Could implementation intentions improve the efficacy of behavioral weight-loss treatment?

Although behavioral weight loss treatment can result in health benefits, the effects on weight loss are generally disappointing, with lost weight regained within a few years. Because these problems appear to be due to a failure to execute the strong weight control intentions formed during treatment, the use of implementation intentions is suggested to bridge that intention-behavior gap. In contrast to behavioral intentions, implementation intentions specify the situation in which the behavior should be enacted. Although implementation intentions have generally been very effective in increasing the likelihood of goal attainment for a wide range of behaviors, results with regard to eating have been mixed. Especially disappointing have been the findings of a large study that added implementation intentions to a clinical weight loss treatment and found no effects either on weight loss or on weight loss maintenance (Kn ¨ auper et al., 2018). I argue that this failure is due to the use of implementation intentions that target specific eating behaviors. According to the goal conflict model of eating, implementation intentions are most effective if they prime the weight control goal; implementation intentions targeting specific behaviors do not reliably do that. I review evidence that the implementation intention to think of dieting when confronted with palatable food items primes weight control thoughts. It also reduces eating of primed foods and even result in weight loss.

Although behavioral weight loss treatment can result in health benefits, the effects on weight loss are generally disappointing, with lost weight regained within a few years. Because these problems appear to be due to a failure to execute the strong weight control intentions formed during treatment, the use of implementation intentions is suggested to bridge that intention-behavior gap. In contrast to behavioral intentions, implementation intentions specify the situation in which the behavior should be enacted. Although implementation intentions have generally been very effective in increasing the likelihood of goal attainment for a wide range of behaviors, results with regard to eating have been mixed. Especially disappointing have been the findings of a large study that added implementation intentions to a clinical weight loss treatment and found no effects either on weight loss or on weight loss maintenance (Knäuper et al., 2018). I argue that this failure is due to the use of implementation intentions that target specific eating behaviors. According to the goal conflict model of eating, implementation intentions are most effective if they prime the weight control goal; implementation intentions targeting specific behaviors do not reliably do that. I review evidence that the implementation intention to think of dieting when confronted with palatable food items primes weight control thoughts. It also reduces eating of primed foods and even result in weight loss.
Although behavioral treatment of individuals with overweight and obesity can result in important health outcomes (e.g., the reduction of the risk to develop diabetes among high-risk individuals), the lack of substantial and persistent weight loss is rather disappointing. According to a recent review of high quality follow-up studies of prospective weight loss trials, average weight loss at the end of treatment was 7.4% of bodyweight, but patients regained weight in the follow-up phase at the rate of 0.14% per month and reached pre-intervention weight at roughly 4.1 years after the intervention (Nordmo et al., 2020). Earlier reviews also concluded that behavioral programs can result in the loss of 8-10% of bodyweight during the first six to eight months of treatment (Look AHEAD Research Group, 2006;Hall & Kahan, 2018;The Look Ahead Research;Stroebe, 2023), but that most patients regain the lost weight within five years after the end of their treatment (Jeffery et al., 2000;Mann et al., 2007).
Treatment methods focus mainly on influencing people's intentions to change their behavior. Although intentions are a proximal determinant of behavior, research on the effect of intention change on behavior indicates that this effect is disappointingly small (McDermott et al., 2016;Rhodes & Dickau, 2012;Webb & Sheeran, 2006). An additional problem with eating behavior is the tempting nature of many high caloric food items. After discussing the goal conflict model of eating to explain why dieting intentions are derailed by exposure to tempting food, I suggest the use of a special form of implementation intentions as a way to strengthen the intention-behavior relationship.

The goal-conflict model of eating
Representative surveys in Western industrialized countries suggest that nearly half the population is trying to lose weight. For example, a large US sample of adults indicated that 42.2% were trying to lose weight in 2015-2016 (Han et al., 2019). Similarly, based on the Health Survey for England, Piernas et al., 2016 reported that 47% of respondents reported dieting attempts. Finally, based on a representative survey of Dutch men and women, De Ridder et al. (2014) found that 63.2% of men and 62.7% of women qualified as restrained eaters according to sex-specific norm scores for the Dutch Eating Behavior Questionnaire (van Strien et al., 1986). Restraint scores were strongly correlated with weight concerns (e.g., I am concerned about my weight and how it will affect my appearance).
The goal conflict model of eating offers a theoretical explanation why chronic dieting (or restrained eating to use the technical term) does not reliably result in weight control or even weight loss (Stroebe, 2018(Stroebe, , 2022(Stroebe, , 2023Stroebe et al., 2013). According to this model the difficulty of chronic dieters to control their eating is due to a conflict between two attractivebut often incompatiblegoals: On the one hand, they would like to lose weight or at least not gain weighttheir weight control goal. But at the same time, they also very much enjoy eating food they liketheir eating enjoyment goal. For people, who love salads or grilled fish, these goals are not incompatible. However, for people, who enjoy deep-fried fish and chips, followed by a dessert of ice cream, the two goals are incompatible. The weight control goal is typically the dominant goal for restrained eaters. When they are in environments that do not contain palatable food stimuli (e.g., their office, place of work), they have no difficulty pursuing their weight control goal. However, once they are exposed to stimuli that reflect or promise palatable food (e.g., smell of cooking, window of a bakery shop displaying cakes and cookies), their eating enjoyment goal gets primed and they feel an urge to eat palatable food.
There is a great deal of empirical evidence supporting this model (for reviews, see, Stroebe, 2018;2023, Stroebe et al., 2013. Whereas restrained eating is assessed with a questionnaire, we had to use social cognition methods to assess the goal conflict, because questions assessing either of the two goals would act as primes that increase the cognitive accessibility of the goal mentioned. For example, asking restrained eaters, who are sitting in front of an ice cream anticipating eating enjoyment about their dieting plans would prime their dieting goal and push it into dominance. For this reason, questionnaires are best suited to measure a static situation, whereas goal conflicts are fluid. Therefore, questionnaire methods are not suitable for studying goal conflicts. Fortunately, cognitive psychologists have developed methods that circumvent these problems. Our studies were conducted in an experimental laboratory, where students sat in individual cubicles. Both stimulus presentation and measurement of responses was done via computers. When we began our research project, we tested two key predictions of the model, namely (1) that exposure to palatable food stimulates eating enjoyment in restrained eaters and (2) that stimulation of the eating enjoyment goal will result in a suppression of the eating control goal. We used the Concern for Dieting subscale of the Restraint Scale (RS) of Herman and Polivy (1980) to measure eating restraint in all our studies. In contrast to the Weight-Fluctuation subscale of the RS, which assesses the consequences of a failure of weight control, the Concern for Dieting subscale consists mainly of questions referring to eating control. It can be considered a measure of weight concern and chronic dieting.
We used the probe-recognition task of McKoon and Ratcliff (1986) to test the first prediction (Papies et al., 2007;Experiment 1). With this task, participants are presented with sentences followed by a probe word. They are asked to decide whether the probe word was part of the sentence. Our experimental sentences described a person eating some tasty food (e.g., Bill is eating a big piece of apple pie). The sentence was then followed by a hedonic probe word (e.g., delicious, tasty) that was not part of the sentence. Participants had to decide as fast as possible whether the probe word had been part of the sentence. In control sentences, Bill was eating something bland (e.g., a piece of rye bread). Our hypothesis was that food words, referring to very tasty food (e.g., apple pie) would elicit in restrained (but not in unrestrained) eaters, thoughts about how tasty such food would be. They would therefore take somewhat longer than unrestrained eaters to decide that the word "tasty" was actually not part of that sentence. In support of our hypothesis, restrained eaters took longer to decide that a sentence did not contain the word "tasty" in sentences that contained palatable rather than non-palatable food. In contrast, the palatability of the food a person was described as eating did not influence the reaction times of unrestrained eaters.
The (second) hypothesis that priming individuals with tasty food words would stimulate the eating enjoyment goal in restrained eaters and as a result lead to the suppression of weight control thoughts was tested in an experiment, where participants were subliminally primed with tasty food words (e.g., chocolate, ice cream) and the cognitive accessibility of dieting concepts was measured with a lexical decision task (Stroebe et al., 2008). We predicted that in restrained eaters, tasty food primes would increase the cognitive accessibility of the eating enjoyment goal and as a result, decrease the cognitive accessibility of eating control thoughts. With a lexical decision task respondents are presented either with words related to weight control (e.g., slimming, diet) or with non-word (nonsense syllables) and have to decide as fast as possible whether the presented word was a proper word. Earlier research had indicated that the time it would take participants to recognize the behavioral concepts would reflect cognitive accessibility (e.g., Aarts & Dijksterhuis, 2000. Eating restraint was again measured with the concern for dieting subscale of the RS (Herman & Polivy, 1980). We predicted that priming with palatable food words would decrease the cognitive accessibility of eating control thoughts in restrained, but not in unrestrained eaters. Our results supported this prediction.
These findings have also been replicated in actual eating situations, where the primes were the sight and smell of palatable food. Although increases in the accessibility of eating enjoyment and the suppression of weight control cannot be measured directly in such situations, they can be inferred from the amount of food eaten. Several such studies have been reported (Jansen & van den Hout, 1991;Fedoroff et al., 1997;Rogers & Hill;1989).
The earliest study was conducted by Rogers and Hill (1989). The experiment consisted of two parts. In the first part, participants in their experimental condition were presented with palatable food items (actual and color photographs) and asked to imagine, how this food would taste. The control group was not exposed to food and instead filled out several rating scales. In the second part of the study all participants were led into a room with a display of tasty biscuits and were asked to taste and rate these biscuits and allowed to eat as much as they liked. After that, they filled in the Three Factor Eating Questionnaire (TFEQ; Stunkard & Messick, 1985). Whereas the amount eaten by control participants correlated negatively with the eating restraint scale of the TFEQ, there was a significant positive correlation for participants in the experimental condition. Having previously been exposed to the sight of tasty food, they consumed more biscuits and the amount eaten was greater, the higher their restraint score. This finding is particularly important, because it demonstrates that the effect is not restricted to restrained eaters identified with the RS of Herman and Polivy (1980). Fedoroff et al. (1997) also produced persuasive evidence that priming restrained eaters with palatable food stimuli can increase consumption. These researchers exposed half their experimental participants, who were led to believe that their task was to rate the taste of different pizzas, beforehand to the smell of pizza being baked in the next room (i.e., the prime). The other half was not exposed to pizza smells. Participants were also asked before and after the experiment how much they craved pizza. Restrained eating was assessed after the experiment with the RS (Herman & Polivy, 1980). They found that in restrained (but not in unrestrained) eaters, exposure to the delicious smell of pizza increased the craving for pizza as well as the number of pizza pieces eaten in the taste test.

Using implementation intentions to bridge the intentionbehavior gap
Although most restrained eaters manage on their own to control their weight or even moderately to reduce it, some of those who fail in their attempts at weight control end up in clinical weight loss trials. The processes described by the goal conflict model identify the psychological reason why weight loss is so difficult and why participants in weight loss trials typically end up regaining the lost weight. Even though they have the intention to keep to their diet (i.e., weight control goal), exposure to palatable food stimuli primes their eating enjoyment goal to an extent that that goal becomes dominant and induces them to violate their dieting intention. Thus, by the time they have the opportunity to eat some palatable food, their eating enjoyment goal has become dominant and their dieting intentions are being suppressed.
There is evidence that reminding individuals of their dieting goal at the time of eating can reduce the risk of dietary violations. This was illustrated in a study by Papies and Hamstra (2010) who made use of the fact that in Dutch butcher shops, tasty bits of (usually fatty) sausage are offered on trays for customers to try. For half of the customers of such a shop, a poster at the door announced a recipe for a dish that was recommended for being low-calorie. (Restrained eating was measured with the RS, just before customers were leaving). This diet-prime reduced consumption of the tasty snacks among restrained but not among unrestrained eaters. Thus, this prime appeared to have restored the weight control goal among restrained eaters and enabled them to resist the temptation to taste the sausage samples on offer. Papies and Veling (2013) replicated these findings in a study that took place in a small café-style restaurant. The daily special consisted of three different main dish salads that were announced on a sheet displayed on the table. In the diet-prime condition, the salads were described as low in calories or for the calorie conscious. In the control condition, these descriptions were not added. In a pilot study, all the dishes offered by the restaurant had been rated according to their perceived calorie content. The experimenters also noted the food choices and asked customers to fill in the RS after they had finished eating. When food choices were categorized according to the ratings established in the pilot study, it was found that restrained eaters made significantly more low calorie choices in the diet prime than the under no-prime conditions. Food choices of unrestrained eaters were not affected by the primes.
Obviously, we cannot make sure that restrained eaters always receive a warning message when being exposed to food temptations. However, we can increase the likelihood that they will remember their dieting intentions in such tempting situation by instructing them to form implementation intentions. In contrast to behavioral intentions, implementation intentions specify the situation in which the intention should be enacted and have shown to remind individuals of the intention when confronted with the specified situation. A meta-analysis of 84 studies testing the effectiveness of implementation intentions for a wide variety of behaviors reported a medium to large effect size of implementation intention intervention on goal attainment (Gollwitzer & Sheeran, 2006). More recent meta-analyses focusing specifically on healthy and unhealthy eating have not only confirmed this finding, but also concluded that effects were stronger for studies that increase healthy eating, rather than diminish unhealthy eating Carrero et al., 2019).

Studies targeting weight loss
Results of studies using implementation intentions to help overweight and obese individuals to reduce their food consumption have been mixed. Luszczynska et al. (2007) conducted the earliest study assessing the effectiveness of implementation intentions with participants in a weight-loss program. Participants were overweight or obese women, who were enrolled in a Weight Watchers (WW) program in the area of Warsaw. Participants were randomly assigned to either a control or an implementation intentions condition. Whereas participants in the control condition participated in the standard WW program, those in the implementation intention condition were in addition asked to make detailed plans about when, where and what to eat and when, where and how to exercise. The trial lasted for two months. At the end of that period, the implementation intention group had lost significantly more weight (4.2 kg) than the control group performing the usual WW program. Thus, adding implementation intentions to the standard WW procedure doubled the average weight-loss achievement over a two months period. Another successful test of the effectiveness of implementation intentions conducted by Armitage et al. (2014) also recruited participants already enrolled in a regular weight loss program. They also used implementation intentions that targeted specific eating behaviors. After one month, the participants in the implementation intention condition lost 0.60 kg more than participants who participated only in the regular sessions. However, other studies failed to find implementation intentions superior to simple goal intentions (e.g., Benyamini et al., 2013;Hayes et al., 2021).
One of the most concerning failures was the randomized controlled trial of Knäuper et al. (2018), who compared the effectiveness of a modified version of the behavioral weight loss treatment developed by the Diabetes Prevention Program (DPP; Diabetes Prevention Program Research Group, 2002), with a group that -in addition to the weight-loss treatment -was asked to form implementation intentions. Although both groups achieved significant weight loss of 9.42% (standard treatment) and 10.63% (implementation added treatment), the differences between the two groups was not statistically significant. There was also no difference in weight regain after 24 months (Knäuper et al., 2020). Both groups had retained approximately 6% weight loss compared to their initial weight.
What went wrong? Why do implementation intentions work in some studies (e.g., Armitage et al., 2014;Luszczynska et al., 2007), but not in others (e.g., Benyamini et al., 2013;Hayes et al., 2021;Knäuper et al., 2018). Because implementation intentions promoting healthy eating are highly effective Carrero et al., 2019), the problem appears to be specifically connected with intentions not to eat certain foods. There is reason to assume that these failures have been due to the use of ineffective implementation intentions. All the studies reviewed so far used instructions that targeted specific behaviors. Because exposure to palatable food tends to prime the eating enjoyment goal according to the goal conflict model (Stroebe, 2018(Stroebe, , 2022(Stroebe, , 2023Stroebe et al., 2013), implementation intentions to refrain from eating palatable food are only effective for restrained eaters, if they prime the eating control goal. Targeting specific eating behaviors might sometimes prime that goal, but the effect will not be reliable, probably because exposure to the palatable food item will often elicit more powerful eating enjoyment thoughts. To reliably prime the weight control goal, participants need to be primed with an implementation intention that contains "think of dieting" as the "then" component, because dieting thoughts constitute weight control primes.

Implementation intentions that prime the weight-control goal
There is evidence that implementation intentions that contain "think of dieting" as the "then" component prime the weight control goal and also reduce eating of high calorie food (Kroese et al., 2011;van Koningsbruggen et al., 2011;Veling et al., 2014). Participants in the study of van Koningsbruggen et al. (2011;Study 1) were first asked to form implementation intentions for five palatable foods (e.g., French fries; chocolate; pizza). Participants in the think of dieting condition were instructed to form the following implementation intention (e.g., for chocolate): "The next time I am tempted to eat chocolate, I will think of dieting". The control group did not form implementation intentions. Participants then had to complete a series of unfinished words in a word completion task. Half of the unfinished words could be completed as words that were either related or unrelated to dieting. Restrained (but not unrestrained) eaters completed significantly more of the critical words as dieting words (e.g., weight loss) after having formed a dieting implementation intention compared to having been in the control group. Kroese et al. (2011; Study 2) used a lexical decision task instead of word completion to assess the effect of "think of dieting" implementation intentions. Participants in their study were women, who wanted to reduce their chocolate consumption, but were not very successful in doing so. Half of the participants were instructed to form the following implementation intention: "If I see or smell chocolate, then I will follow my dieting goal". A control group was not given any instructions. Participants were then asked to perform a lexical decision task, where half of 42 targets were non-word, 18 were neutral words and three were diet-related words (dieting, slim, thin). Before each of the three dieting targets, the word chocolate was presented. Before the other targets, neutral words appeared. Participants, who had formed the dieting goal implementation intentions, were significantly faster recognizing the dieting targets than participants, who had not formed an implementation intention. There was no difference in reaction times to neutral targets.
Both van Koningsbruggen et al. (2011) and Kroese et al. (2011) also assessed whether the formation of implementation intentions that were likely to prime the weight control goal would influence the (reported) consumption of the foods specified in the implementation intention. In Study 2, van Koningsbruggen et al. (2011) had half of their participants form implementation intentions with regard to the same five palatable foods used in Study 1. There were two control conditions. In one control condition participants followed the same steps as with the "think of dieting" conditions, but were instead asked to think "then I will not eat it". Participants in the other control condition were not asked to form any implementation intention. Two weeks later, participants were contacted again via e-mail and asked to report, how much of each of the foods they had eaten. Whereas the implementation intention to think of dieting was associated with a reduction in the consumption of these palatable food items, there was no reduction in the other two control conditions. The instruction "the next time I am tempted to eat chocolate, I will not eat it" was not effective in reducing consumption, even though it fulfills the "if-then" character of an implementation intention. Thus, unlike the instruction to think of dieting, the instruction not to eat chocolate did not appear to have primed the eating control goal. Kroese et al. (2011), who had asked participants at baseline, how many portions of chocolates they had consumed on average per day during the past week, reported that the implementation intention, which primed the dieting goal with regard to chocolate, significantly reduced reported chocolate consumption one week later (compared to a control group with no instructions). But even more interesting, are the results of a regression analysis that included the portions of chocolate consumed during the baseline week in Step 1, and reaction times to diet words assessed with the lexical decision task in Step 2, with the dependent variable being the portions of chocolate consumed during the following week. They found a main effect of portions of chocolate eaten at baseline, which explained 27.7% of the variance. However, reaction times measured with the temptation-primed lexical decision explained an additional 8% of the variance in reported chocolate consumption. Thus, participants, who ate more chocolate at baseline continued to do so. But in addition, the reaction time measure that indicated the effectiveness of the "think of dieting" manipulation accounted for another 8% of the variance in chocolate consumption indicating the effectiveness of the "think of dieting" instructions.
To summarize, both studies reported evidence that forming the implementation intention to "think of dieting" or "follow the dieting goal" when confronted with specific palatable food items did prime dieting thoughts in participants, who wanted to avoid the high calorie foods specified in the implementation intention. They further demonstrated that this kind of implementation intention reduced (reported) consumption of such foods. However, each study reported important additional findings. Kroese et al. reported that the impact of implementation intentions on chocolate consumption was moderated by the strength of priming effects assessed with the lexical decision task. Thus, the more effective the implementation intention was in increasing the cognitive accessibility of dieting thoughts when confronted with (the word) chocolate, the less chocolate participants reported consuming in the following week. Van Koningsbruggen et al. (2011) demonstrated that an implementation intention that specifies "then I will not eat it" rather "then I will think of dieting" did not reduce consumption of the food specified in the "if component" of the implementation intention. Thus, consistent with the goal conflict model, implementation intentions are only effective, if they prime the weight control goal.
Despite these interesting findings, both studies could be criticized for the fact that the consumption was self-reported. Self-reports could have been influenced by demand characteristics. However, this was unlikely in both studies. In the study of van Koningsbruggen et al. (2011) one would have expected stronger demand effects in the condition with the implementation intention not to eat any of the attractive food items. And yet, that implementation intention did not affect consumption. In the Kroese et al. experiment, demand effects could not explain the pattern reported in the regression analysis.
Nevertheless, it was important to test whether "think of dieting" implementation intentions would also be effective with the outcome being actual weight loss. Veling et al. (2014) conducted a study that assessed the impact of "think of dieting" implementation intentions on body weight. The study was conducted via the Internet, with only the first session taking place in an experimental laboratory, where participants were instructed individually. Weight was measured in the laboratory before the first and after the last session. The majority of the participants were students, but a minority consisted also of non-students. Participants were told that dieting could be facilitated by planning. In the dieting implementation condition, they were asked to form implementation intentions with regard to five eating occasions (e. g., breakfast, lunch, snacks during the day, dinner, snacks after dinner). For each of these occasions, participants were first asked to think about the typical way such an occasion unfolded. They were then asked for the best moment to be reminded of dieting and then to think of dieting on these occasions. For the control implementation condition, participants were asked to make implementation intentions on taking it easy for five occasions (days of the week). This training program was repeated each week for four weeks. After the initial session, participants were contacted via the Internet on each Tuesday. Dieting goal strength was measured at the beginning of the study with a six-item scale (e.g., "I will do my best to lose weight in the coming month", "I want to reduce my calorie content in the coming month". When participants were weighed again after four weeks, those in the dieting implementation intention condition lost significantly more weight than participants in the control condition. Furthermore, weight loss was moderated by the strength of participants dieting intentions. Participants with strong weight-loss intention lost significantly more weight (1.34 kg) than participants with weak weight loss intentions (0.22 kg). As expected, the strength of the weight loss intention had no effect on weight loss in the control condition. The fact that weight-loss goal strength is strongly associated with the amount of weight lost in the implementation condition makes this procedure an ideal combination with weight loss trials. Participants in a weight loss trial are likely to be much more motivated to lose weight than the participants in the Veling et al. (2014) study.

Conclusions
The finding that patients, who lost weight during behavioral weightloss treatment regain the lost weight within a few years suggests a weakening of the intentions formed during treatment. Because there is evidence that the formation of implementation intentions can to some extent prevent such deterioration, it would appear worthwhile to add implementation intention procedures as part of behavioral treatment of obesity. As multiple meta-analyses (e.g., Adriaanse et al., 2011;Carrero et al., 2019;Gollwitzer & Sheeran, 2006;Vila et al., 2017), have demonstrated, implementation intentions are highly effective in strengthening the intention-behavior relationship in a wide variety of behavioral contexts. It is therefore surprising that implementations have not reliably influenced dieting behavior aimed at weight-loss (Benyamini et al., 2013;Hayes et al., 2020Hayes et al., , 2021. In the context of trying to improve the effectiveness of behavioral weight loss treatments, the findings of Knäuper et al., (2018;, that adding implementation intentions to engage in specific diet-related behaviors to an acclaimed behavioral weight loss program failed to improve both weight loss and weight loss maintenance is particularly concerning.
The goal conflict model of eating offers an explanation for these failures. According to this model, dieters are confronted with a conflict between their goal of eating enjoyment and weight control. Exposure to palatable food primes their eating enjoyment goal, which often tends to overpower their weight control motivation. The instruction not to eat certain palatable food does not address this problem, because dieters typically know that they should not eat tasty high calorie food. However, once eating enjoyment has been primed, they are tempted to do so anyway. In contrast, the instruction to think of dieting when confronted with some palatable food strengthens the weight control goal and helps them to keep to their weight control intentions.
The studies of van Koningsbruggen et al. (Experiment 1; 2011) and Kroese et al. (Experiment 2; demonstrated that the implementation intention "to think of dieting" (or to "follow your dieting goal") when confronted with some specified highly liked food had the effect that restrained eaters did not only think about dieting when exposed to these foods, they also ate less of them during the two week period of these studies. Kroese et al. (2011) further demonstrated that this effect was stronger, the stronger the effect of the implementation intention in eliciting dieting thoughts when confronted with (the word) chocolate. Van Koningsbruggen also found that the implementation intention to think "then I will not eat it", when confronted with some specified attractive food had no effect on consumption. This last finding can explain, why the standard implementation intentions used in weight loss research are often ineffective. They might sometimes trigger more general thoughts about dieting, but this effect does not appear to be reliable.
I would like to add that the "think of dieting" implementation intention is no miracle cure. It is a theoretically-based implementation intention that -as research has shown -is more effective than the implementation intention not to eat some specific foods. How successful it will be in helping restrained eaters to resist food temptations will depend on a myriad of other factors, such as the attractiveness of the food people are trying not to eat, their general level of self-control and whether their control is not momentarily depleted by tiredness, alcohol consumption or ego depletion. Nevertheless, the evidence is sufficiently promising to justify making this type of implementation intention a standard part of behavioral treatment of overweight and obesity.

Ethical approval statement
All research reviewed in this article has been published in international journals. With regard to the two (published) Utrecht studies, ethical approval has been given by the ethical advisor of the Faculty of Social Sciences of Utrecht University.

Declaration of competing interest
There are no conflicts of interest.