"Food for memory": pictorial food-related memory bias and the role of thought suppression in high and low restrained eaters.
According to the leading cognitive theory of eating pathology, our thinking is organized by so-called schemas (Williamson et al. 2004). Schemas are knowledge structures, involved in guiding information processing (Neisser 1976). In eating pathology, schemas reflect an overconcern with food, weight, and/or shape (Cooper and Fairburn 1992). It has been suggested that food-focussed schemas not only exist in clinical eating-disordered patients (see Mizes and Christiano 1995, for a review) but also characterize restrained eaters (Hoffmeister et al. 2010; Morris et al. 2001).
The three best-known theoretical consequences of the activation of such schemas are (a) selective attention, (b)judge-ment bias, and (c) memory bias for schema-consistent stimuli (Vitousek and Hollon 1990; Williamson et al. 2004). These cognitive biases are considered pivotal in the psychopathology of eating disorders and are assumed to instigate maladap-tive weight control behaviors (Overduin et al. 1995). Selective attention toward food cues or problems in directing attention away from food cues have been established quite extensively in eating-disordered and restrained eaters (for review, see Brooks et al. 2011a, b). Although studied to a lesser extent, there is also evidence for a schema-consistent perceptual judgement bias (i.e., how the perception of food-related stimuli is altered in relation to an individual's current concerns), particularly in restrained eaters (e.g., Brooks et al. 2011a, b). Studies about memory biases, however, are still scarce. In patients with eating disorders, studies so far have yielded mixed results. Some studies indicated a memory bias for food- or body-related cues in (satiated) women with anorexia nervosa (e.g., Nikcndei et al. 2008; Tekcan et al. 2008), yet other studies failed to find a memory bias (e.g., Hunt and Cooper 2001) or even found poorer recollection of body-related stimuli in individuals with bulimia nervosa (Legenbauer et al. 2010). Despite their obvious relevance, studies about food-related memory biases in restrained caters are almost non-existent. To our knowledge, only two studies have been conducted to date. In a study by Boon et al. (2000), restrained eaters were found to need less time in recalling food words compared to neutral words. Israeli and Stewart (2001) found that contrary to their predictionN, high restrained eaters did not remember more food cues than low restrained eaters did. However, the former group did remember more food words than animal words, which according to the authors, may suggest a relative memory bias for food cues in restrained eaters. Given the scarcity and inconclusiveness of previous research on memory biases in this relevant group of restrained eaters on the one hand, and the central role of memory biases in cognitive theory on the other hand, further research is definitely warranted.
Hence, the prime focus of the current study is to examine a food-related memory bias in restrained eaters. Unique to the present study, three novel aims are included. First of all, the two studies with restrained eaters conducted so far have both used food words as cues. Yet, in real-life situations, food-related information is more commonly presented in the form of images (e.g., billboards, commercials) than written words. A study by Storniark and Torkildsen (2004) already showed the relevance of including pictorial food stimuli in studying selective attentional processing in patients with eating disorders. Moreover, it is possible that memorizing words may not trigger food-related schemas as strongly as pictorial stimuli (Legenbauer et al. 2010). Hence, in the present study, ecologically more valid pictorial cues will be used for the first time, instead of using written cues. Second, no studies so far have investigated a delayed memory bias for food cues in restrained eaters. Yet, if food-related cognitive biases are presumed to trigger maladaptive weight control behaviors later on (Overduin et al. 1995), one would expect these cognitive biases to be present over a longer period of time. Consequently, the existence of a delayed food-related memory bias will be examined in the present study. Finally, an additional aim is to examine the potential role of thought suppression in schema activation. As mentioned above, memory biases for food stimuli stem from the activation of food-related schemas (Williamson et al. 2004). It is therefore important to understand the particular processes that contribute to schema activation. Obviously, the mere sight of palatable foods may enhance the cognitive focus on food cues (Mogg et al. 1998); yet, other, more intricate mechanisms may also play an essential role. These mechanisms include thought suppression, denoting deliberate attempts to suppress unwanted thoughts (Wegner 1994). Research on thought suppression has repeatedly demonstrated that trying to avoid or suppress unwanted thoughts can backfire and cause an ironic increase in unwanted thoughts (for a review, see Abramowitz et al. 2001). This phenomenon is known as the post-suppression rebound effect (Wegner 1994). According to Davies and Clark (1998), this effect occurs especially for thoughts that are relevant to current concerns. The similarity with the counterproductive effects of restrained eating is striking. Restrained eaters also try to control a certain urge (i.e., eating), yet often find themselves overindulging (Stice and Agras 1998). Herman and Polivy (1980) even posit that those who score highly on restraint ironically are the ones who are most likely to relapse and binge eat.
Applying Wegner's (1994) theory to food-related thoughts, it can be expected that since suppression makes individuals hypervigilant for these thoughts, they may well act as a trigger for food-related schemas in eating-concerned individuals (Soetens et al. 2006). Consequently, one can assume that the post-suppression rebound effect can present itself as a memory bias (Meier et al. 2011; Nixon et al. 2009). Indeed, Kircanslci et al. (2008) established that threat-related thought suppression can lead to an elevated explicit memory bias for threat words. To our knowledge, only one study so far (Soetens and Bract 2007) has looked at the effects of food-related thought suppression on memory for food words, but no effects of thought suppression on memory were found. It was suggested that the absence of a memory bias might have been due to the particular suppression paradigm used in that study not being sensitive enough to detect effects. Lavy and van den Hout (1994) established that standard 5-min thought suppression instructions are quite fragile and may not be extensive enough to have a prolonged influence on later performance tasks. Therefore, in the current study, suppression instructions will be extended to 10 min.
In sum, to investigate our research questions, high and low restrained eaters will be asked either to suppress thoughts about food (Suppression food) or animals (Suppression animals) or merely to monitor thoughts about food (Control food) or animals (Control animals). Afterward, all participants will be exposed to photographs of attractive foods and animals and will receive two free recall tasks for these photographs as measurements of memory, once immediately after the pictorial exposure and once after a short delay (6 min). First, it is hypothesized that high restrained eaters overall will remember more pictorial food cues than low restrained eaters (i.e., main effect of Restraint group on direct memory for food cues) and relative to remembering Control animal cues (see Israeli and Stewart 2001). Second, we assume that after food-related thought suppression, participants will remember more food cues in the recall task than participants not suppressing food-related thoughts (i.e., main effect of Suppression condition on direct memory for food; see Kircanski et al. 2008; Meier et al. 2011). Additionally, based on the findings by Davies and Clark (1998), which showed that it is more difficult to suppress thoughts that are relevant to current concerns, we will investigate whether the alleged effects of food-related thought suppression on memory for food cues are the strongest for the high restrained eaters (i.e., interaction effect Restraint group x Suppression condition on direct memory for food cues). Finally, we expect similar effects to occur after the 6-min delay (i.e., main effects of Restraint group, better recall of food cues versus animal cues in high restrained eaters, main effect of suppression condition, as well as an interaction effect Restraint group x Suppression condition on delayed memory for food cues).
Participants The sample consisted of 70 undergraduate psychology students (61 women and 9 men), with a mean age of 18.91 years (SD=2.23). The average amount of time that had passed since the last meal was 3.51 h (SD=2.11). The mean BMI (Body Mass Index: weight/height x height) was 21.18 (SD=3.29). One participant was excluded from the analyses because the number of target thoughts during thought suppression deviated more than 3 standard deviations from the mean. Two additional participants were excluded from the analyses because their scores for dietary restraint did not meet the cutoff criteria for defining high and low restrained eaters (also see "Measures"). All participants received a course credit for their participation in the study. Informed consent was appropriately obtained, and the experiment was approved by a college's ethical committee.
Dutch Eating Behaviour Questionnaire (DEBQ) The DEBQ (Van Strien et al. 1986b; originally published in Dutch by Van Strien et al. 1986a) is a 33-item questionnaire designed to assess three characteristics of eating behavior: restrained eating (10 items), emotional eating (13 items), and external eating (10 items). The subscale Restrained Eating was used in the present study. The questionnaire consists of 5-point Likert scales with categories ranging from never (1) to vety often (5). In previous research, both reliability (Cronbach's alpha > 0.80) and validity of the scale were proven to be adequate (Banasialc et al. 2001). The DEBQ can be used in both adult and adolescent populations. Cronbach's alpha in the present study amounted to 0.93 for the Restrained Eating scale. The DEBQ Restrained Eating scale is a well-known and commonly used measure of dietary restraint, and (Dutch) norms for females and male are available. This was of particular importance to this study since we compared each participant's individual score to the norm scores for dietary restraint, accounting for gender, after all data were collected. A norm-based cutoff point has the advantage that it is not sample dependent (contrary to for example a median split; see Whisman and McClelland 2005). Mean scores of 2.65 or higher for females and 2.26 or higher for males were deemed to be above average to very high (M+SEM[less than or equal to]score[less than or equal to]99th percentile) and were labeled as high restrained eaters (n=24). Scores of 2.56 and lower for females and 2.15 or lower for males were deemed to be below average to very low (1st pereentile[less than or equal to]score[less than or equal to]M-SEM) and were labeled as low restrained eaters (n=46).
Food- and Animal-Related Thought Recording In line with previous studies (see Abramowitz et al. 2001), the number of food- or animal-related thoughts expressed during the Suppression and Control conditions was registered by asking participants to press a clicking device with their dominant hand each time they had a target thought (food or animal, depending on the condition; see "Procedure"), while simultaneously reporting the thought out loud. The experimenter registered the number of clicks throughout the experiment, as well as the expressed thoughts. An independent referee checked all responses using a coding manual that was developed prior to the experiment, in order to omit potential accidental clicks from the analyses. This coding manual described which thoughts were regarded as references to food or animals, respectively (e.g., thoughts about particular foods and animals but also references to brands of food, desire, and willingness to eat, and all actions associated with the process of eating, or animals).
Pictorial Sorting Risk In order to consciously expose each participant to the same food and animal cues, a pictorial sorting task was developed. The task consisted of a total of 40 photographs, including 20 pictures of high-caloric, palatable foods (e.g., French fries, ice cream, pastry, soda, and hamburger) and 20 pictures of charming animals (e.g., cat, dog, dolphin, squirrel, and hedgehog). The photo-giaphs were selected based on pretest valence ratings in a small sample of young adults (n=8, mean age=27 years). The sample granted similar (neutral to positive) valence ratings to the food cues and the animal cues. Care was taken in matching the pictures for size of the target cue, brightness, color scheme, and camera angle as well as possible. The photographs were randomly shuffled and placed on one pile in front of the participant (photographs faced down). Participants were asked to turn the photographs one by one, to name out loud what they saw and to sort the photographs on two piles, according to category (animals on one side and foods on the other side), as fast and as accurate as possible.
Free Recall Task Two free recall tasks were employed as measurements of (explicit) memory (e.g., see Paunovic et al. 2002). Participants were given no prior notice of these recall tasks, and they were asked to write down all the cues they could remember from the pictorial sorting task on a blank piece of paper. No further sequence or spelling restrictions were enforced.
Probing for Suspicion All participants were asked to write down in their own words what they thought the study was about. This way, we were able to see whether they figured out the real intentions of the study, potentially tailoring their responses. None of the participants indicated that they had figured out the real intentions of the study.
Valence Questionnahr To check for potential differences in general valence perception of the cues in the pictorial sorting task, participants were asked to rate the words on a word valence rating scale after the experiment. Each word was rated on a 7-point rating scale, ranging from very negative to very positive. Mean valence scores were calculated by summing scores per category (animal/food).
Demographic Questionnaire Participants were asked about their age, sex, height, weight, time since last meal, and whether they were currently dieting (yes/no) or had been on a diet in the past (yes/no).
Participants were randomly assigned to one of four conditions before arrival at the laboratory: Suppression food (n=21), Suppression animals (n=17), Control food (n=16), and Control animals (n=16). There was a slight inequality in the number of participants per group, which was due to the fact that some of the participants didn't show up at the time of appointment, despite being enrolled in the course credit system. All participants were told that they would participate in a study about "thoughts and thinking."
In the conditions Suppression food and Control food, all participants were first asked to write down a description of their favorite food and indicate why this was their favorite food. Likewise, in the conditions Suppression animals and Control animals, all participants wrote down a description of their favorite animal and indicated why they liked that particular animal. The purpose of this procedure was to expose all participants equally to the target thought of their particular condition and to make these thoughts conscious to them. For the manipulation of thought suppression, a modified version of Wegner et al.'s (1987) original thought suppression paradigm was used (also see Abramowitz et al. 2001). The experiment consisted of three subsequent trial periods, over a 5-min time span each. During the first trial period (baseline), all participants were told to think about anything they liked, to report these thoughts out loud and to press a clicking device each time a target thought came to mind (i.e., thoughts referring to "food" in the conditions Suppression food and Control food and thoughts referring to "animals" in the conditions Suppression animals and Control animals). This way, participants monitored whether thoughts about food or animals occurred, while expressing every thought they experienced out loud. During the identical second and third trial period, participants in the two suppression conditions were explicitly asked not to think about food (condition Suppression food) or animals (condition Suppression animals), while participants in the two control conditions were again given the instructions to think about anything they wanted. All four groups were again asked to monitor their thoughts about food or animals, while expressing their thoughts. The frequency of references to the target thought (food or animal) in the instructions was held constant across conditions in order to control for unequal priming effects. In line with Lavy and van den Hout (1994), during the second and third trial period, electrodes connected to a computer were placed on the participants' temples, seemingly to monitor brain waves. Participants were told that we were interested in monitoring their brain activity during the suppression/monitoring of their thoughts. The purpose of this procedure was to stress the importance of compliance with the (suppression or control) instructions. After the suppression phase, participants in the suppression conditions were told that they could again think about anything they wanted.
Immediately after the suppression/control instructions, all participants were given the pictorial sorting task (see "Measures"). Following this task, all participants completed the first unannounced free recall task (see "Measures"). Following a 6-min distracter task (i.e., participants were given mazes, to be completed on paper), participants were asked to complete the same free recall task for a second time. Finally, the probing for suspicion check, the valence questionnaire, the demographic questionnaire, and the DEBQ Restrained Eating subscale were administered (see "Measures"). The total experiment took about 45 min for each participant. Written debriefing was provided collectively after the completion of the experiment.
First, preliminary analyses of variance (ANOVAs), independent sample t tests, and chi square tests were used to examine potential baseline differences (e.g., in gender, dietary restraint, time since last meal, BMI) between the four Suppression conditions and the two Restraint groups, respectively. If the Levene's Test for Equality of Variances was significant, results were based on analyses with equal variances not assumed. Second, an ANOVA with repeated measures design was executed to test whether the participants in the four conditions were able to suppress/monitor their thoughts as requested. Trial period (produced number of target thoughts per trial period) was entered as a three-level within-subjects factor, and Suppression condition (Suppression food/Suppression animals/Control food/Control animals) was entered as a four-level between-subjects factor. Planned comparisons were executed when appropriate.
Next, to test our main hypotheses regarding immediate and delayed memory for food cues, two subsequent ANOVAs were conducted. In the first ANOVA, the number of immediately recalled food cues was subjected to a 2 (Restraint group: high restrained eaters/low restrained eaters) x4 (Suppression condition: Suppression food/Suppression animals/Control food/Control animals) ANOVA. To test within-group differences, two follow-up paired sample t tests were executed, one within the high restraint group and one within the low restraint group, allowing comparison between remembrance of food versus animal cues within each restraint group. Bonferroni corrections were applied to correct for false positives (Type I errors) arising from multiple comparisons and, in this case, set the critical significance level at 0.025 (0.05/2; Howell 1997). When significant baseline differences between Suppression conditions and/or Restraint groups appeared, these variables were entered as covariates and the ANOVAs were re-executed as ANCOVAs, to eliminate the potential bias of these confounds (Field 2009). Planned comparisons were performed when appropriate. In the second ANOVA, the same 2 (Restraint group) x 4 (Suppression condition) analysis was executed but with the number of recalled food cues after delay, as dependent variable. Again, paired sample t tests with Bonferroni correction, ANCOVAs, and planned comparisons were executed when appropriate. Reported significance tests are all two-tailed.
Descriptive Statistics for Suppression Conditions and Restraint Groups
ANOVAs revealed no baseline differences between the four suppression conditions in terms of their levels of dietary restraint, time since last meal, BMI, age, amount of errors made during the pictorial sorting task, time spent sorting, and valence of animal cues or valence of food cues, all Fs <1.0, all [p.sup.s]>0.66. A chi square test revealed that there was no difference in sex between the conditions either, [[chi].sup.2](3)=2.58, p= 0.46. Comparison between the high and low restrained eaters, using independent sample t tests, revealed that the high restrained eaters scored significantly higher on dietary restraint and time spent on sorting the photographs. Additionally, there was a trend in the direction of higher BMI (p=0.10) and lower valence ratings for food cues (p=0.08) in the high restrained eaters (see Table 1). Chi square tests revealed no sex differences between restraint groups, [[chi].sup.2](1)=0.67, p=0.41. Not surprisingly, high restrained eaters indicated more frequently that they were currently on a diet and/or had been on a diet in the past, compared to the low restrained eaters, [[chi].sup.2](1)=29.00, p<0.001. Finally, time since last meal did not correlate with the number of immediately remembered food cues (r=--0.12, p= 0.34), nor with remembered food cues after delay (r=-0.04, p=0.76).
Table 1 Descriptive statistics and group differences for the high and low restrained eaters Statistic High Low restraint restraint (n=24) (n=24) M SD M SD t(68) Age 19.25 1.82 18.74 2,42 0.91 BMI 22.07 2.80 20.56 3.59 1.66*** Last meal (in 200.21 133.24 216.11 123.99 -0.50 minutes) Restraint 3.21 0.44 1.76 0.48 12.32* Valence animal 108.43 9.72 105.85 12.92 0.85 cues Valence food 105.04 16.78 111.74 14.19 -1.76*** cues Time sorting (in 85.04 21.47 98.26 24.16 -2.25** seconds) Errors sorting 0.08 0.28 074 0.85 -0.87 BMI Body Mass Index; last meal elapsed time since last meal (in minutes); restraint Dutch Eating Behaviour Questionnaire Restrained Eating scale; valence animal cues rated valence for animal cues seen in the pictorial sorting task; valence food cues rated valence for food cues seen in the pictorial sorting task; time sorting duration time of the pictorial sorting task (in seconds); errors sorting average amount of errors (misplacing) made during the pictorial sorting task * p[less than or equal to]0.001. ** p[less than or equal to]0.05. *** p[less than or equal to]0.10
Suppression Manipulation Check
A 3 (Trial period) x 4 (Suppression condition) ANOVA with repeated measures design was executed to test whether the participants in the four conditions were able to suppress/monitor their thoughts as requested. Results revealed a significant main effect of Suppression condition, F(3, 66)=3.74, p=0.01, and a significant interaction effect between Suppression condition and Trial period, F(6, 132)= 2.84, p=0.01, as was expected. Planned comparisons revealed that in the Suppression food condition, participants expressed significantly more thoughts about food during the first trial period (baseline; M=5.62, SD=4.50), when compared to the second trial period (suppression; M=3.24, SD=2.95), F(1, 66)=10.77, p<0.01. and the third trial period (suppression; M=3.05, SD=2.76), F(1, 66)=13.44, p<0.001. The number of food thoughts during the two suppression periods was comparable, F(1, 66)=0.10, p= 0.75. These results indicate that the participants were able to follow up on the suppression instructions. Also as expected, in the Control food condition, the number of food thoughts was comparable across trial periods, trial period 1 (M=3.50, SD=2.31) through trial period 2 (M=4.19, SD= 3.06), F(1, 66)=0.68, p=0.41; trial period 1 through trial period 3 (M=4.25, SD=3.17), F(1, 66)=0.87, p=0.35; trial period 2 through trial period 3, F(1, 66)=0.01, p=0.93.
Contrary to expectations, in the Suppression animals condition, results indicated that the number of mentioned animal thoughts was comparable across all trial periods, trial period 1 (M=2.29, SD=1.86) through trial period 2 (M=1.76, SD= 1.42), F(1, 66)=0.43, p=0.51; trial period 1 through trial period 3 (M=1.65, SD=1.37), F(1, 66)=0.69, p=0.41; trial period 2 through trial period 3, F(1, 66)=0.03, p=0.86, suggesting that the participants in this group failed to comply with the suppression instructions. Therefore a valid interpretation of the data in terms of thought suppression effects seems compromised in this particular condition. In the Control animals condition, there were no differences between trial periods, trial period 1 (M=2.69, SD=2.65) through trial period 2 (M=3.44, SD=2.80), F (1, 66)=0.81, p=0.37; trial period 1 through trial period 3 (M=2.81, SD= 2.37), F(1, 66)=0.02,p= 0.88; trial period 2 through trial period 3, F(1, 66)=0.86, p= 0.36, which was in line with expectations.
Results on Immediate Memory for Food Cues A 2 (Restraint group: high restrained eaters/low restrained eaters) x 4 (Suppression condition: Suppression food/Suppression animals/Control food/Control animals) ANOVA with the number of immediately remembered food cues in the first free recall task as dependent variable revealed a main effect of Restraint group, F(1, 62)=4.21, p=0.04, and Suppression condition, F(3, 62)=3.60, p=0.02, as was expected. The interaction between Suppression condition x Restraint group was not significant, F(3, 62)=1.59, p=0.20). (1)
First, a closer inspection of the main effect of Restraint group showed that the high restrained eaters remembered significantly more food cues (M=10.13, SD=2.80) than did the low restrained eaters (M=8.98, SD=2.05; see Fig. 1). There was no such difference between high and low restrained eaters in number of remembered animal cues, F(1, 62)=0.12, p=0.73. These findings were in line with our hypotheses. Follow-up paired sample t tests with Bonfcrroni correction showed that the low restrained eaters on average remembered more animal cues (M=10.09, SD=2.39) than food cues, t(45)=-2.84, p<0.01. In the high restrained eaters, there was no such ditTerence, with an equal amount of remembered food cues and animal cues (M=10.50, SD=2.02), t(23)=-0.72, p= 0.48.
[FIGURE 1 OMITTED]
Second, the main effect of Suppression condition was further examined. When the number of remembered food cues in the Suppression food condition was compared with the mean number of remembered food cues in the other three conditions (combined control group) using planned comparisons, results indicated a significant difference, F(1, 66)=6.40, p=0.01. After food-related thought suppression, participants on average remembered significantly more food cues (M= 10.43, SD=3.06) than in the control conditions (M=8.91, SD= 1.87). More detailed follow-up comparisons indicated that after food-related thought suppression, participants remembered more food cues than participants in the Control food condition (M=8.56, SD=2.06), F(1, 66)=5.96, p=0.02; and the Control animals condition (M=8.75, SD= 1.61), F(1, 66)=4.82, p=0.03. The difference between the Suppression food condition and the Suppression animals condition (M= 9.41, SD= 1.94) did not reach significance, F(1, 66)=1.83, p= 0.18.
Finally, a follow-up ANCOVA on remembered food cues controlling for the baseline differences between restraint groups (i.e., entering time spent sorting pictures, BMI, and valence of food cues as covariates) yielded results that were comparable to the first ANOVA, again revealing a main effect of Restraint group, F(1, 59)=4.31, p= 0.04, and Suppression condition, F(3, 59)=3.42, p=0.02, in the same directions.
Results on Delayed Memory for Food Cues
A 2 (Restraint group: high restrained eaters/low restrained eaters) x 4 (Suppression condition: Suppression food/Suppression animals/Control food/Control animals) ANOVA with the number of delayed remembered food cues in the second free recall task as dependent variable did not reveal any significant effects, all Fs<2.10, all ps> 0.15. Yet, when time spent sorting the pictures, BMI, and valence of food cues were included as covariates to control for baseline differences between groups in a follow-up ANCOVA, results did show a trend in the direction of a main effect of Restraint group, F(1, 59)=3.51, p=0.06. High restrained eaters on average remembered more food cues (M=9.41, SE=0.50) than did low restrained eaters (M=8.24, SE=0.37) (2) (see Fig. 1). There was no difference between high and low restrained eaters in number of remembered animal cues, F(1, 58)=0.01, p=0.91.
Follow-up paired t tests with Bonferroni correction showed that after the delay, the low restrained eaters on average remembered more animal cues (M=9.84, SD= 2.16) than food cues, 444)=-3.75, p=0.001. In the high restrained eaters, the average number of remembered food cues did not differ from the average number of remembered animal cues (M=9.59, SD= 1.86), t(23)=-1.05, p=0.31.
Building on cognitive theory for eating pathology (Williamson et al. 2004), the present study is the first to assess the existence of a pictorial food-related memory bias in high restrained eaters, both immediately and after a short delay. According to the leading cognitive theory for eating pathology, high restrained eaters show hypersensitivity for food cues, which should, inter alia, present itself as a memory bias for this type of cues (Williamson et al. 2004). An additional novel aim was to assess the potential role of food-related thought suppression on memory bias.
Our results provide at least partial evidence for a memory bias for pictorial food cues in high restrained eaters, both immediately and after a short delay. First of all, in line with expectations, high restrained eaters remembered significantly more food cues than did low restrained eaters immediately after a pictorial card sorting task. When controlling for baseline differences between high and low restrained eaters, this effect remained and there was also a trend for a (6-min) delayed memory bias for pictorial food cues in high restrained eaters. Also according to expectations, no such differences between high and low restrained eaters were found for the animal control cues.
Somewhat surprisingly, the high restrained eaters did not remember more food cues than animal cues. Yet, when looking at the low restrained eaters (who act as an unbiased control group regarding food cues), they on average remembered significantly fewer food cues than animal cues. This difference disappeared in the high restrained eaters. The observation that the high restrainers did not remember more food cues than animal cues is not in line with the results of Israeli and Stewart (2001). These authors found a better remembrance of food cues compared to animal cues in high restrained eaters. Yet, there is an important difference between the study by Israeli and Stewart (2001) and the present study, which may hinder comparison. In particular, Israeli and Stewart (2001) used food and animal woids as cues, whereas in the present study, pictorial cues were used (also see Stormark and Torkildsen 2004). Some studies suggest that animal pictures are in general visually more distinctive (more variety in types of animals) than food pictures and are therefore better remembered overall. In particular, Martin and colleagues (2010) established a better remembrance of animal pictures compared to food pictures, in a group of obese and normal-weight adults. This distinctiveness issue only seems applicable to pictorial stimuli and not to animal versus food words. Summarized, our results do seem to provide at least partial support for a memory bias for food cues in high restrained eaters, as predicted by cognitive theory (Williamson et al. 2004). To our knowledge, the present study was the first to illustrate this phenomenon using pictorial food stimuli. The current study was also the first to provide some evidence for a (6-min) delayed memory bias, suggesting a more prolonged influence of food-related schema activation on memory in high restrainers.
As a further point of interest and based on Wegner's theory (1994), the present study wanted to establish whether food-related thought suppression would result in a rebound effect for these thoughts, presenting itself as a memory bias. The analyses of the effects of suppression on memory processing yielded mixed results. Participants overall remembered significantly more food cues during the first recall task after suppression of food-related thoughts than did participants in the other conditions (combined). After the 6-min delay, no effects of food-related suppression on memory were found. This suggests a short-term rebound effect in the form of a food-related memory bias after food-related thought suppression. A closer inspection showed that more food cues were remembered after food-related thought suppression compared with merely monitoring food-related thoughts (without suppressing them) and compared with monitoring of animal-related thoughts. The difference with participants who had suppressed animal thoughts, however, did not reach significance. Yet, interpretation of the analyses in the Suppression animals condition seems compromised, since participants in this condition did not report a significant decrease in target thoughts during suppression, indicating that they were not able to comply with the suppression instructions. Since much care was taken in the administration of the suppression instructions (similar to earlier studies; see Abramowitz et al. 2001), even placing bogus electrodes to stress the importance of the instructions (see Lavy and van den Hout 1994), and since none of the participants indicated that they figured out the real intentions of the study, we have no reason to assume that participants intentionally did not follow the suppression instructions. Perhaps the number of animal thoughts at baseline was merely too low to begin with (despite prior priming) in the Suppression animals condition, thereby limiting a further decrease during suppression. Indeed, the number of target thoughts at baseline was the lowest of all conditions in the Suppression animals condition, potentially leaving too few thoughts to suppress. Perhaps in future studies a stronger priming of the target thought (e.g., by explicitly asking participants to think about the target thought out loud) prior to suppression/control instructions would resolve the low number of target thoughts at baseline. In sum, considering all findings, the present results seem to provide some support for a short but intensifying effect of food-related thought suppression on memory for food cues, but given the apparent problems of compliance with suppression instructions in the Suppression animals condition, further research is warranted.
As a final research interest, and given the finding that post-suppression rebound effects may be the strongest for thoughts with an emotional valence (e.g., Davies and Clark 1998), we wanted to examine whether high restrained eaters would show a stronger memory bias for food cues after food-related thought suppression, compared to low restrained eaters. In the present study, no evidence was found for such an interaction effect. High restrainers did not remember more food cues after food-related thought suppression than did low restrainers. In the present study, the subdivision between high and low restrained eaters was performed after the experiment had ended and was based on the norms of the DEBQ Restrained Eating scale (Van Strien et al. 1986a, b). Although a norm-based cutoff point has the advantage that it is not sample dependent (contrary to, e.g., a median split), this procedure led to an unequal number of high and low restrained eaters. As a result, the number of participants in some cells was small, thereby compromising the power and interpretation of the interaction effect, making it unwarranted to draw firm conclusions. For future research, it would be interesting to a priori select high and low restrained eaters, without raising suspicion as to why the experimenter wants to know this information (e.g., by including a measure of dietary restraint in a larger bundle of tests) and preferably using even more stringent cutoff scores for the high versus low restrained eaters (e.g., 25th and 75th percentile). With the norm-based subdivision between high and low restrainers in the present study, there was a trend for a higher BMI in the high restrained eaters. This supports Herman and Polivy's (1980) notion that people who attempt to restrain food intake are often prone to relapse.
The present study holds a number of limitations and directions for future research. First, the results do not delineate whether the observed bias is due to selective encoding, selective storage, selective retrieval, or biases at all stages of memory. It would, for example, be interesting for future studies to look into the depth of encoding as well as to include a link with attention processing in a single study, for example, by administrating a free recall task after an exogeneous cueing paradigm, which in its turn distinguishes between guiding attention toward or away from certain cues (Posner 1980). Second, recent neuroscientific insights show that in individuals with severe dietary restraint tendencies (i.e., individuals with anorexia nervosa), restraint of appetite is linked to structural brain changes, including a brain region related to working memory function (Brooks et al. 2011a, b). It would be interesting for future studies to address how this may impact on the emergence (or maintenance) of cognitive biases in high restrained eaters in the general population as well. Third, the current study only focused on memory for food cues. Future studies might also want to address memory processing for other schema-consistent information, such as body shape and weight in restrained eaters, using attention and memory paradigms. Furthermore, rather unexpectedly, in the present study, the high restrained eaters rated the food cues as having lower emotional valence than the low restrained eaters. The selected pictures were all of attractive (high caloric) foods, and participants were asked to give their explicit evaluation. It could well be that high restrained eaters, given their restraint status, responded more socially desirable, giving "lower" ratings to these unhealthy foods. Indeed, previous studies indicated that restrained eaters tend to implicitly show stronger positive associations with high caloric foods than do unrestrained eaters (Houbcn et al. 2010), but they may be reluctant to express these associations explicitly. Nevertheless, given that the results of the ANCOVAs in the present study (controlling for valence amongst other variables) were comparable to the initial ANOVAs, we believe that the established results were indeed content specific and not merely valence dependent and that they hence do not compromise our conclusions. In the same line, a recent study by Brooks and colleagues (2012) showed that subliminal, as opposed to consciously presented, food images can interfere with memory processes in individuals with anorexia nervosa. Likewise, a study by Dejonckheere et al. (2003) showed that suppression of thoughts about sweets led to paying more attention to these cues in a subsequent Stroop task, only when these cues were presented subliminally. Hence, for future research, it may be interesting not only to use cues that appeal to conscious, cognitively driven recognition but also to include subliminal presented cues that activate subcortical responses (Fazio and Olson 2003). Also, the present study included mostly females with a small number of males. Although the males were equally distributed across suppression conditions and restraint groups and gender-appropriate cutoff scores were used for dietary restraint, as a check, we tested the main hypotheses of this study for the female participants only and obtained results similar to those obtained for the full sample. The number of men in the present sample was too small to compare the experimental paradigm across men and women. However, we believe a gender comparison may be an interesting direction for future research. Finally, there is still some controversy in literature about the role of food deprivation on memory for food cues. While some studies find a correlation between hunger and recall of food words in individuals with bulimia nervosa (Hunt and Cooper 2001), suggesting that enhanced recall for food-related information may be state dependent, other studies with individuals with anorexia nervosa indicate that cognitive processing of food stimuli does not depend on food deprivation (Pietrowsky et al. 2002). In the present study, participants were asked about the time since their last meal, which did not correlate with the number of remembered food cues (immediately and after delay). Yet, it would be interesting for future research to include additional measures of food deprivation (e.g., content of last meal, time until next meal, how much participants think they could eat).
Regarding the clinical ramifications, the established memory bias for high caloric food cues in high restrained eaters seems to reflect a predisposition in the cognitive processing of food-related information. Dietitians should be aware that such preconceived notions may hinder the attempts of restrained eaters to lose weight. When your thoughts are "in the refrigerator all day", this may place an extra burden on self-control attempts (Wegner 1989). Moreover, thought suppression does not seem to be the best way to deal with unwanted food thoughts.
In conclusion, this study is the first to provide support for a (delayed) memory bias for food pictures in high restrained eaters, reflecting selective remembrance of "forbidden" food information, relative to low restrained eaters. As such, this study provides at least partial support for the existence of food-related schemas in high restrained eaters, as is the case in eating-disordered and obese individuals. There was also some evidence for a brief but enhancing effect of food-related thought suppression on memory for food cues. We hence believe the present findings provide a promising starting point for future research.
Published online: 20 March 201
[C] Association of Behavior Analysis International 2014
Abramowitz, J. S., Tolin, D. F., & Street, G. P. (2001). Paradoxical effects of thought suppression: a meta-analysis of controlled studies. Clinical Psychology Review 21, 683-703. doi:10.1016/S0272-7358(00)00057-X.
Banasiak, S. J., Wertheim, E. H., Koerner, J., & Voudouris, N. J. (2001). Test-retest reliability and internal consistency of a variety of measures of dietary restraint and body concerns in a sample of adolescent girls. International Journal of Eating Disorders, 29, 85-89. doi :10.1002/1098- 08X(200101)29:1<85::AID-EAT14>3.3.00;2-7.
Boon, B., Vogelzang, L., & Jansen, A. (2000). Do restrained eaters show attention toward or away from food, shape and weight stimuli? European Eating Disorders Review 8, 51-58. doi:10.1002/(SICI) 1099-0968(200002)8:1<51::AID-ERV306>3.0. CO;2-E.
Brooks, S. J., Barker, G. J., O'Daly, 0. G., Brammer, M., Williams, S. C. R., Benedict, C., et al. (2011a). Restraint of appetite and reduced regional brain volumes in anorexia nervosa: a voxel-based morphometric study. BMC Psychiatry, 11, 179. doi:10.1186/1471-244X-11-179.
Brooks, S. J., Prince, A., Stahl, D., Campbell, I. C., & Treasure, J. (2011b). A systematic review and meta-analysis of cognitive bias to food stimuli in people with disordered eating behaviour. Clinical Psychology Review 31,37-5 I. doi:10.1016/j.cpr.2010.09.006.
Brooks, S. J., O'Daly, 0. G., Uher, R., Schioth, H. B., Treasure, J., & Campbell, I. C. (2012). Subliminal food images compromise superior working memory performance in women with restricting anorexia nervosa. Consciousness and Cognition, 21, 751-763. doi:10.1016/j.cpr.2012.02.006.
Cooper, M. J., & Fairburn, C. G. (1992). Selective processing of eating, weight and shape related words in patients with eating disorders and dieters. British Journal of Clinical Psychology 3, 363-365. doi:10.1016/j.2044-8260.1992.tb01007.x.
Daee, A., Robinson, P., Lawson, M., Turpin, J. A., Gregory, B., & Tobias, J. D. (2002). Psychologic and physiologic effects of dieting in adolescents. Southern Medical Journal, 95, 1032-1041. doi:10.1097/00007611-200295090-00016.
Davies, M. I., & Clark, D. M. (1998). Thought suppression produces a rebound effect with analogue post-traumatic intrusions. Behaviour Researrh and Therapy, 36, 571-582. doi :10.1016/S0005-7967(98) 00051-5.
Dejonckheere, P. J. N., Braet, C., & Soetens, B. (2003). Effects of thought suppression on subliminally and supraliminally presented food-related stimuli. Behaviour Change, 20, 223-230. doi:10.1375/bech.220.127.116.11379.
Fazio, R. H., & Olson, M. A. (2003). Implicit measures in social cognition research: their meaning and use. Annual Review of Psychology 54, 297-327. doi:10.1146/annurev.psych.54.101601.145225.
Field, A. (2009). Discovering statistics using SPSS (3rd ed.). Thousand Oaks: Sage.
Herman, C. P., & Mack, D. (1975). Restrained and unrestrained eating. Journal of Personality, 43, 647-660. doi:10.1111/j.1467-6494. 1975.tb00727.x.
Herman, C. P., & Polivy, J. (1980). Restrained eating. In E. Stunkard (Ed.), Obesity (pp. 208-225). Philadelphia: Saunders.
Hoffineister, K., Teige-Mocigemba, S., Blechert, J., Klauer, K. C., & Tuschen-Caffier, B. (2010). Is implicit self-esteem linked to shape and weight concerns in restrained and unrestrained eaters? Journal of Behavior Therapy and Experimental Psychiatry 41, 31-38. doi: 10.1016/j.jbtep.2009.08.009.
Houben, K., Roefs, A., & Jansen, A. (2010). Guilty pleasures. Implicit preferences for high calorie food in restrained eating. Appetite, 55, 18-24. doi:10.1016/j.appet.2010.03.003.
Howell, D. C. (1997). Statistical methods in psychology (4th ed.). Belmont: Duxbury Press.
Hunt, J., & Cooper, M. (2001). Selective memory bias in women with bulimia nervosa and women with depression. Behavioural and Cognitive Psychotherapy, 29, 93-102. doi:10.1017/S1352465801001102.
Israeli, A. L., & Stewart, S. H. (2001). Memory bias for forbidden food cues in restrained eaters. Cognitive Therapy and Research, 25, 3747. doi:10.1023/A:1026422731313.
Kircanski, K., Craske, M. G., & Bjork, R. A. (2008). Thought suppression enhances memory bias for threat material. Behaviour Research and Therapy, 46, 462-476. doi:10.1016/j.brat2008.01.009.
Lavy, E. H., & van den Flout, M. A. (1994). Cognitive avoidance and attentional bias: causal relationships. Cognitive Therapy and Research, 18, 179-191. doi:10.1007/BF02357223.
Legenbauer, T., Maul, B., Ruhl, I., Kleinstauber, M., & Hiller, W. (2010). Memory bias for schema-relevant stimuli in individuals with bulimia nervosa. Journal of Clinical Psychology. 66, 302-316. doi:10.1002/jclp.20651.
Martin, L. E., Holsen, L. M., Chambers, R. J., Bruce, A. S., Brooks, W. M.,.Zarcone, J. R., et al. (2010). Neural mechanisms associated with food motivation in obese and healthy weight adults. Obesity 18, 254-260.
Meier, B., Konig, A., Parak, S., & Henke, K. (2011). Suppressed, but not forgotten. Swiss Journal of Psychology 70, 5-11. doi:10.1024/1421-0185/a000033.
Mizes, J. S, & Christiano, B. A. (1-995). Assessment of cognitive variables relevant to cognitive-behavioral perspectives on anorexia-nervosa and bulimia-nervosa. Behaviour Research and Therapy, 33, 95-105. doi:10.1016/0005-7967(94)E0032-E.
Mogg, K., Bradley, B. P., Hyare, H., & Lee, S. (1998). Selective attention to food-related stimuli in hunger: are attention biases specific to emotional and psychopathological states, or are they also found in normal drive states? Behaviour Research and Therapy 36, 227237. doi:10.1016/S0005-7967(97)00062-4.
Morris, M. E., Goldsmith, T. E., Roll, S., & Smith, J. E. (2001). A model for assessing the self-schemas of restrained eaters. Cognitive Therapy and Research, 25, 201-214. doi:10.1023/A:1026472009449.
Neisser, U. (1976). Cognition and reality: Principles and implications of cognitive psychology. San Francisco: WH Freeman.
Nikendei, C., Weisbrod, M., Schild, S., Bender, S., Walther, S., Herzog, W., et al. (2008). Anorexia nervosa: selective processing of food-related word and pictorial stimuli in recognition and free recall tests. International Journal of Eating Disorders, 5, 439-447. doi:10.1002/eat.20518.
Nixon, R. D. V., Cain, N., Nehmy, T., & Seymour, M. (2009). The influence of thought suppression and cognitive load on intrusions and memory processes following an analogue stressor. Behavior Therapy 40, 368-379. doi:10.1016/j.beth.2008.10.004.
Overduin, J., Jansen, A., & Louwerse, E. (1995). Stroop interference and food intake. International Journal of Eating Disorders, 18, 277285. doi:10.1002/1098-108X(199511)18:3<277::AID-EAT2260180310>3.0.00;2-3.
Paunovic, N., Lundh, L. G., & Ost, L. G. (2002). Attentional and memory bias for emotional information in crime victims with acute posttraumatic stress disorder (PTSD). Journal of Anxiety Disorders, 16, 675-692. doi:10.1016/S0887-6185(02)00136-6.
Pietrowsky, R., Krug, R., Fehm, H. L., & Born, J. (2002). Food deprivation fails to affect preoccupation with thoughts of food in anorectic patients. British Journal of clinical Psychology, 41, 321-326. doi: 10.1348/014466502760379172.
Posner, M. L (1980). Orientation of attention. Quarterly Journal of Experimental Psychology, 32, 3-25. doi: 10.1080/00335558008248231.
Soetens, B., & Bract, C. (2007). Information processing of food cues in overweight and normal weight adolescents. British Journal of Health Psychology, 12,285-304. doi:10.1348/135910706X107604.
Soetcns, B., Braet, C., Dejoncklieerc, P., & Roets, A. (2006). When suppression backfires: the ironic effects of suppressing eating-related thoughts. Journal of Health Psychology; 11, 655-668. doi: 10.1177/1359105306066615.
Stice, E., & Agras, W. S. (1998). Predicting onset and cessation of bulimic behaviours during adolescence: a longitudinal grouping analysis. Behavior Therapy 29, 257-276. doi: 10.1016/S 00057894(98)80006-3.
Stice, E., Marti, C. N., & Durant, S. (2011). Risk factors for onset of eating disorders: evidence of multiple risk pathways from a 8-year prospective study. Behaviour Reseatch and Therapy 49, 622-627. doi:10.1016/j.bmt.2011.06.009.
Stonnark, K. M., & Torkildsen, 0. (2004). Selective processing of linguistic and pictorial food stimuli in females with anorexia and bulimia nervosa. Eating Behaviors, 5, 27-33. doi:10.1016/j.eatbeh. 2003.07.002.
Tekcan, A. I., Tas, A. C., Topcuoglu, V, & Yucel, B. (2008). Memory bias in anorexia nervosa: evidence from directed forgetting. Journal of Behavior Therapy and Experimental Psychiato.; 39, 369-380. doi:10.1016/j.jbtep.2007.09.005.
Van Strien, T., Frijters, J.E.R., Bergers, G.P.A., & Defares, P.B. (1986a). Nederlandse Vragenlijst voor Eetgedrag (NVE), handleiding [Dutch Eating Behavior Questionnaire (DEBQ) manual]. Lisse, the Netherlands: Swets & Zeitlinger b.v.
Van Strien, T., Frijters, J. E. R., Bergers, G. P. A., & Defares, P. B. (1986b). The Dutch Eating Behavior Questionnaire (DEBQ) for assessment of restrained, emotional and external eating behavior. International Journal qf Eating Disorders, 5, 295-315. doi:10.1002/1098-108X(198602)5:2<295::AID-EAT2260050209>3.0.00;2-T.
Vitousek, K. B., & Hollon, S. D. (1990). The investigation of schema content and processing in eating disorders. cognitive Therapy and Research, 14, 191-214. doi:10.1007/BF01176209.
Wegner, D. M. (1989). White bears and other unwanted thoughts: Suppression, obsession, and the psychology of mental contral. New York: The Guilford Press.
Wegner, D. M. (1994). Ironic processes of mental control. Psychological Review 10, 34-52. doi:10.1037/0033-295X.101.1.34.
Wegner, D. M., Schneider. D. J., Carter, S. R., & White, T. L. (1987). Paradoxical effects of thought suppression. Journal of Personality and Social Psychology 53,5-13. doi:10.1037//0022-3518.104.22.168.
Whisman, M. A., & McClelland, G. H. (2005). Designing, testing and interpreting interactions and moderator effects in family research. Journal of Family Psychology 19, 111-120. doi:10.1037/0893-322.214.171.124.
Williamson, D. A., White, M. A., York-Crowe, E., & Stewart, T. M. (2004). Cognitive-behavioral theories of eating disorders. Behavior Modification, 28,711-738. doi:10.1177/0145445503259853.
(1.) We conducted the same ANOVA including only the female participants and found similar, yet less strong effects (given the reduction in statistical power). There was a trend for a main effect of Restraint group, F(1,53)= 2.88,p=0.09, and for Suppression condition, F(3, 53)=2.64, p=0.06), in the same directions.
(2.) The same ANCOVA including only the female participants yielded similar results, namely a trend for a main effect for Restraint group in the same direction, F(1, 50)=2.97, p=0.09.
B. Soetens (*)
Department of Applied Psychology, Thomas More University College, Jozef De Bomstraat 11, 2018 Antwerp, Belgium
B. Soetens * F. Raes
Centre for the Psychology of Learning and Experimental Psychopathology, KU Leuven, Leuven, Belgium
Department of Developmental, Personality and Social Psychology, Ghent University, Ghent, Belgium
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||ORIGINAL ARTICLE|
|Author:||Soetens, Barbara; Roets, Arne; Raes, Filip|
|Publication:||The Psychological Record|
|Date:||Jan 1, 2014|
|Previous Article:||Effects of an Online Stimulus Equivalence Teaching Procedure on Research Design Open-Ended Questions Performance of International Undergraduate...|
|Next Article:||What is the role of tone in the phonological similarity effect?|