*3.2. Results*

As presented in Table 1, there were no differences between the two groups of restrained eaters in age, BMI, and DEBQ-R. Mean reaction time (RT) in the FVCT was calculated for each participant. The food–valence association effect was calculated for each participant in each valence (negative vs. neutral) and time (pre- vs. post-training) condition.

To ensure proper task engagement, we calculated two measures for the F-SST task: the nsRT (RT in no-stop-signal trials) and the nsACC (accuracy in no-stop-signal trials). There were no significant differences between the groups in both nsRT (food-response training: mean = 592, food-response/inhibition training: mean = 583; *t*(45) = 0.33, *p* = 0.740) and nsACC (food-response training: mean = 0.93, food-response/inhibition training: mean = 0.94; *t*(45) = 1.40, *p* = 0.166), indicating similar task engagement in both groups.

In order to validate our measurement of implicit association toward food (food–valence association effect), we tested whether this effect at baseline (prior to practice) correlated with the DEBQ-R scores. To that end, we subtracted the food–valence association effect for the negative valence from the food–valence association effect for the positive valence (i.e., a more positive attitude toward food or a less negative attitude toward food will result in a larger estimate). Results yielded a significant correlation between this estimate and DEBQ-R scores at baseline (*r*(46) = -0.31, *p* = 0.035), indicating that highly restrained eaters will exhibited more negative and less positive attitudes toward food.

A three-way mixed model ANOVA was carried out on the food–valence association effect, with group (food-response vs. food-response/inhibition training) as a between-subject factor, and valence (negative vs. positive) and time (pre-training vs. post-training) as within-subject factors (Figure 4 and Table 2). The results revealed a significant main effect for valence, (*F*(1, 45) = 58.57, *p* < 0.001, ηp <sup>2</sup> = 0.57), indicating a larger food–valence association effect for the positive compared to negative valence condition. There was also a significant main effect for time (*F*(1, 45) = 5.40, *p* = 0.025, ηp <sup>2</sup> = 0.11), indicating a larger food–valence association effect pre-training (mean effect = <sup>−</sup>7.3, SD = 34.3) compared to post-training (mean effect = 6.0, SD = 22.9). The main effect for group was not significant (*F*(1, 45) = 2.58, *p* = 0.116). Importantly, the three-way interaction between group, valence, and time was significant, (*F*(1, 45) = 5.12, *p* = 0.029, η<sup>p</sup> <sup>2</sup> = 0.10). To further investigate this interaction, planned comparisons were carried out to examine the effects of time and group for each valence condition separately. In the positive associations condition, the results showed a significant increase in the food–valence association effect (i.e., an increase in implicit positive attitudes toward food) pre- to post-training in the food-response/inhibition training group (*F*(1,44) = 13.052, *p* < 0.001, η<sup>p</sup> <sup>2</sup> = 0.22), but not in the food-response training group (*F*(1,44) = 1.094, *p* = 0.301, η<sup>p</sup> <sup>2</sup> = 0.02). In the negative association condition, there were no pre to post differences in the food–valence association effect in the food-response/inhibition group (*F*(1,44) = 2.257, *p* = 0.14, η<sup>p</sup> <sup>2</sup> = 0.04), nor in the food-response training group (*F*(1,44) = 1.21, *p* = 0.276, η<sup>p</sup> <sup>2</sup> = 0.02) (Figure 4 and Table 2).


**Table 2.** Results of Experiment 2—food–valence compatibility task (FVCT).

Note: Mean reaction time, (SD), and [accuracy] for the different conditions of Experiment 2.

**Figure 4.** Differences in the food–valence association effect as a function of group, time, and valence. The food–valence association effect was calculated as mean response time (RT) in the non-food condition minus that in the food conditions. On the left panel (positive-valence condition) higher scores indicate greater positive associations whilst on the right panel (negative-valence condition) higher scores indicate greater negative associations. Error bars represent 1 standard error from the mean. \* indicates *p* < 0.05.

#### *3.3. Discussion Experiment 2*

The results of Experiment 2 revealed an increase in positive implicit attitudes toward palatable foods following the food-response/inhibition training, but not following the food-response training. There were no differences between the training groups in negative implicit attitudes toward food. Previous studies have shown that inhibiting motor responses while being exposed to food stimuli leads to a food devaluation effect. Specifically, inhibited food stimuli were rated as less attractive following such training procedures [23,24]. Our results showed that balanced response inhibition training can increase, rather than reduce, positive implicit associations regarding food stimuli among female restrained eaters. This is especially important considering that restrained eaters already hold negative attitudes toward food [29]. Therefore, finding ways to increase positive attitudes toward food among these individuals, while retaining self-control, is clinically important.

#### **4. Discussion**

The current study compared two response inhibition trainings on food consumption, food-related anxiety (Experiment 1), and implicit attitudes toward food (Experiment 2) among female restrained eaters. Results yielded that a food-response/inhibition training that balanced the requirement to inhibit and respond to food and non-food stimuli reduced food consumption and increased positive implicit attitudes toward food. On the other hand, a food-response training procedure that encouraged a response to food stimuli and inhibition to non-food stimuli increased food-related anxiety and had no influence on implicit attitudes toward food.

The current study contributes to the existing literature in several important ways. First, response inhibition trainings are currently being investigated as means for influencing food consumption. However, most studies form consistent associations between food and stopping a response [19,20]. Besides reduction of food consumption, such training procedures also elicit negative attitudes toward food among participants [22–24]. In contrast to those studies, the present study showed that a response inhibition training procedure that balances the requirement to stop a response to food and non-food stimuli reduces food consumption but also improves implicit attitudes toward food. This finding is clinically important because therapeutically, we would like to help restrained eaters achieve more flexible eating behaviors while retaining positive attitudes toward food (i.e., not experiencing food

as a threat). Second, previous studies have shown that restrained eaters have a general deficit in response inhibition to non-food stimuli [16–18] and exert more inhibitory resources in order to inhibit their response to food stimuli compared to non-restrained eaters [18,31]. It has been postulated that in the long run, the exhaustion of inhibitory resources may subsequently lead to disinhibited eating behaviors such as overeating [3]. These findings suggest that overeating among restrained eaters may be the result of a lack of balance between over-activation and under-activation of inhibitory reactions to food and non-food stimuli. The present study supports this theory by showing that self-control over eating may be achieved by training restrained eaters to balance response inhibition and execution between food and non-food stimuli. Our results indicate that such balancing may improve restrained eaters' self-control in the presence of food.

In a broader perceptive, the current study adds to the mounting evidence suggesting that response inhibition is a modular process that can be trained and result in real-life behavioral changes. Specifically, individuals differ in their ability to use response inhibition as was previously indicated in behavioral and imaging studies [32–34]. Individual differences in response inhibition mean that some individuals experience marked difficulties recruiting inhibitory resources in everyday situations. As noted earlier, general response inhibition failures have been documented among restrained eaters, which may result in difficulty engaging in self-control in the presence of palatable foods [16,18]. Nevertheless, as the current study and others demonstrate, the modular nature of response inhibition allows training an individual to stop an automatic response in the presence of specific environmental cues. Indeed, previous studies have shown that conditioning response inhibition to activate during exposure to specific-environmental cues such as beer, cigarettes, and chocolate in lab-based experiments can subsequently alter how the trained individual behave when presented with these type of cues in real-life situations [35–37].

Increasing the knowledge on how response inhibition interacts with environmental cues and how it can be trained may have implications on various psychological disorders that are characterized by stimulus-driven behaviors or impulse-control problems. In fact, recent studies have provided preliminary reports that such conditioning can be beneficial in augmenting treatments for clinical populations (e.g., [38]). For example, in a recent study, a modified version of the stop-signal task was used to condition automatic inhibition in treatment of refractory patients with obsessive–compulsive disorder [21]. This study demonstrated that associating disorder-specific stimuli with stopping can not only change behaviors but also reduce unwanted intrusive negative cognitions. Similarly, the training procedures used in the current study, not only influenced food intake but also had an effect on food-related anxiety and attitudes toward food. Other studies have also shown that response inhibition can result in attitudinal changes regarding food cues that were associated with stopping a response [22–24]. Taken together, converging evidence suggest that response inhibition trainings can act to regulate emotions, thoughts, and behaviors.

With respect to clinical implications, although restrained eating is not considered a psychiatric disorder, it is associated with elevated levels of depression and anxiety [4]. Moreover, studies have reported that dietary restraint is a proxy for developing eating disorders such as binge eating disorder and bulimia nervosa (for review see [39]). Restrained eating is also associated with weight gain and obesity [40]. Therefore, there is great importance in identifying ways to improve restrained eaters' control over eating while retaining positive attitudes toward food. Response inhibition training such as that tested in the current study may, in the future, show promise as a means to regulate disordered eating patterns among individuals who are at high risk for developing eating disorders. Nevertheless, the field of cognitive training using appetitive food cues is relatively new. There are still many questions that require answers before such training can be offered as clinical interventions. For example, it is still not clear whether response inhibition trainings using appetitive cues have a lasting effect. Additionally, the exact amount of training sufficient for eliciting long-term effects is yet to be determined. Lastly, most response-inhibition trainings using appetitive food cues focus on changing eating behaviors. As such, there is still much to learn regarding the impact of such trainings on emotional and attitudinal

factors that can be clinically meaningful and influence one's emotional experience during exposure to various foods. In the future, it would also be interesting to assess such training procedures as potential add-on treatments for eating disorders in which over- and under-activation of response inhibition in the presence of food represent core clinical symptoms of the disorders such as self-starvation in anorexia nervosa and binge eating in bulimia nervosa and binge eating disorder [8,13].

Several limitations of the current study should be addressed. Experiment 1 did not include a baseline measurement of food consumption so that the purpose of the training would not be revealed. However, a lack of baseline food consumption measurement makes it difficult to determine whether the difference found in snack consumption between the training groups is because of a reduction of food intake in the food-response/inhibition group or an increase in food intake in the food-response training group. Future studies should include a baseline measurement of food intake or include a third control group that does not perform any training. Nevertheless, the results showed that there were no baseline differences between the groups in hunger level. A second limitation is that Experiment 2 was run online using a modest sample. Our initial plan was to replicate the results on food consumption and add the implicit attitude measures, but due to COVID-19, we could not conduct a lab-based experiment with a taste test. Thus, future studies will need to replicate the results with a larger sample in order to affirm the beneficial role of the food-response/inhibition task on food consumption and implicit attitudes toward food. Finally, the DEBQ-R does not have a standardized threshold for defining high-restrained eating. Therefore, it could be that other thresholds than that used in the current study would yield different results. However, it is important to note that the same threshold was used in both training groups as the study only tested high-restrained eaters.

To conclude, the current study revealed that response inhibition training that balances the requirement to stop a response to food and non-food stimuli can reduce food consumption and improve positive attitudes toward food among restrained eaters. This study adds to the existing knowledge regarding how eating behaviors can be modulated using cognitive training procedures that target neurocognitive mechanisms suggested to underlie disordered eating. Future studies should investigate the utility of such training procedures as intervention programs with a goal to achieving long-term effects on eating-related thoughts, emotions, and behaviors.

**Author Contributions:** Conceptualization, E.K. (Eyal Kalanthroff), and N.W; data collection, E.K. (Eldad Keha), and H.L.; formal analysis, E.K. (Eyal Kalanthroff), E.K. (Eldad Keha), and H.L.; writing—original draft preparation, N.W., E.K. (Eyal Kalanthroff), E.K. (Eldad Keha), and H.L.; writing—review and editing, N.W. and E.K. (Eyal Kalanthroff). All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Israel Science Foundation (1341/18, (PI: Eyal Kalanthroff and the National Institute for Psychobiology, Israel (215-17-18b, (PI: Eyal Kalanthroff)).

**Acknowledgments:** We thank Hadar Naftalovich for her useful input on this article. We thank Ashely Ackerling and Noa Bar for their assistance in data collection.

**Conflicts of Interest:** The authors declare no conflict of interest.
