescent AN patients will spend more time viewing FBS images when compared with controls.

Finally, based on previous studies that suggest that symptomatic women are less attentive to images of fat body shapes and more attentive to images of thin body shapes (Janelle et al., 2003, Blechert et al., 2009), we hypothesize the following:

• When presented with images of both TBSs and FBSs, adolescent AN patients will spend more time viewing TBS images than FBS images.

We carried out an experiment with three different conditions to test the above hypotheses. Data from the first and second conditions were used to test hypothesis 1 and 2, and hypothesis 3 was tested with data from condition 3.