Effectiveness of a Smartphone-Delivered Approach-Avoidance Intervention in Dietary Behavior - A Randomized Controlled Trial

Effectiveness of a Smartphone-Delivered Approach-Avoidance Intervention in Dietary Behavior - A Randomized Controlled Trial | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Effectiveness of a Smartphone-Delivered Approach-Avoidance Intervention in Dietary Behavior - A Randomized Controlled Trial Matthias Burkard Aulbach, Mareike Roettger, Hannah Alebeek, Sercan Kahveci, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6751171/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 28 Nov, 2025 Read the published version in International Journal of Behavioral Nutrition and Physical Activity → Version 1 posted 9 You are reading this latest preprint version Abstract Background Given the therapeutic potential of Approach-Avoidance interventions (AAIs) in the alcohol domain, research has increasingly applied them to the food domain. In AAIs, harmful stimuli are avoided while healthy ones are approached, for example by respectively moving a phone away from or towards oneself. Methods We administered a phone-based AAI six times over two weeks to 156 participants in a pre-registered randomized-controlled trial to reduce intake of six “decrease-foods” and increase intake of six “increase-foods”, selected according to each participant’s individual dietary goals. The control group received a placebo task in which all stimuli were equally often approached and avoided. Food craving and intake were the outcomes, measured daily during the training period, four days before and after, and once during a follow-up one month after training. Per-food approach bias was recorded before and after training, and at follow-up. Results Compared to placebo, active training reduced the level of decrease-food craving without affecting how often craving occurred. Restrained eaters and those with low past dietary success showed the strongest craving strength reduction. Active training also reduced approach bias for decrease-foods. We found no intervention effects on increase-foods on any outcome. There were no interpretable training effects for food intake and no changes were maintained at follow-up. Conclusions We find support for the use of AAI against food cravings for goal-incongruent foods, especially for those who struggle with their diet. It remains to future research how this can be effectively translated into reduced food intake. Trial registration This study was registered in the German Clinical Trials Register, ID DRKS00030780. cognitive bias modification approach bias modification approach-avoidance task dietary change eating behavior mHealth restraint eating intervention food craving ecological momentary assessment randomized controlled trial Figures Figure 1 Figure 2 Figure 3 Background Overweight and obesity have become major health concerns in many countries [1, 2], with the main cause for overweight being an imbalance between energy input and output [3, 4]. Calorie-dense foods are readily available in obesogenic environments, which often leads to excessive consumption and insufficient intake of healthier foods such as fruits and vegetables. While many people know that diet, overweight, and health are related, and many people intend to adhere to healthier diets [5], they often fail to actually follow through on these intentions [6]. Many psychological models struggle to explain this phenomenon, as they model behavior mainly as originating from thoroughly reflected intentions [7–9]. Dual-process models address this shortcoming by emphasizing that behavior is not only determined by such reflective processes but also by impulsive ones [10, 11]. These models have led to the development of computer-based reaction time (RT) tasks that aim to measure such impulsive determinants of behavior. One such task is the Approach-Avoidance Assessment task (AAA), in which participants respond to images of different categories (e.g., palatable vs non-palatable foods or objects) on a screen by performing movements that represent approach or avoidance, such as pulling or pushing a joystick [12] or moving a smartphone towards or away from their body [13]. There are often differences in RTs between approach and avoidance trials, and this difference is often greater for one stimulus category than another; this latter difference is termed an “approach bias”. This approach bias is stronger towards foods in individuals with high trait food craving [14] and towards chocolate in those with currently strong state chocolate craving [15]. It is also related to increased food consumption in impulsive individuals [16] and in people prone to external or emotional eating [17]. Further, approach bias relates to participants’ desire to eat specific foods [18, 19]. Past Approach-Avoidance Interventions and their Pitfalls Moving from observation to intervention, researchers have devised ways to intervene on consumption behavior using an adapted task setup. In Approach-Avoidance Interventions (AAI), images are systematically paired with specific reactions, usually unbeknownst to the participant: participants for example always have to approach healthier and avoid less healthy foods. Such interventions haven been shown to effectively reduce food craving [20] and intake of trained foods, at least for some people and under certain circumstances [21–25]: For example, interventions are more effective for those who struggle with self-regulation [23, 26]. Not all interventions have been effective however [21], spurring discussions about reasons for a lack of effectiveness, and three prominent aspects have arisen, which we discuss in the following. First, the exact instructions and setup of the task differ between studies such that participants either approach or avoid based on the stimulus category (“relevant-feature” AAA/AAI) or some other aspect of the image (e.g. its spatial orientation; “irrelevant-feature” AAA/AAI). Since a relevant-feature task focuses attention on the image, it performs better as an approach bias measure than the irrelevant-feature version [27, 28]. While data for interventions are rare [29–31], it seems likely that relevant-feature AAIs would be superior to an irrelevant-feature version due to this aspect of attentional focus. Second, many studies present pre-selected images that are not personalized to the participant. This likely dilutes training effectiveness, as some participants may not like a given food, or have no intention of changing their intake of it. Therefore, allowing participants to select images of foods they would like to eat less should help in enhancing effectiveness. However, intentions for behavior change have mostly been ignored in this literature. As a third reason for mixed findings, most studies have delivered only single intervention sessions in a laboratory or online; delivering an AAI repeatedly should enhance effectiveness [32, 33]. Smartphones easily enable this and have the further advantage of bringing the intervention closer to day-to-day consumption behavior. In addition to these three potential reasons limiting the effectiveness of AAIs, most AAI studies to date mainly focused on reducing the intake of unhealthy foods, somewhat neglecting the fact that attaining a healthier diet also requires increasing the intake of healthier foods. This is not only important for health reasons, as many people fail to consume the recommended amount of fruits and vegetables [34]; it is also important from a behavior change perspective, as it is more effective to substitute an unwanted behavior with an alternative, similar behavior, than it is to simply suppress the unwanted behavior [35–37]. AAIs are promising in this regard as they can be programmed to train both approach of healthy foods and avoidance of unhealthy foods simultaneously, and could thus facilitate such behavior substitution [38]. The current study In the current study we deploy a mobile AAI that addresses these shortcomings of prior studies, and we investigate whether this AAI is able to effectively support dietary changes. Specifically, we deliver a smartphone-based multisession relevant-feature AAI that allows participants to select which food images they will work with during the study based on their intention to change how often they eat them. This results in two stimulus sets: individualized foods that participants want to eat more (“increase-foods”) and those they want to eat less (“decrease-foods”). The AAI is accompanied by twice-daily ecological momentary assessment (EMA) that includes measures of intake, food craving, and other relevant day-level variables. To control for potential placebo effects that may come with repeated performance of an eating-related behavioral task, we compare this to an active control in which increase-foods and decrease-foods are approached and avoided equally often. In the assessment of approach bias, we include neutral control stimuli to ensure that approach bias for increase-foods can be separately interpreted/calculated from the approach bias of decrease-foods. As preregistered, we expect that participants in the AAI will display stronger increases in consumption of, craving for, and approach bias towards increase-foods and stronger decreases in consumption of, craving for, and approach bias towards decrease-foods compared to the control condition. We also explored potential person-level moderators of the effect to determine which population benefits most from AAI. Methods The full methods of this randomized controlled trial (RCT) are outlined in the corresponding protocol paper [39] and the preregistration https://osf.io/4k3q9/. Here, we only describe the aspects of the trial relevant to this study and refer the reader to those documents for more detail. Participants We recruited 157 participants (132 women, 25 men, M age [ SD ] = 27.6 [9.73]; see protocol paper for power analysis [39]) through university and social networks, and through word of mouth. Participants had to be between 18 and 60 years old, not pregnant, and report living without an eating disorder. Participants needed to have a goal of changing their eating behavior in the upcoming weeks. Six participants were excluded from all analyses as their AAI/AAA sessions were all either missing, or excluded (see below for exclusion criteria for AAI/AAA sessions; see consort flowchart in the supplementary materials). One participant was excluded from all analyses that had approach bias scores as the outcome of interest, as they seemed to have technical issues with the smartphone application. Questionnaires For each scale we report both Cronbach’s alpha using the R-package psych [40], and McDonald’s omega with bootstrapped 95% confidence intervals using the R-package MBESS [41]. Perceived Self-Regulatory Success in Dieting Scale (PSRS) The Perceived Self-Regulatory Success in Dieting Scale (PSRS) [42] is used to differentiate between successful and unsuccessful dieters. It consists of three items rated on a 7-point Likert scale with higher scores indicating more successful self-regulation. Internal consistency was questionable: ω = .69 [.56, .76], α = .68. Dutch Eating Behavior Questionnaire restrained eating (DEBQ-res) and external eating (DEBQ-ext) subscales From the Dutch Eating Behavior Questionnaire (DEBQ) [43], we administered the subscales of restrained eating (DEBQ-res) and external eating (DEBQ-ext). Each subscale consists of 10 items and is rated on a 5-point Likert scale from 1 (never) to 5 (very often). Internal consistency was good for both the restrained eating subscale: ω = .89 [.86, .91], α = .89; and the external eating subscale: ω = .85 [.80, .89], α = .85. UPPS Impulsive Behavior Scale (UPPS) The UPPS impulsive behavior scale (UPPS) [44] consists of four subscales that each measure a facet of impulsivity: urgency, lack of pre-meditation, lack of perseverance, and sensation seeking. Each subscale consists of five statements rated on a 4-point Likert scale from 1 (strongly disagree) to 4 (strongly agree). Higher scores on an UPPS subscale are interpreted as a higher level of impulsivity in the respective impulsivity facet. Internal consistency was questionable for the urgency subscale, ω = .68 [.57, .77], α = .68; but acceptable for the other three subscales, pre-meditation: ω = .74 [.65, .81], α = .73; perseverance: ω = .79 [.72, .85], α = .79; sensation seeking: ω = .73 [.65, .79], α = .70. Materials and procedure Participants gave informed consent online, followed by a range of questionnaires including demographics. After the questionnaires, they rated 90 preselected images of foods and drinks from the food.pics [45] and CROCUFID [46] databases, photographs taken by the study authors, and license-free online stock photograph databases. The ratings included two questions for each image: “In the last three weeks, on how many days have you eaten/drunk this food/drink?” (recent intake) and “In the next three weeks, on how many days do you want to eat/drink this food/drink?” (intended intake). Answers to both questions were indicated on a slider ranging from 0 to 21 days. The six images with the largest (recent intake > intended intake: “decrease-food”) and the six images with the smallest difference (recent intake < intended intake: “increase-food”) between the two scales were included in the study, with a random selection of four images per category included in the AAA/AAI. A randomly selected 8 out of 12 images of office items from the food.pics [45] and FRIDa [47] databases served as control stimuli. We excluded participants that did not have six decrease-foods which they consumed at least twice weekly. At this point, an R-script (The function sample randomly outputs the number ‘1’ or ‘2’ which correspond to the conditions.) randomized participants to either the intervention or control group with the condition unknown to the study team. Next, a member of the research team conducted a setup call with each participant to install and explain the use of two smartphone applications: m-path [48] for EMA, and the app to conduct the AAA/AAI. The remainder of the study consisted of four phases (Fig. 1): a baseline phase on days 1–4, an intervention phase on days 5–16, a post-intervention phase on days 17–20, and a follow-up phase 4 weeks after day 20. Participants completed a single measurement-AAA during the baseline phase (day 4) and also during the post-intervention phase (day 17), while performing an AAI or AAA every other day during the intervention phase, depending on condition (days 6, 8, 10, 12, 14, and 16). EMA was collected throughout these three phases. The follow-up consisted of a single EMA prompt and AAA measurement (day 48, four weeks after the last post-intervention day). For the AAA/AAI, we used a modified smartphone application [13, 49]. In this task, users held the phone horizontally in front of them while responding to stimuli shown on the screen. After a fixation dot for 1500 ms at the start of the trial, the app presented a food or object stimulus. Depending on image type (food or object), participants had to move the phone towards or away from themselves, representing an approach or avoidance response, respectively. Two types of Approach-Avoidance Tasks were administered: measurement-AAAs and training-AAIs. Each measurement-AAA block consisted of 24 trials each, while each training-AAI/AAA consisted of 16 trials per block. Each session consisted of 4 blocks, with each block preceded by 4 practice trials. Before each block, instructions were presented, with blocks 1 and 3 instructing participants to approach foods and avoid objects (“approach-food-blocks”) and blocks 2 and 4 giving the opposite instruction (“avoid-food-blocks”). In the AAI (intervention condition), the approach-food-blocks exclusively featured increase-foods and objects, and the avoid-food-blocks exclusively featured decrease-foods and objects; thus, increase-foods were always approached while decrease-foods were always avoided. In the AAA (control condition), all food images were presented in each block. During the intervention phase (days 5–17), after the mid-day prompt (content not relevant here), participants were prompted to conduct a session of AAA/AAI on every other day (days 6, 8, 10, 12, 14, 16) resulting in six AAA/AAI sessions. In addition, all participants conducted an AAA on days 4 and 17 and during the follow-up. Throughout the study, participants followed a twice-daily EMA schedule, with the mid-day prompt just before their typical lunch time and the other at the end of the day (timing agreed upon individually) with all EMA items using a virtual 0-100 slider (where not reported differently). Relevant items measured craving for (“How strongly have you been craving this food today?”) and intake of (“How much have you eaten of this food today?”) each of the included foods. Data preprocessing and analysis Approach bias To pre-process the AAA data, we removed trials with non-responses, movements in the wrong direction, or RTs above 2000 or below 200 ms. After this, exclusions continued within each session within each participant, as we removed trials that deviated more than 3 SDs from the mean. AAA sessions with more than 25% of trials missing or removed were then excluded from analysis entirely. Separately for approach and avoidance trials as well as for sessions, RTs were averaged across AAA blocks and all object stimuli. For foods, RTs were averaged across AAA blocks only. The average approach or avoid response for objects on a session was subtracted from stimulus-specific food approach or avoidance response on that session to achieve single-difference approach and avoidance scores: Stimulus-specific approach = [food-specific approach]-[average object approach] Stimulus-specific avoidance = [food-specific avoidance]-[average object avoidance] Double-difference scores were used to achieve a full bias score per food stimulus and session: [food-specific avoidance]-[food-specific approach]-[average object avoidance]-[average object approach]. Hence, positive values imply the food was approached faster than it was avoided, relative to the difference between approach and avoidance RTs for objects. Analysis As multiple datapoints are nested within participants and stimuli, we used multilevel models (MLM) to assess the hypotheses. Since our outcome measures craving and intake displayed strong skew and many zeroes, we used Bayesian multilevel two-part hurdle models [50]. In brief, these regression models consist of a hurdle part akin to logistic regression that estimates the probability of the outcome not being zero, and a continuous part following a gamma distribution that estimates the size of the non-zero values. The hurdle model therefore allows to separately determine which variables predict whether intake or craving occurred and, if so, how much was eaten/how strong the craving was. Since these models flexibly deal with missing values they approximate an intention to treat analysis with multiple imputation.. Note that missing data were rare and similarly spread across groups (see Table 1), making selective drop out unlikely. For each model, we report the regression coefficient of interest as well as both the 95% and the 89% highest density interval (HDI), reflecting the most credible values of the respective model parameter. This is equivalent to a 5% or 11% alpha level, respectively. If the 89%-HDI does not include 0, we describe the model parameter as being estimated as “significantly” above or below 0. Analyses include only data from the baseline and post-intervention phase (4 days each) and include all 12 food stimuli. In separate models, we respectively predicted the stimulus-level intake, craving, and approach bias scores for increase and decrease-foods separately using the interaction between Intervention group (0 = 50/50 AAA vs. 1 = 100/0 AAI) and time (0 = Pre vs. 1 = Post) according to this formula: Intake/Craving/Approach Bias increase/decrease−foods ~ Intervention Group * time + (time | Subject) + (time | Stimulus) To assess intervention effects to follow-up, we then ran equivalent models containing only data from post-intervention and follow-up for all outcomes. For moderation analyses we added the relevant moderator and all interaction terms to the models comparing pre- to post-intervention values. We computed post-hoc contrasts using the R-package emmeans [51] for all models where we found evidence of an effect to assess between- and within-condition differences. We report the estimate as well as the 95% HDI for the contrasts of interest. Results Table 1 Descriptive Statistics and EMA Compliance by group Training group Placebo control group Sample size 72 79 Demographics Women (%) 60 (83.33%) 68 (86.08%) Age in years (SD) 25.61 (8.16) 29.66 (10.85) BMI (kg/m 2 ) 23.49 (6.22) 24.47 (3.95) Compliance N completed training sessions (out of 6) 5.51 5.57 Baseline EMA compliance (mean N completed questionnaires of 4 assessment points) 3.79 3.72 Post-intervention EMA compliance (mean N completed questionnaires of 4 assessment points) 3.58 3.60 Means (SDs) for questionnaire scores PSRS 3.56 (1.11) 3.53 (1.23) DEBQ-res 2.31 (0.73) 2.60 (0.75) DEBQ-ext 3.37 (0.70) 3.39 (0.61) UPPS lack of perseverance 3.06 (0.57) 2.90 (0.64) UPPS lack of premeditation 3.04 (0.54) 2.98 (0.56) UPPS urgency 2.20 (0.56) 2.28 (0.53) UPPS sensation seeking 2.53 (0.64) 2.34 (0.73) Means (SDs) for baseline phase (4 days) Intake decrease-foods (Slider rating 0-100) 19.3 (11.3) 20.2 (11.7) Intake increase-foods (Slider rating 0-100) 16.6 (11.4) 14.1 (10.6) Craving decrease-foods (Slider rating 0-100) 32.7 (18.3) 35.3 (16.7) Craving increase-foods (Slider rating 0-100) 23.8 (14.7) 25.3 (15.5) Means (SDs) for intervention perception Expectancy 2.52 (0.96) 2.41 (1.00) Contingency perception for trained decrease-foods 48.9 (20.9) 52.5 (14.8) Contingency perception for trained increase-foods 31.2 (19.5) 42.0 (19.5) We checked for differences in compliance between the two groups with two-sample t-tests. There was no difference in baseline EMA compliance ( t (143.09) = -0.82, p = .41), training compliance ( t (149) = 0.32, p = .75) or post-intervention EMA compliance ( t (145.99) = 0.16, p = .87). Intervention effects from pre- to post-intervention and follow-up Approach Bias: reduced for decrease-foods For decrease-food s, the group × time interaction for bias scores from baseline assessment to post-assessment was significant on an 89% HDI level (n = 144, b = -38.10; 89% HDI [-73.82, -0.39]; 95% HDI [-81.88, 8.29]). Bias for decrease-foods decreased more in the intervention group than in the control group (Fig. 2 , panel A). There was no significant interaction for increase foods (n = 144, b = -15.92; 89% HDI [-59.81, 28.95]; 95% HDI [-67.82, 40.36]). The models examining changes from post-intervention to follow-up also showed a significant effect for decrease-food s on an 89% HDI level (n = 141, b = 35.06; 89% HDI. [0.98, 66.89]; 95% HDI [-5.61, 75.64]), indicating a stronger increase of bias values in the intervention group than in the control group. Again, there was no effect for increase-food s (n = 142, b = 11.06; 89% HDI. [-24.84, 49.27]; 95% HDI [-32.84, 57.55]). Post-hoc contrasts did not show evidence for a group difference in bias scores for decrease-food s at the post-intervention measurement (89% HDI [-3.18, 51.09]; 95% HDI [-8.53, 58.0]) or at follow-up measurement (89% HDI [-39.53, 15.16]; 95% HDI [-45.37, 20.8]). Craving: reduced for decrease-foods We separately report changes in the likelihood of reporting (no) craving (hurdle part) , and changes in craving intensity when any craving above zero is reported (continuous part). The intervention group showed stronger reductions in craving intensity for decrease-foods compared to the control group from pre- to post-intervention, similar to the effect in bias scores. (n = 151, continuous part: b = -0.19; 89% HDI [-0.29, -0.08]; 95% HDI [-0.31, -0.05], Fig. 2 , panel B). This difference was not present for the probability of experiencing craving (n = 151, hurdle part: b = -0.25; 89% HDI [-0.62, 0.12]; 95% HDI [-0.70, 0.20]). Regarding increase-food s, again, there was neither a significant time × condition interaction for craving intensity nor for the probability of experiencing craving (n = 151, hurdle part: b = -0.31; 89% HDI [-0.70, 0.06]; 95% HDI [-0.77, 0.16]; continuous part: b = -0.04; 89% HDI [-0.19, 0.09]; 95% HDI [-0.21, 0.12]). When looking at the effects from post-intervention to follow-up 4 weeks later, we found a significant interaction for craving intensity (but not probability) for decrease-foods on a 89% HDI level (n = 149, hurdle part: b = 0.06; 89% HDI [-0.39, 0.51]; 95% HDI [-0.51, 0.61]; continuous part: b = -0.13; 89% HDI [-0.24, -0.003]; 95% HDI [-0.27, 0.02]), indicating that the intervention group’s craving intensity increased significantly more than the control group’s from post-intervention to follow-up. We found no effects for increase-foods (n = 149, hurdle part: b = -0.03; 89% HDI [-0.59, 0.27]; 95% HDI [-0.70, 0.36]; continuous part: b = -0.16; 89% HDI [-0.15, 0.10]; 95% HDI [-0.18, 0.12]. Post-hoc contrasts showed evidence for a difference in the craving intensity between the control group and the intervention group at the post-intervention measurement (89% HDI [0.11, 0.35]; 95% HDI [.08, .38]), but the apparent difference in the raw values at follow-up did not prove statistically reliable (89% HDI [-0.04, 0.24]; 95% HDI [-.07, .27]). The craving intensity within the intervention group decreased from baseline to post-intervention measurement (estimated difference = 0.29; 89% HDI [0.21, 0.37], 95% HDI [.19, .38]) and increased again from post-intervention to follow-up measurement (89% HDI [0.13, 0.31]; 95% HDI [.11, .34]). Intake: no group differences There was no significant time × condition interaction on intake for both increase-food s or decrease-food s for hurdle and continuous parts both in the pre-post and the post to follow-up-analyses. Test statistics and plots showing these results can be found in the supplementary materials. Moderation of craving effects by participant characteristics: restraint and perceived self-regulatory success matter Participants’ restrained eating and perceived self-regulatory success moderated the baseline to post-training change in craving intensity for decrease-foods (n = 151, three-way interaction for restrained eating in the continuous part: b = -0.19; 89% HDI [-0.32, -0.05]; 95% HDI [-0.36, -0.03], Fig. 3 (top); and for perceived self-regulatory success: b = 0.13; 89% HDI [0.04, 0.22]; 95% HDI [0.03, 0.24], Fig. 3 (bottom)). There was no evidence for an interaction in the hurdle part (restrained eating: b = 0.41; 89% HDI [-0.05, 0.90]; 95% HDI [-0.20, 0.98], perceived self-regulatory success: b = -0.13; 89% HDI [-0.43, 0.20]; 95% HDI [-0.53, 0.24]). The intervention was most effective in reducing craving intensity for participants with high restrained eating and low past perceived self-regulatory success. These two questionnaires were not significantly correlated; r (149) = − .17, p = .091. The analyses including the UPPS and external eating scales as well as those using other outcomes did not yield significant results (see supplementary materials). Discussion In this RCT, we found that a multisession mobile AAI reduced craving intensity for foods that participants wanted to eat less of. We further found weaker evidence that the intervention reduced approach bias for these foods as well. These effects did not transfer to intake, nor did we find effects on foods that participants wanted to eat more of. However, we found that restrained eaters and those with low perceived self-regulatory success in dieting showed the largest reductions in craving intensity for decrease-foods, indicating they might benefit most from the intervention. The current study was designed to address several shortcomings of earlier research on AAIs. In particular, we expected that the repeated delivery of the intervention (6 sessions/384 trials) would significantly boost its effectiveness. Indeed, the achieved reductions in craving intensity seem promising, as craving is a common obstacle to successful dietary goal pursuit [ 52 , 53 ] and is perceived as unpleasant, especially in dieters [ 54 , 55 ]. Like craving, approach bias for decrease-foods also decreased from pre- to post-intervention. This is in line with findings that craving and approach bias are coupled [ 18 , 56 ] as part of a cognitive-behavioral pattern that prepares for ingestion [ 54 ]. Similar tasks have also been shown to reduce food valuation based on neuroimaging [ 57 , 58 ]. In the current study, however, the achieved reductions in craving and approach bias did not result in intake reductions, in line with earlier studies that showed AAIs and similar tasks reduce bias or food liking without changing intake [ 22 , 59 ]. Several factors might have prevented the craving and bias reduction from translating into consumption reduction: recent research has shown that food cravings are associated with higher self-regulatory efforts, indicating that individuals adjust their self-regulatory efforts to experienced craving strength [ 60 ]. If the AAI reduced craving strength, participants might, in consequence, have reduced subsequent self-regulatory effort, leading to similar consumption levels as pre-intervention. In addition, eating behavior is context dependent: availability of craved foods, homeostatic meal planning, and social context are factors that influence (overt) eating [ 52 , 61 ], but not (covert) craving. Studies that have successfully changed actual eating behavior using the AAI and similar tasks (such as the Go-/No-Go task) have typically been conducted in controlled laboratory settings and thus ignored such real-life circumstances imminent in our EMA design [ 62 – 64 ]. We further found two eating behavior traits to moderate the intervention effects on decrease-food craving intensity: perceived self-regulatory success in dieting [ 42 ] and eating restraint [ 43 , 65 ]. Those who reported difficulties in regulating their dieting showed the strongest effects, indicating that AAI can support dietary goal pursuit of individuals who need it the most [ 23 ], and may be better suited for that purpose than for enhancing the dieting capabilities of those who are already successful. In light of dual-process models, this might be because those struggling with self-regulation often experience goal-incongruent behavioral impulses and AAIs might help to reduce such impulses. Additionally, independent from self-regulatory success, restrained eaters showed stronger effects than unrestrained eaters, which underlines the importance of having a dietary goal for AAI to have an effect [ 65 ]: Restrained eaters care strongly about what they (don’t) eat. AAI might be able to support this dietary goal through reducing craving. This seems particularly important as restrained eaters report stronger tendencies to experience craving on a trait level [ 66 ] and respond strongly to perceived breaches of their dietary rules [ 67 , 68 ]. While restraint eating is clearly not equivalent to caloric dieting, the latter likely brings about similar processes as in restrained eating (dieting goal) and thus future research could apply the present intervention trial to weight reduction dieting. We had expected to find not only goal-congruent changes for decrease-foods but also in increase-foods, increasing the appeal and consumption of foods participants wanted to eat more of (e.g. for their health benefits), based on earlier findings [ 38 , 69 , 70 ]. However, we found no main effects in this goal-congruent food category on any of our outcome variables. It might be that participants already chose foods they generally liked as their increase-foods and therefore, further increases in craving and approach bias might be hard to achieve. Future Directions In addition to the analyses reported here, we had assessed other variables that have been suggested to work as moderators or mechanisms of the effect, including the awareness of contingencies between stimuli and required reactions [ 38 , 71 ], expectations regarding training effects [ 72 – 74 ], as well as transfer of training effects to non-trained stimuli. However, none of the analyses on these variables yielded conclusive results and we therefore refrain from commenting on these theoretical debates (see the supplementary materials for all analyses). However, we urge future researchers to design their studies explicitly to examine such potential mechanisms to improve our understanding of how AAI effects come about in applied settings. We make a number of propositions for future research into dietary enhancement through AAI. First, research may benefit from an exclusive focus on restrained eaters with low dietary success, since our training was most effective in reducing craving in this subsample. Second, it may help to get participants to commit to their dietary intentions, as participants in the current study were free to pursue or abandon their dietary intentions. Third, future research could study changes in food intake without the influence of food availability, by ensuring participants always have their decrease and increase-foods available. Fourth, we suggest studying the combination of AAIs with other behavior change techniques that focus on different mechanisms of eating behavior [ 75 ]. Strengths and Limitations This study used a rigorous design to test an intensive, multi-session, mobile version of a task in a field setting that has thus far mainly been tested in laboratory environments. Its strengths thus include (1) high ecological validity due to the mobile nature of the intervention and data assessment; (2) comparison with a closely matched, active control group within a double-blind design to isolate effects of stimulus-action couplings from mere stimulus exposure, task performance, and experimenter as well as tracking effects; (3) a large sample that (4) provided many data points on craving, consumption, and approach bias for (5) a wide range of personalized food stimuli. Regarding limitations, allowing participants to freely choose food images might have resulted in fuzzy categories of decrease- and increase-foods which might have impaired effectiveness. Similarly, stimulus selections of some participants seem to indicate that the food item selection did not work fully as intended (e.g., one participant choosing ice cream as an increase-food), adding to the fuzziness of the categories. We recommend using a more restricted stimulus set in future studies, for example by limiting stimuli to snack foods and pre-classifying stimuli as possible increase or decrease-foods rather than leaving the choice entirely to participants. Conclusion The mobile AAI tested in this multi-session intervention study reduced craving intensity and approach bias towards goal-incongruent foods throughout the 12-day intervention period. The effects on craving were most pronounced for those participants who reported to struggle with dietary self-regulation and restrained eaters. While no effects were observed for actual food intake, our study adds to previous work showing the potential of cognitive bias modification interventions in health and psychopathology and instills hope that also eating behavior - being known to be highly resistant to change and multiply determined - can be changed in structured cognitive bias modification tasks that take up nor more than a few minutes per day and that can be disseminated at low costs on a global level. Abbreviations AAI Approach-Avoidance Intervention AAA Approach-Avoidance Assessment BMI Body-Mass Index DEBQ Dutch Eating Behavior Questionnaire PSRS Perceived Self-Regulatory Success in Dieting Scale RCT Randomized-Controlled Trial UPPS (Negative) Urgency, (lack of) Premeditation, (lack of) Perseverance, Sensation Seeking Declarations Ethics approval and consent to participate The study has received ethical approval from the ethics board of the University of Salzburg (reference number 27/2018, Add 2) and is conducted in accordance with the declaration of Helsinki. All participants gave informed consent to participate in this study. Consent for publication Not applicable. Availability of data and materials The datasets generated and/or analysed during the current study are available in the Open Science Framework repository, https://osf.io/yn7kt Competing interests The authors declare that they have no competing interests. Funding This research was funded in whole, or in part, by the Austrian Science Fund (FWF) [grant number P 34542-B]. SK and HvA were supported by the Doctoral College 'Imaging the Mind' (FWF; W1233-B). HvA was additionally supported by the project: Mapping neural mechanisms of appetitive behaviour (FWF; KLI762-B). The funder plays no role in the study design; collection, management, analysis, and interpretation of data; writing of the report; and the decision to submit the report for publication. For the purpose of open access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. Authors' contributions Conceptualization: MBA, HvA, SK, JB. Funding acquisition: JB. Formal analysis: MBA, MR. Investigation: MBA, MR, HvA, SK. Project administration: MBA. Visualization: MR. Writing - original draft: MBA, MR. Methodology: MBA, HvB, SK, JB. Supervision: JB. Writing—review and editing: MBA, MR, HvB, SK, JS, JB. All authors read and approved the final manuscript. Acknowledgements The authors acknowledge the computational resources and services provided by Salzburg Collaborative Computing (SCC), funded by the Federal Ministry of Education, Science and Research (BMBWF) and the State of Salzburg. References Global Health Observatory. WHO | Overweight and obesity. WHO; 2017. World Health Organization. World Health Statistics. World Health Organ Cardiovasc Dis Factsheet 112; 2015. World Health Organization. (2018) Obesity and Overweight. https://doi.org/10.1007/978-3-319-33228-4_447 World Health Organization. Healthy diet. World Health Organization. Regional Office for the Eastern Mediterranean; 2019. Santos I, Sniehotta FF, Marques MM, Carraça EV, Teixeira PJ. Prevalence of personal weight control attempts in adults: a systematic review and meta-analysis. Obes Rev. 2017;18:32–50. Haynes A, Kersbergen I, Sutin A, Daly M, Robinson E. A systematic review of the relationship between weight status perceptions and weight loss attempts, strategies, behaviours and outcomes. Obes Rev. 2018;19:347–63. Ajzen I. The theory of planned behavior. Organ Behav Hum Decis Process. 1991;50:179–211. Fishbein M. (1979) A theory of reasoned action: some applications and implications. Fishbein M, Ajzen I. (2010) Predicting and changing behavior: The reasoned action approach. https://doi.org/10.4324/9780203937082 Hofmann W, Friese M, Wiers RW. Impulsive versus reflective influences on health behavior: a theoretical framework and empirical review. Health Psychol Rev. 2008;2:111–37. Strack F, Deutsch R. Reflective and impulsive determinants of social behavior. Personal Soc Psychol Rev Off J Soc Personal Soc Psychol Inc. 2004;8:220–47. Rinck M, Becker ES. Approach and avoidance in fear of spiders. J Behav Ther Exp Psychiatry. 2007;38:105–20. Zech HG, Rotteveel M, van Dijk WW, van Dillen LF. A mobile approach-avoidance task. Behav Res Methods. 2020;52:2085–97. Brockmeyer T, Hahn C, Reetz C, Schmidt U, Friederich H-CC. Approach bias and cue reactivity towards food in people with high versus low levels of food craving. Appetite. 2015;95:197–202. Meule A, Lender A, Richard A, Dinic R, Blechert J. Approach–avoidance tendencies towards food: Measurement on a touchscreen and the role of attention and food craving. Appetite. 2019;137:145–51. Booth C, Spronk D, Grol M, Fox E. Uncontrolled eating in adolescents: The role of impulsivity and automatic approach bias for food. Appetite. 2018;120:636–43. Kakoschke N, Kemps E, Tiggemann M. Differential effects of approach bias and eating style on unhealthy food consumption in overweight and normal weight women. Psychol Health. 2017;32:1371–85. Kahveci S, Meule A, Lender A, Blechert J. Food approach bias is moderated by the desire to eat specific foods. Appetite. 2020;154:104758. Kahveci S, van Alebeek H, Berking M, Blechert J. Touchscreen-based assessment of food approach biases: Investigating reliability and item-specific preferences. Appetite. 2021;163:105190. Brockmeyer T, Hahn C, Reetz C, Schmidt U, Friederich HC. Approach Bias Modification in Food Craving - A Proof-of-Concept Study. Eur Eat Disord Rev. 2015;23:352–60. Aulbach MB, Knittle K, Haukkala A. Implicit process interventions in eating behaviour: a meta-analysis examining mediators and moderators. Health Psychol Rev. 2019;13:179–208. Dickson H, Kavanagh DJ, MacLeod C. The pulling power of chocolate: Effects of approach–avoidance training on approach bias and consumption. Appetite. 2016;99:46–51. Kakoschke N, Kemps E, Tiggemann M. Impulsivity moderates the effect of approach bias modification on healthy food consumption. Appetite. 2017;117:117–25. Kakoschke N, Hawker C, Castine B, de Courten B, Verdejo-Garcia A. Smartphone-based cognitive bias modification training improves healthy food choice in obesity: A pilot study. Eur Eat Disord Rev. 2018;26:526–32. Veling H, Verpaalen IAM, Liu H, Mosannenzadeh F, Becker D, Holland RW. (2021) How can food choice best be trained? Approach-avoidance versus go/no-go training. Appetite 105226. Chen Z, Veling H, Dijksterhuis A, Holland RW. (2017) Do impulsive individuals benefit more from food go/no-go training? Testing the role of inhibition capacity in the no-go devaluation effect. Appetite. https://doi.org/10.1016/j.appet.2017.04.024 Kahveci S, Rinck M, van Alebeek H, Blechert J. How pre-processing decisions affect the reliability and validity of the approach–avoidance task: Evidence from simulations and multiverse analyses with six datasets. Behav Res Methods. 2023;56:1551–82. Lender A, Meule A, Rinck M, Brockmeyer T, Blechert J. Measurement of food-related approach–avoidance biases: Larger biases when food stimuli are task relevant. Appetite. 2018;125:42–7. Fishbach A, Shah JY. Self-control in action: implicit dispositions toward goals and away from temptations. J Pers Soc Psychol. 2006;90:820–32. Meule A, Richard A, Dinic R, Blechert J. Effects of a Smartphone-Based Approach-Avoidance Intervention on Chocolate Craving and Consumption: Randomized Controlled Trial. JMIR MHealth UHealth. 2019;7:e12298. Schakel L, Veldhuijzen DS, van Middendorp H, Dessel PV, Houwer JD, Bidarra R, Evers AWM. The effects of a gamified approach avoidance training and verbal suggestions on food outcomes. PLoS ONE. 2018;13:e0201309. Aulbach MB, Knittle K, van Beurden SB, Haukkala A, Lawrence NS. App-based food Go/No-Go training: User engagement and dietary intake in an opportunistic observational study. Appetite. 2021;165:105315. Eberl C, Wiers RW, Pawelczack S, Rinck M, Becker ES, Lindenmeyer J. Implementation of Approach Bias Re-Training in Alcoholism—How Many Sessions are Needed? Alcohol Clin Exp Res. 2014;38:587–94. Nour M, Sui Z, Grech A, Rangan A, McGeechan K, Allman-Farinelli M. The fruit and vegetable intake of young Australian adults: a population perspective. Public Health Nutr. 2017;20:2499–512. Adriaanse MA, van Oosten JMF, de Ridder DTD, de Wit JBF, Evers C. Planning What Not to Eat: Ironic Effects of Implementation Intentions Negating Unhealthy Habits. Pers Soc Psychol Bull. 2011;37:69–81. Gardner B, Rebar AL. Habit Formation and Behavior Change. Oxf Res Encycl Psychol. 2019. https://doi.org/10.1093/acrefore/9780190236557.013.129 . Lally P, Wardle J, Gardner B. Experiences of habit formation: A qualitative study. Psychol Health Med. 2011;16:484–9. Kahveci S, Van Alebeek H, Blechert J. (2024) The Dual-Feature Approach-Avoidance Task: Validity, Training Efficacy, and the Role of Contingency Awareness in Changing Food Preference. Cogn Emot 1–23. Aulbach MB, Van Alebeek H, Kahveci S, Blechert J. Testing the effectiveness of a mobile approach avoidance intervention and measuring approach biases in an ecological momentary assessment context: study protocol for a randomised-controlled trial. BMJ Open. 2023;13:e070443. Revelle W. (2022) psych: Procedures for psychological, psychometric, and personality research. Kelley K. (2022) MBESS. Meule A, Papies EK, Kübler A. Differentiating between successful and unsuccessful dieters. Validity and reliability of the Perceived Self-Regulatory Success in Dieting Scale. Appetite. 2012;58:822–6. van Strien T, Frijters JER, Bergers GPA, Defares PB. The Dutch Eating Behavior Questionnaire (Debq) for Assessment of Restrained, Emotional, and External Eating Behavior Internal Structure and Measurement Invariance of the Dutch Eating Behavior Questionnaire (DEBQ) in a (Nearly) Representative Dutch. Int J Eat Disord. 1986;5:295–315. Whiteside SP, Lynam DR, Miller JD, Reynolds SK. Validation of the UPPS impulsive behaviour scale: a four-factor model of impulsivity. Eur J Personal. 2005;19:559–74. Blechert J, Lender A, Polk S, Busch NA, Ohla K. Food-Pics_Extended—An Image Database for Experimental Research on Eating and Appetite: Additional Images, Normative Ratings and an Updated Review. Front Psychol. 2019;10:307. Toet A, Kaneko D, de Kruijf I, Ushiama S, van Schaik MG, Brouwer A-M, Kallen V, van Erp JBF. CROCUFID: A Cross-Cultural Food Image Database for Research on Food Elicited Affective Responses. Front Psychol. 2019;10:58. Foroni F, Pergola G, Argiris G, Rumiati RI. The FoodCast research image database (FRIDa). Front Hum Neurosci. 2013;7:51. Mestdagh M, Verdonck S, Piot M, Niemeijer K, Kilani G, Tuerlinckx F, Kuppens P, Dejonckheere E. (2023) m-Path: an easy-to-use and highly tailorable platform for ecological momentary assessment and intervention in behavioral research and clinical practice. Front Digit Health 5. van Beers JJ, Kaneko D, Stuldreher IV, Zech HG, Brouwer A-M. An Accessible Tool to Measure Implicit Approach-Avoidance Tendencies Towards Food Outside the Lab. Companion Publ. 2020 Int. Conf. Multimodal Interact. Virtual Event Netherlands: ACM; 2020. pp. 307–11. Ruf A, Neubauer AB, Ebner-Priemer U, Reif A, Matura S. Studying dietary intake in daily life through multilevel two-part modelling: a novel analytical approach and its practical application. Int J Behav Nutr Phys Act. 2021;18:130. Lenth RV. (2023) emmeans: Estimated Marginal Means, aka Least-Squares Means. Aulbach MB, van Alebeek H, Jones CM, Blechert J. Why we don’t eat as intended: Moderators of the short-term intention–behaviour relation in food intake. Br J Health Psychol. 2024. https://doi.org/10.1111/bjhp.12714 . Hofmann W, Van Dillen L. Desire: The New Hot Spot in Self-Control Research. Curr Dir Psychol Sci. 2012;21:317–22. Kavanagh DJ, Andrade J, May J. Imaginary Relish and Exquisite Torture: The Elaborated Intrusion Theory of Desire. Psychol Rev. 2005;112:446–67. May J, Andrade J, Kavanagh DJ, Hetherington M. Elaborated Intrusion Theory: A Cognitive-Emotional Theory of Food Craving. Curr Obes Rep. 2012;1:114–21. van Alebeek H, Röttger M, Kahveci S, Blechert J, Aulbach MB. (2024) The Only Constant is Change: Stable vs. Variable Aspects of Food Approach Bias Relate Differently to Food Craving and Intake. Appetite 107726. Stice E, Yokum S, Veling H, Kemps E, Lawrence NS. Pilot test of a novel food response and attention training treatment for obesity: Brain imaging data suggest actions shape valuation. Behav Res Ther. 2017;94:60–70. Yang Y, Morys F, Wu Q, Li J, Chen H. Pilot study of food-specific go/no-go training for overweight individuals: brain imaging data suggest inhibition shapes food evaluation. Soc. Cogn. Affect. Neurosci; 2021. Adams RC, Button KS, Hickey L, et al. Food-related inhibitory control training reduces food liking but not snacking frequency or weight in a large healthy adult sample. Appetite. 2021;167:105601. Saunders B, Milyavskaya M, More KR, Anderson J. Food cravings are associated with increased self-regulation, even in the face of strong instigation habits: A longitudinal study of the transition to plant-based eating. Appl Psychol Health Well-Being. 2025;17:e12629. Elliston KG, Ferguson SG, Schüz B. Personal and situational predictors of everyday snacking: An application of temporal self-regulation theory. Br J Health Psychol. 2017;22:854–71. Schumacher SE, Kemps E, Tiggemann M. Bias modification training can alter approach bias and chocolate consumption. Appetite. 2016;96:219–24. Houben K, Jansen A. Training inhibitory control. A recipe for resisting sweet temptations. Appetite. 2011;56:345–9. Veling H, Aarts H, Stroebe W. Stop signals decrease choices for palatable foods through decreased food evaluation. Front Psychol. 2013;4:1–7. Herman CP, Polivy J. Restraint Eating. Philadelphia: In: Obesity. Saunders; 1980. pp. 208–25. Adams RC, Chambers CD, Lawrence NS. Do restrained eaters show increased BMI, food craving and disinhibited eating? A comparison of the Restraint Scale and the Restrained Eating scale of the Dutch Eating Behaviour Questionnaire. R Soc Open Sci. 2019;6:190174. Herman CP, Mack D. Restrained and unrestrained eating. J Pers. 1975;43:647–60. Ruderman AJ. Dietary restraint: a theoretical and empirical review. Psychol Bull. 1986;99:247. van Alebeek H, Veling H, Blechert J. Disentangling go/no-go from motivational orientation to foods: Approaching is more than just responding. Food Qual Prefer. 2023;106:104821. Yang Y, Shields GS, Wu Q, Liu Y, Chen H, Guo C. Cognitive training on eating behaviour and weight loss: A meta-analysis and systematic review. Obes Rev. 2019. https://doi.org/10.1111/obr.12916 . Van Dessel P, De Houwer J, Gast A. Approach–Avoidance Training Effects Are Moderated by Awareness of Stimulus–Action Contingencies. Pers Soc Psychol Bull. 2016;42:81–93. Van Dessel P, Hughes S, De Houwer J. How Do Actions Influence Attitudes? An Inferential Account of the Impact of Action Performance on Stimulus Evaluation. Personal Soc Psychol Rev. 2019;23:267–84. Van Dessel P, Gawronski B, Smith CT, De Houwer J. Mechanisms underlying approach-avoidance instruction effects on implicit evaluation: Results of a preregistered adversarial collaboration. J Exp Soc Psychol. 2017;69:23–32. Masterton S, Hardman CA, Jones A. Don’t stop believing’: The role of training beliefs in cognitive bias modification paradigms. Appetite. 2022;174:106041. Michie S, Richardson M, Johnston M, Abraham C, Francis J, Hardeman W, Eccles MP, Cane J, Wood CE. The behavior change technique taxonomy (v1) of 93 hierarchically clustered techniques: Building an international consensus for the reporting of behavior change interventions. Ann Behav Med. 2013;46:81–95. Footnotes We had preregistered further analyses and report those in the supplementary materials. Due to apparently interchanged data for height and weight of one participant, we excluded this participant from the BMI computation. N = 71 Note that the parameter estimates for the hurdle part relate to the probability that craving is zero, that is, the probability of craving being absent. As pre-registered, we calculated interaction models with additional participant-level, stimulus-level and training-level variables for the changes from pre- to post training. An overview of effects as well as all model results can be found in the supplements. In this paper, we focus on the variables with consistent effects (i.e., effects for which we found evidence within different outcome variables or different similar moderator variables). Additional Declarations No competing interests reported. Supplementary Files Supplements.html consortdiagram.doc Cite Share Download PDF Status: Published Journal Publication published 28 Nov, 2025 Read the published version in International Journal of Behavioral Nutrition and Physical Activity → Version 1 posted Editorial decision: Revision requested 04 Aug, 2025 Reviews received at journal 30 Jul, 2025 Reviewers agreed at journal 04 Jul, 2025 Reviews received at journal 25 Jun, 2025 Reviewers agreed at journal 24 Jun, 2025 Reviewers invited by journal 29 May, 2025 Editor assigned by journal 27 May, 2025 Submission checks completed at journal 27 May, 2025 First submitted to journal 26 May, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6751171","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":463995917,"identity":"e5086b89-3d85-4bee-8086-07835f2180d5","order_by":0,"name":"Matthias Burkard Aulbach","email":"data:image/png;base64,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","orcid":"","institution":"Paris-Lodron-University of Salzburg","correspondingAuthor":true,"prefix":"","firstName":"Matthias","middleName":"Burkard","lastName":"Aulbach","suffix":""},{"id":463995918,"identity":"4822e384-d0b3-436e-bda5-ac8c7012a680","order_by":1,"name":"Mareike Roettger","email":"","orcid":"","institution":"Paris-Lodron-University of Salzburg","correspondingAuthor":false,"prefix":"","firstName":"Mareike","middleName":"","lastName":"Roettger","suffix":""},{"id":463995919,"identity":"3f8b2374-c1d5-4567-b32e-3c3c788047d2","order_by":2,"name":"Hannah Alebeek","email":"","orcid":"","institution":"Paris-Lodron-University of Salzburg","correspondingAuthor":false,"prefix":"","firstName":"Hannah","middleName":"","lastName":"Alebeek","suffix":""},{"id":463995920,"identity":"d4538c3a-8e01-4243-9d6f-73d95ad44f9d","order_by":3,"name":"Sercan Kahveci","email":"","orcid":"","institution":"Paris-Lodron-University of Salzburg","correspondingAuthor":false,"prefix":"","firstName":"Sercan","middleName":"","lastName":"Kahveci","suffix":""},{"id":463995921,"identity":"e52f9dbd-8c72-49c6-9d58-6f4742956f45","order_by":4,"name":"Jennifer Schmidt","email":"","orcid":"","institution":"FH Münster University of Applied Sciences","correspondingAuthor":false,"prefix":"","firstName":"Jennifer","middleName":"","lastName":"Schmidt","suffix":""},{"id":463995922,"identity":"5ab13a7a-fb9c-4856-a51d-dea965dd90fe","order_by":5,"name":"Jens Blechert","email":"","orcid":"","institution":"Paris-Lodron-University of Salzburg","correspondingAuthor":false,"prefix":"","firstName":"Jens","middleName":"","lastName":"Blechert","suffix":""}],"badges":[],"createdAt":"2025-05-26 13:08:53","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6751171/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6751171/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12966-025-01836-2","type":"published","date":"2025-11-28T15:57:37+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":83813199,"identity":"bc6886f2-8573-4fb4-ad36-cb0f73d73af2","added_by":"auto","created_at":"2025-06-03 07:19:27","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":36605,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMobile AAI/ AAA and Procedure. \u003c/strong\u003eNote. AAI = Approach-Avoidance Intervention, AAA = Approach-Avoidance Assessment Task. Panel A adapted from [13].\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6751171/v1/e5a992fa0cb687c3c1e8ba36.jpg"},{"id":83813198,"identity":"870999bb-05dd-45cf-8811-174bbc441a93","added_by":"auto","created_at":"2025-06-03 07:19:27","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":23953,"visible":true,"origin":"","legend":"\u003cp\u003eLong-term effects on bias strength as well as craving intensity for decrease-foods\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNote\u003c/em\u003e. These plots were created based on the raw data. Panel A shows the means of bias strength, Panel B shows the means of craving intensity. The error bars show the standard errors of these means.\u003c/p\u003e","description":"","filename":"Picture2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6751171/v1/6908528dcef0050c8efa9ceb.jpg"},{"id":83813202,"identity":"3904e4ee-5a91-4168-9b10-895ec5b80c33","added_by":"auto","created_at":"2025-06-03 07:19:27","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":43659,"visible":true,"origin":"","legend":"\u003cp\u003eModeration of training effectiveness by restrained eating and perceived self-regulatory success. The graphs show the means and standard errors of the means from the raw data.\u003c/p\u003e","description":"","filename":"Picture3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6751171/v1/17949774604c4512e4613bca.jpg"},{"id":97178740,"identity":"65bb558d-719f-4c8d-9736-e4d7517db067","added_by":"auto","created_at":"2025-12-01 16:13:11","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1001583,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6751171/v1/ecb7fb44-43a5-4c50-bb32-73a58f9c8663.pdf"},{"id":83813205,"identity":"8487ebcc-e36f-4ca5-9631-70633f138922","added_by":"auto","created_at":"2025-06-03 07:19:28","extension":"html","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":5612054,"visible":true,"origin":"","legend":"","description":"","filename":"Supplements.html","url":"https://assets-eu.researchsquare.com/files/rs-6751171/v1/5e6aefd6fe90117772495202.html"},{"id":83814495,"identity":"ad5d4475-a259-4f9d-af95-20c1b210d8fa","added_by":"auto","created_at":"2025-06-03 07:27:27","extension":"doc","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":16820,"visible":true,"origin":"","legend":"","description":"","filename":"consortdiagram.doc","url":"https://assets-eu.researchsquare.com/files/rs-6751171/v1/1de28622bdc725ba24e96adb.doc"}],"financialInterests":"No competing interests reported.","formattedTitle":"Effectiveness of a Smartphone-Delivered Approach-Avoidance Intervention in Dietary Behavior - A Randomized Controlled Trial","fulltext":[{"header":"Background","content":"\u003cp\u003eOverweight and obesity have become major health concerns in many countries [1, 2], with the main cause for overweight being an imbalance between energy input and output [3, 4]. Calorie-dense foods are readily available in obesogenic environments, which often leads to excessive consumption and insufficient intake of healthier foods such as fruits and vegetables. While many people know that diet, overweight, and health are related, and many people intend to adhere to healthier diets [5], they often fail to actually follow through on these intentions [6].\u003c/p\u003e\n\u003cp\u003eMany psychological models struggle to explain this phenomenon, as they model behavior mainly as originating from thoroughly reflected intentions [7–9]. Dual-process models address this shortcoming by emphasizing that behavior is not only determined by such reflective processes but also by impulsive ones [10, 11]. These models have led to the development of computer-based reaction time (RT) tasks that aim to measure such impulsive determinants of behavior. One such task is the Approach-Avoidance Assessment task (AAA), in which participants respond to images of different categories (e.g., palatable vs non-palatable foods or objects) on a screen by performing movements that represent approach or avoidance, such as pulling or pushing a joystick [12] or moving a smartphone towards or away from their body [13]. There are often differences in RTs between approach and avoidance trials, and this difference is often greater for one stimulus category than another; this latter difference is termed an “approach bias”. This approach bias is stronger towards foods in individuals with high trait food craving [14] and towards chocolate in those with currently strong state chocolate craving [15]. It is also related to increased food consumption in impulsive individuals [16] and in people prone to external or emotional eating [17]. Further, approach bias relates to participants’ desire to eat specific foods [18, 19].\u003c/p\u003e\n\u003ch3\u003ePast Approach-Avoidance Interventions and their Pitfalls\u003c/h3\u003e\n\u003cp\u003eMoving from observation to intervention, researchers have devised ways to intervene on consumption behavior using an adapted task setup. In Approach-Avoidance \u003cem\u003eInterventions\u003c/em\u003e (AAI), images are systematically paired with specific reactions, usually unbeknownst to the participant: participants for example always have to approach healthier and avoid less healthy foods. Such interventions haven been shown to effectively reduce food craving [20] and intake of trained foods, at least for some people and under certain circumstances [21–25]: For example, interventions are more effective for those who struggle with self-regulation [23, 26]. Not all interventions have been effective however [21], spurring discussions about reasons for a lack of effectiveness, and three prominent aspects have arisen, which we discuss in the following.\u003c/p\u003e\n\u003cp\u003eFirst, the exact instructions and setup of the task differ between studies such that participants either approach or avoid based on the stimulus category (“relevant-feature” AAA/AAI) or some other aspect of the image (e.g. its spatial orientation; “irrelevant-feature” AAA/AAI). Since a relevant-feature task focuses attention on the image, it performs better as an approach bias measure than the irrelevant-feature version [27, 28]. While data for interventions are rare [29–31], it seems likely that relevant-feature AAIs would be superior to an irrelevant-feature version due to this aspect of attentional focus. Second, many studies present pre-selected images that are not personalized to the participant. This likely dilutes training effectiveness, as some participants may not like a given food, or have no intention of changing their intake of it. Therefore, allowing participants to select images of foods they would like to eat less should help in enhancing effectiveness. However, intentions for behavior change have mostly been ignored in this literature. As a third reason for mixed findings, most studies have delivered only single intervention sessions in a laboratory or online; delivering an AAI repeatedly should enhance effectiveness [32, 33]. Smartphones easily enable this and have the further advantage of bringing the intervention closer to day-to-day consumption behavior.\u003c/p\u003e\n\u003cp\u003eIn addition to these three potential reasons limiting the effectiveness of AAIs, most AAI studies to date mainly focused on \u003cem\u003ereducing\u003c/em\u003e the intake of unhealthy foods, somewhat neglecting the fact that attaining a healthier diet also requires \u003cem\u003eincreasing\u003c/em\u003e the intake of healthier foods. This is not only important for health reasons, as many people fail to consume the recommended amount of fruits and vegetables [34]; it is also important from a behavior change perspective, as it is more effective to substitute an unwanted behavior with an alternative, similar behavior, than it is to simply suppress the unwanted behavior [35–37]. AAIs are promising in this regard as they can be programmed to train both approach of healthy foods and avoidance of unhealthy foods simultaneously, and could thus facilitate such behavior substitution [38].\u003c/p\u003e\n\u003cdiv id=\"Sec3\"\u003e\n \u003ch2\u003eThe current study\u003c/h2\u003e\n \u003cp\u003eIn the current study we deploy a mobile AAI that addresses these shortcomings of prior studies, and we investigate whether this AAI is able to effectively support dietary changes. Specifically, we deliver a smartphone-based multisession relevant-feature AAI that allows participants to select which food images they will work with during the study based on their intention to change how often they eat them. This results in two stimulus sets: individualized foods that participants want to eat more (“increase-foods”) and those they want to eat less (“decrease-foods”). The AAI is accompanied by twice-daily ecological momentary assessment (EMA) that includes measures of intake, food craving, and other relevant day-level variables. To control for potential placebo effects that may come with repeated performance of an eating-related behavioral task, we compare this to an active control in which increase-foods and decrease-foods are approached and avoided equally often. In the assessment of approach bias, we include neutral control stimuli to ensure that approach bias for increase-foods can be separately interpreted/calculated from the approach bias of decrease-foods. As preregistered, we expect that participants in the AAI will display stronger increases in consumption of, craving for, and approach bias towards increase-foods and stronger decreases in consumption of, craving for, and approach bias towards decrease-foods compared to the control condition. We also explored potential person-level moderators of the effect to determine which population benefits most from AAI.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Methods","content":"\u003cp\u003eThe full methods of this randomized controlled trial (RCT) are outlined in the corresponding protocol paper [39] and the preregistration https://osf.io/4k3q9/. Here, we only describe the aspects of the trial relevant to this study and refer the reader to those documents for more detail.\u003c/p\u003e\n\u003ch3\u003eParticipants\u003c/h3\u003e\n\u003cp\u003eWe recruited 157 participants (132 women, 25 men, \u003cem\u003eM\u003c/em\u003e\u003csub\u003eage\u003c/sub\u003e [\u003cem\u003eSD\u003c/em\u003e] \u003cem\u003e=\u003c/em\u003e 27.6 [9.73]; see protocol paper for power analysis [39]) through university and social networks, and through word of mouth. Participants had to be between 18 and 60 years old, not pregnant, and report living without an eating disorder. Participants needed to have a goal of changing their eating behavior in the upcoming weeks. Six participants were excluded from all analyses as their AAI/AAA sessions were all either missing, or excluded (see below for exclusion criteria for AAI/AAA sessions; see consort flowchart in the supplementary materials). One participant was excluded from all analyses that had approach bias scores as the outcome of interest, as they seemed to have technical issues with the smartphone application.\u003c/p\u003e\n\u003ch3\u003eQuestionnaires\u003c/h3\u003e\n\u003cp\u003eFor each scale we report both Cronbach’s alpha using the R-package psych [40], and McDonald’s omega with bootstrapped 95% confidence intervals using the R-package MBESS [41].\u003c/p\u003e\n\u003ch3\u003ePerceived Self-Regulatory Success in Dieting Scale (PSRS)\u003c/h3\u003e\n\u003cp\u003eThe Perceived Self-Regulatory Success in Dieting Scale (PSRS) [42] is used to differentiate between successful and unsuccessful dieters. It consists of three items rated on a 7-point Likert scale with higher scores indicating more successful self-regulation. Internal consistency was questionable: ω = .69 [.56, .76], α = .68.\u003c/p\u003e\n\u003cdiv id=\"Sec8\"\u003e\n \u003ch2\u003eDutch Eating Behavior Questionnaire restrained eating (DEBQ-res) and external eating (DEBQ-ext) subscales\u003c/h2\u003e\n \u003cp\u003eFrom the Dutch Eating Behavior Questionnaire (DEBQ) [43], we administered the subscales of restrained eating (DEBQ-res) and external eating (DEBQ-ext). Each subscale consists of 10 items and is rated on a 5-point Likert scale from 1 (never) to 5 (very often). Internal consistency was good for both the restrained eating subscale: ω = .89 [.86, .91], α = .89; and the external eating subscale: ω = .85 [.80, .89], α = .85.\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003eUPPS Impulsive Behavior Scale (UPPS)\u003c/h3\u003e\n\u003cp\u003eThe UPPS impulsive behavior scale (UPPS) [44] consists of four subscales that each measure a facet of impulsivity: urgency, lack of pre-meditation, lack of perseverance, and sensation seeking. Each subscale consists of five statements rated on a 4-point Likert scale from 1 (strongly disagree) to 4 (strongly agree). Higher scores on an UPPS subscale are interpreted as a higher level of impulsivity in the respective impulsivity facet. Internal consistency was questionable for the urgency subscale, ω = .68 [.57, .77], α = .68; but acceptable for the other three subscales, pre-meditation: ω = .74 [.65, .81], α = .73; perseverance: ω = .79 [.72, .85], α = .79; sensation seeking: ω = .73 [.65, .79], α = .70.\u003c/p\u003e\n\u003ch3\u003eMaterials and procedure\u003c/h3\u003e\n\u003cp\u003e Participants gave informed consent online, followed by a range of questionnaires including demographics. After the questionnaires, they rated 90 preselected images of foods and drinks from the food.pics [45] and CROCUFID [46] databases, photographs taken by the study authors, and license-free online stock photograph databases. The ratings included two questions for each image: “In the last three weeks, on how many days have you eaten/drunk this food/drink?” (recent intake) and “In the next three weeks, on how many days do you want to eat/drink this food/drink?” (intended intake). Answers to both questions were indicated on a slider ranging from 0 to 21 days. The six images with the largest (recent intake \u0026gt; intended intake: “decrease-food”) and the six images with the smallest difference (recent intake \u0026lt; intended intake: “increase-food”) between the two scales were included in the study, with a random selection of four images per category included in the AAA/AAI. A randomly selected 8 out of 12 images of office items from the food.pics [45] and FRIDa [47] databases served as control stimuli. We excluded participants that did not have six decrease-foods which they consumed at least twice weekly. At this point, an R-script (The function sample randomly outputs the number ‘1’ or ‘2’ which correspond to the conditions.) randomized participants to either the intervention or control group with the condition unknown to the study team.\u003c/p\u003e\n\u003cp\u003eNext, a member of the research team conducted a setup call with each participant to install and explain the use of two smartphone applications: m-path [48] for EMA, and the app to conduct the AAA/AAI. The remainder of the study consisted of four phases (Fig.\u0026nbsp;1): a baseline phase on days 1–4, an intervention phase on days 5–16, a post-intervention phase on days 17–20, and a follow-up phase 4 weeks after day 20. Participants completed a single measurement-AAA during the baseline phase (day 4) and also during the post-intervention phase (day 17), while performing an AAI or AAA every other day during the intervention phase, depending on condition (days 6, 8, 10, 12, 14, and 16). EMA was collected throughout these three phases. The follow-up consisted of a single EMA prompt and AAA measurement (day 48, four weeks after the last post-intervention day).\u003c/p\u003e\n\u003cp\u003eFor the AAA/AAI, we used a modified smartphone application [13, 49]. In this task, users held the phone horizontally in front of them while responding to stimuli shown on the screen. After a fixation dot for 1500 ms at the start of the trial, the app presented a food or object stimulus. Depending on image type (food or object), participants had to move the phone towards or away from themselves, representing an approach or avoidance response, respectively.\u003c/p\u003e\n\u003cp\u003eTwo types of Approach-Avoidance Tasks were administered: measurement-AAAs and training-AAIs. Each measurement-AAA block consisted of 24 trials each, while each training-AAI/AAA consisted of 16 trials per block. Each session consisted of 4 blocks, with each block preceded by 4 practice trials. Before each block, instructions were presented, with blocks 1 and 3 instructing participants to approach foods and avoid objects (“approach-food-blocks”) and blocks 2 and 4 giving the opposite instruction (“avoid-food-blocks”). In the AAI (intervention condition), the approach-food-blocks exclusively featured increase-foods and objects, and the avoid-food-blocks exclusively featured decrease-foods and objects; thus, increase-foods were always approached while decrease-foods were always avoided. In the AAA (control condition), all food images were presented in each block. During the intervention phase (days 5–17), after the mid-day prompt (content not relevant here), participants were prompted to conduct a session of AAA/AAI on every other day (days 6, 8, 10, 12, 14, 16) resulting in six AAA/AAI sessions. In addition, all participants conducted an AAA on days 4 and 17 and during the follow-up.\u003c/p\u003e\n\u003cp\u003eThroughout the study, participants followed a twice-daily EMA schedule, with the mid-day prompt just before their typical lunch time and the other at the end of the day (timing agreed upon individually) with all EMA items using a virtual 0-100 slider (where not reported differently). Relevant items measured \u003cem\u003ecraving\u003c/em\u003e for (“How strongly have you been craving this food today?”) and \u003cem\u003eintake\u003c/em\u003e of (“How much have you eaten of this food today?”) each of the included foods.\u003c/p\u003e\n\u003cdiv id=\"Sec11\"\u003e\n \u003ch2\u003eData preprocessing and analysis\u003c/h2\u003e\n \u003cdiv id=\"Sec12\"\u003e\n \u003ch2\u003eApproach bias\u003c/h2\u003e\n \u003cp\u003eTo pre-process the AAA data, we removed trials with non-responses, movements in the wrong direction, or RTs above 2000 or below 200 ms. After this, exclusions continued within each session within each participant, as we removed trials that deviated more than 3 SDs from the mean. AAA sessions with more than 25% of trials missing or removed were then excluded from analysis entirely.\u003c/p\u003e\n \u003cp\u003eSeparately for approach and avoidance trials as well as for sessions, RTs were averaged across AAA blocks and all object stimuli. For foods, RTs were averaged across AAA blocks only. The average approach or avoid response for objects on a session was subtracted from stimulus-specific food approach or avoidance response on that session to achieve single-difference approach and avoidance scores:\u003c/p\u003e\n \u003cp\u003eStimulus-specific approach = [food-specific approach]-[average object approach]\u003c/p\u003e\n \u003cp\u003eStimulus-specific avoidance = [food-specific avoidance]-[average object avoidance]\u003c/p\u003e\n \u003cp\u003eDouble-difference scores were used to achieve a full bias score per food stimulus and session:\u003c/p\u003e\n \u003cp\u003e[food-specific avoidance]-[food-specific approach]-[average object avoidance]-[average object approach]. Hence, positive values imply the food was approached faster than it was avoided, relative to the difference between approach and avoidance RTs for objects.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec13\"\u003e\n \u003ch2\u003eAnalysis\u003c/h2\u003e\n \u003cp\u003eAs multiple datapoints are nested within participants and stimuli, we used multilevel models (MLM) to assess the hypotheses. Since our outcome measures craving and intake displayed strong skew and many zeroes, we used Bayesian multilevel two-part hurdle models [50]. In brief, these regression models consist of a hurdle part akin to logistic regression that estimates the probability of the outcome not being zero, and a continuous part following a gamma distribution that estimates the size of the non-zero values. The hurdle model therefore allows to separately determine which variables predict whether intake or craving occurred and, if so, how much was eaten/how strong the craving was. Since these models flexibly deal with missing values they approximate an intention to treat analysis with multiple imputation.. Note that missing data were rare and similarly spread across groups (see Table\u0026nbsp;1), making selective drop out unlikely.\u003c/p\u003e\n \u003cp\u003eFor each model, we report the regression coefficient of interest as well as both the 95% and the 89% highest density interval (HDI), reflecting the most credible values of the respective model parameter. This is equivalent to a 5% or 11% alpha level, respectively. If the 89%-HDI does not include 0, we describe the model parameter as being estimated as “significantly” above or below 0.\u003c/p\u003e\n \u003cp\u003eAnalyses include only data from the baseline and post-intervention phase (4 days each) and include all 12 food stimuli. In separate models, we respectively predicted the stimulus-level intake, craving, and approach bias scores for increase and decrease-foods separately using the interaction between Intervention group (0 = 50/50 AAA vs. 1 = 100/0 AAI) and time (0 = Pre vs. 1 = Post) according to this formula:\u003c/p\u003e\n \u003cp\u003eIntake/Craving/Approach Bias\u003csub\u003eincrease/decrease−foods\u003c/sub\u003e ~ Intervention Group * time + (time | Subject) + (time | Stimulus)\u003c/p\u003e\n \u003cp\u003eTo assess intervention effects to follow-up, we then ran equivalent models containing only data from post-intervention and follow-up for all outcomes. For moderation analyses we added the relevant moderator and all interaction terms to the models comparing pre- to post-intervention values. We computed post-hoc contrasts using the R-package emmeans [51] for all models where we found evidence of an effect to assess between- and within-condition differences. We report the estimate as well as the 95% HDI for the contrasts of interest.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDescriptive Statistics and EMA Compliance by group\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTraining group\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePlacebo control group\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSample size\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e79\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eDemographics\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWomen (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e60 (83.33%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e68 (86.08%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge in years (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e25.61 (8.16)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e29.66 (10.85)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI (kg/m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e23.49 (6.22)\u003ca class=\"FNLink\" href=\"#Fn2\" id=\"#FNLinkFn2\"\u003e\u003c/a\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e24.47 (3.95)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eCompliance\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eN completed training sessions (out of 6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5.51\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5.57\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBaseline EMA compliance (mean N completed questionnaires of 4 assessment points)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.72\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePost-intervention EMA compliance (mean N completed questionnaires of 4 assessment points)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.58\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.60\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eMeans (SDs) for questionnaire scores\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePSRS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.56 (1.11)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.53 (1.23)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDEBQ-res\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.31 (0.73)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.60 (0.75)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDEBQ-ext\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.37 (0.70)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.39 (0.61)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUPPS lack of perseverance\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.06 (0.57)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.90 (0.64)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUPPS lack of premeditation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.04 (0.54)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.98 (0.56)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUPPS urgency\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.20 (0.56)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.28 (0.53)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUPPS sensation seeking\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.53 (0.64)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.34 (0.73)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eMeans (SDs) for baseline phase (4 days)\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIntake decrease-foods (Slider rating 0-100)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e19.3 (11.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e20.2 (11.7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIntake increase-foods (Slider rating 0-100)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e16.6 (11.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e14.1 (10.6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCraving decrease-foods (Slider rating 0-100)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e32.7 (18.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e35.3 (16.7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCraving increase-foods (Slider rating 0-100)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e23.8 (14.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e25.3 (15.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eMeans (SDs) for intervention perception\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExpectancy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.52 (0.96)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.41 (1.00)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eContingency perception for trained decrease-foods\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e48.9 (20.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e52.5 (14.8)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eContingency perception for trained increase-foods\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e31.2 (19.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e42.0 (19.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eWe checked for differences in compliance between the two groups with two-sample t-tests. There was no difference in baseline EMA compliance (\u003cem\u003et\u003c/em\u003e(143.09) = -0.82, p\u0026thinsp;=\u0026thinsp;.41), training compliance (\u003cem\u003et\u003c/em\u003e(149)\u0026thinsp;=\u0026thinsp;0.32, p\u0026thinsp;=\u0026thinsp;.75) or post-intervention EMA compliance (\u003cem\u003et\u003c/em\u003e(145.99)\u0026thinsp;=\u0026thinsp;0.16, p\u0026thinsp;=\u0026thinsp;.87).\u003c/p\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eIntervention effects from pre- to post-intervention and follow-up\u003c/h2\u003e \u003cdiv id=\"Sec16\" class=\"Section3\"\u003e \u003ch2\u003eApproach Bias: reduced for decrease-foods\u003c/h2\u003e \u003cp\u003eFor \u003cem\u003edecrease-food\u003c/em\u003es, the group \u0026times; time interaction for bias scores from baseline assessment to post-assessment was significant on an 89% HDI level (n\u0026thinsp;=\u0026thinsp;144, b = -38.10; 89% HDI [-73.82, -0.39]; 95% HDI [-81.88, 8.29]). Bias for decrease-foods decreased more in the intervention group than in the control group (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, panel A). There was no significant interaction for \u003cem\u003eincrease\u003c/em\u003e foods (n\u0026thinsp;=\u0026thinsp;144, b = -15.92; 89% HDI [-59.81, 28.95]; 95% HDI [-67.82, 40.36]). The models examining changes from post-intervention to follow-up also showed a significant effect for \u003cem\u003edecrease-food\u003c/em\u003es on an 89% HDI level (n\u0026thinsp;=\u0026thinsp;141, b\u0026thinsp;=\u0026thinsp;35.06; 89% HDI. [0.98, 66.89]; 95% HDI [-5.61, 75.64]), indicating a stronger increase of bias values in the intervention group than in the control group. Again, there was no effect for \u003cem\u003eincrease-food\u003c/em\u003es (n\u0026thinsp;=\u0026thinsp;142, b\u0026thinsp;=\u0026thinsp;11.06; 89% HDI. [-24.84, 49.27]; 95% HDI [-32.84, 57.55]). Post-hoc contrasts did not show evidence for a group difference in bias scores for \u003cem\u003edecrease-food\u003c/em\u003es at the post-intervention measurement (89% HDI [-3.18, 51.09]; 95% HDI [-8.53, 58.0]) or at follow-up measurement (89% HDI [-39.53, 15.16]; 95% HDI [-45.37, 20.8]).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eCraving: reduced for decrease-foods\u003c/h2\u003e \u003cp\u003eWe separately report changes in the likelihood of reporting (no) craving (hurdle part) \u003ca class=\"FNLink\" href=\"#Fn3\" id=\"#FNLinkFn3\"\u003e\u003c/a\u003e, and changes in craving intensity when any craving above zero is reported (continuous part). The intervention group showed stronger reductions in craving intensity for \u003cem\u003edecrease-foods\u003c/em\u003e compared to the control group from pre- to post-intervention, similar to the effect in bias scores. (n\u0026thinsp;=\u0026thinsp;151, continuous part: b = -0.19; 89% HDI [-0.29, -0.08]; 95% HDI [-0.31, -0.05], Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, panel B). This difference was not present for the probability of experiencing craving (n\u0026thinsp;=\u0026thinsp;151, hurdle part: b = -0.25; 89% HDI [-0.62, 0.12]; 95% HDI [-0.70, 0.20]). Regarding \u003cem\u003eincrease-food\u003c/em\u003es, again, there was neither a significant time \u0026times; condition interaction for craving intensity nor for the probability of experiencing craving (n\u0026thinsp;=\u0026thinsp;151, hurdle part: b = -0.31; 89% HDI [-0.70, 0.06]; 95% HDI [-0.77, 0.16]; continuous part: b = -0.04; 89% HDI [-0.19, 0.09]; 95% HDI [-0.21, 0.12]).\u003c/p\u003e \u003cp\u003eWhen looking at the effects from post-intervention to follow-up 4 weeks later, we found a significant interaction for craving intensity (but not probability) for \u003cem\u003edecrease-foods\u003c/em\u003e on a 89% HDI level (n\u0026thinsp;=\u0026thinsp;149, hurdle part: b\u0026thinsp;=\u0026thinsp;0.06; 89% HDI [-0.39, 0.51]; 95% HDI [-0.51, 0.61]; continuous part: b = -0.13; 89% HDI [-0.24, -0.003]; 95% HDI [-0.27, 0.02]), indicating that the intervention group\u0026rsquo;s craving intensity increased significantly more than the control group\u0026rsquo;s from post-intervention to follow-up. We found no effects for increase-foods (n\u0026thinsp;=\u0026thinsp;149, hurdle part: b = -0.03; 89% HDI [-0.59, 0.27]; 95% HDI [-0.70, 0.36]; continuous part: b = -0.16; 89% HDI [-0.15, 0.10]; 95% HDI [-0.18, 0.12].\u003c/p\u003e \u003cp\u003ePost-hoc contrasts showed evidence for a difference in the craving intensity between the control group and the intervention group at the post-intervention measurement (89% HDI [0.11, 0.35]; 95% HDI [.08, .38]), but the apparent difference in the raw values at follow-up did not prove statistically reliable (89% HDI [-0.04, 0.24]; 95% HDI [-.07, .27]). The craving intensity within the intervention group decreased from baseline to post-intervention measurement (estimated difference\u0026thinsp;=\u0026thinsp;0.29; 89% HDI [0.21, 0.37], 95% HDI [.19, .38]) and increased again from post-intervention to follow-up measurement (89% HDI [0.13, 0.31]; 95% HDI [.11, .34]).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eIntake: no group differences\u003c/h2\u003e \u003cp\u003eThere was no significant time \u0026times; condition interaction on intake for both \u003cem\u003eincrease-food\u003c/em\u003es or \u003cem\u003edecrease-food\u003c/em\u003es for hurdle and continuous parts both in the pre-post and the post to follow-up-analyses. Test statistics and plots showing these results can be found in the supplementary materials.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eModeration of craving effects by participant characteristics: restraint and perceived self-regulatory success matter\u003c/h2\u003e \u003cp\u003eParticipants\u0026rsquo; restrained eating and perceived self-regulatory success moderated the baseline to post-training change in craving intensity for \u003cem\u003edecrease-foods\u003c/em\u003e (n\u0026thinsp;=\u0026thinsp;151, three-way interaction for restrained eating in the continuous part: b = -0.19; 89% HDI [-0.32, -0.05]; 95% HDI [-0.36, -0.03], Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e (top); and for perceived self-regulatory success: b\u0026thinsp;=\u0026thinsp;0.13; 89% HDI [0.04, 0.22]; 95% HDI [0.03, 0.24], Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e (bottom)). There was no evidence for an interaction in the hurdle part (restrained eating: b\u0026thinsp;=\u0026thinsp;0.41; 89% HDI [-0.05, 0.90]; 95% HDI [-0.20, 0.98], perceived self-regulatory success: b = -0.13; 89% HDI [-0.43, 0.20]; 95% HDI [-0.53, 0.24]). The intervention was most effective in reducing craving intensity for participants with high restrained eating and low past perceived self-regulatory success. These two questionnaires were not significantly correlated; \u003cem\u003er\u003c/em\u003e (149)\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;.17, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.091.\u003ca class=\"FNLink\" href=\"#Fn4\" id=\"#FNLinkFn4\"\u003e\u003c/a\u003e The analyses including the UPPS and external eating scales as well as those using other outcomes did not yield significant results (see supplementary materials).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this RCT, we found that a multisession mobile AAI reduced craving intensity for foods that participants wanted to eat less of. We further found weaker evidence that the intervention reduced approach bias for these foods as well. These effects did not transfer to intake, nor did we find effects on foods that participants wanted to eat more of. However, we found that restrained eaters and those with low perceived self-regulatory success in dieting showed the largest reductions in craving intensity for decrease-foods, indicating they might benefit most from the intervention.\u003c/p\u003e \u003cp\u003eThe current study was designed to address several shortcomings of earlier research on AAIs. In particular, we expected that the repeated delivery of the intervention (6 sessions/384 trials) would significantly boost its effectiveness. Indeed, the achieved reductions in craving intensity seem promising, as craving is a common obstacle to successful dietary goal pursuit [\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e, \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e] and is perceived as unpleasant, especially in dieters [\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e, \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e]. Like craving, approach bias for decrease-foods also decreased from pre- to post-intervention. This is in line with findings that craving and approach bias are coupled [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e] as part of a cognitive-behavioral pattern that prepares for ingestion [\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e]. Similar tasks have also been shown to reduce food valuation based on neuroimaging [\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e, \u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e]. In the current study, however, the achieved reductions in craving and approach bias did not result in intake reductions, in line with earlier studies that showed AAIs and similar tasks reduce bias or food liking without changing intake [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e]. Several factors might have prevented the craving and bias reduction from translating into consumption reduction: recent research has shown that food cravings are associated with higher self-regulatory efforts, indicating that individuals adjust their self-regulatory efforts to experienced craving strength [\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e]. If the AAI reduced craving strength, participants might, in consequence, have reduced subsequent self-regulatory effort, leading to similar consumption levels as pre-intervention. In addition, eating behavior is context dependent: availability of craved foods, homeostatic meal planning, and social context are factors that influence (overt) eating [\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e, \u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e], but not (covert) craving. Studies that have successfully changed actual eating behavior using the AAI and similar tasks (such as the Go-/No-Go task) have typically been conducted in controlled laboratory settings and thus ignored such real-life circumstances imminent in our EMA design [\u003cspan additionalcitationids=\"CR63\" citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eWe further found two eating behavior traits to moderate the intervention effects on decrease-food craving intensity: perceived self-regulatory success in dieting [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e] and eating restraint [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e, \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e]. Those who reported difficulties in regulating their dieting showed the strongest effects, indicating that AAI can support dietary goal pursuit of individuals who need it the most [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e], and may be better suited for that purpose than for enhancing the dieting capabilities of those who are already successful. In light of dual-process models, this might be because those struggling with self-regulation often experience goal-incongruent behavioral impulses and AAIs might help to reduce such impulses. Additionally, independent from self-regulatory success, restrained eaters showed stronger effects than unrestrained eaters, which underlines the importance of having a dietary goal for AAI to have an effect [\u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e]: Restrained eaters care strongly about what they (don\u0026rsquo;t) eat. AAI might be able to support this dietary goal through reducing craving. This seems particularly important as restrained eaters report stronger tendencies to experience craving on a trait level [\u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e66\u003c/span\u003e] and respond strongly to perceived breaches of their dietary rules [\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e, \u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e]. While restraint eating is clearly not equivalent to caloric dieting, the latter likely brings about similar processes as in restrained eating (dieting goal) and thus future research could apply the present intervention trial to weight reduction dieting.\u003c/p\u003e \u003cp\u003e We had expected to find not only goal-congruent changes for decrease-foods but also in increase-foods, increasing the appeal and consumption of foods participants wanted to eat more of (e.g. for their health benefits), based on earlier findings [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e69\u003c/span\u003e, \u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e]. However, we found no main effects in this goal-congruent food category on any of our outcome variables. It might be that participants already chose foods they generally liked as their increase-foods and therefore, further increases in craving and approach bias might be hard to achieve.\u003c/p\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eFuture Directions\u003c/h2\u003e \u003cp\u003eIn addition to the analyses reported here, we had assessed other variables that have been suggested to work as moderators or mechanisms of the effect, including the awareness of contingencies between stimuli and required reactions [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e], expectations regarding training effects [\u003cspan additionalcitationids=\"CR73\" citationid=\"CR72\" class=\"CitationRef\"\u003e72\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e74\u003c/span\u003e], as well as transfer of training effects to non-trained stimuli. However, none of the analyses on these variables yielded conclusive results and we therefore refrain from commenting on these theoretical debates (see the supplementary materials for all analyses). However, we urge future researchers to design their studies explicitly to examine such potential mechanisms to improve our understanding of how AAI effects come about in applied settings.\u003c/p\u003e \u003cp\u003eWe make a number of propositions for future research into dietary enhancement through AAI. First, research may benefit from an exclusive focus on restrained eaters with low dietary success, since our training was most effective in reducing craving in this subsample. Second, it may help to get participants to commit to their dietary intentions, as participants in the current study were free to pursue or abandon their dietary intentions. Third, future research could study changes in food intake without the influence of food availability, by ensuring participants always have their decrease and increase-foods available. Fourth, we suggest studying the combination of AAIs with other behavior change techniques that focus on different mechanisms of eating behavior [\u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e75\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eStrengths and Limitations\u003c/h2\u003e \u003cp\u003eThis study used a rigorous design to test an intensive, multi-session, mobile version of a task in a field setting that has thus far mainly been tested in laboratory environments. Its strengths thus include (1) high ecological validity due to the mobile nature of the intervention and data assessment; (2) comparison with a closely matched, active control group within a double-blind design to isolate effects of stimulus-action couplings from mere stimulus exposure, task performance, and experimenter as well as tracking effects; (3) a large sample that (4) provided many data points on craving, consumption, and approach bias for (5) a wide range of personalized food stimuli.\u003c/p\u003e \u003cp\u003e Regarding limitations, allowing participants to freely choose food images might have resulted in fuzzy categories of decrease- and increase-foods which might have impaired effectiveness. Similarly, stimulus selections of some participants seem to indicate that the food item selection did not work fully as intended (e.g., one participant choosing ice cream as an increase-food), adding to the fuzziness of the categories. We recommend using a more restricted stimulus set in future studies, for example by limiting stimuli to snack foods and pre-classifying stimuli as possible increase or decrease-foods rather than leaving the choice entirely to participants.\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThe mobile AAI tested in this multi-session intervention study reduced craving intensity and approach bias towards goal-incongruent foods throughout the 12-day intervention period. The effects on craving were most pronounced for those participants who reported to struggle with dietary self-regulation and restrained eaters. While no effects were observed for actual food intake, our study adds to previous work showing the potential of cognitive bias modification interventions in health and psychopathology and instills hope that also eating behavior - being known to be highly resistant to change and multiply determined - can be changed in structured cognitive bias modification tasks that take up nor more than a few minutes per day and that can be disseminated at low costs on a global level.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAAI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eApproach-Avoidance Intervention\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAAA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eApproach-Avoidance Assessment\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eBMI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eBody-Mass Index\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDEBQ\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDutch Eating Behavior Questionnaire\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePSRS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePerceived Self-Regulatory Success in Dieting Scale\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eRCT\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eRandomized-Controlled Trial\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eUPPS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003e(Negative) Urgency, (lack of) Premeditation, (lack of) Perseverance, Sensation Seeking\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cem\u003eEthics approval and consent to participate\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe study has received ethical approval from the ethics board of the University of Salzburg (reference number 27/2018, Add 2) and is conducted in accordance with the declaration of Helsinki. All participants gave informed consent to participate in this study.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eConsent for publication\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAvailability of data and materials\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated and/or analysed during the current study are available in the Open Science Framework repository, https://osf.io/yn7kt\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCompeting interests\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eFunding\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThis research was funded in whole, or in part, by the Austrian Science Fund (FWF) [grant number P 34542-B]. SK and HvA were supported by the Doctoral College \u0026apos;Imaging the Mind\u0026apos; (FWF; W1233-B). HvA was additionally supported by the project: Mapping neural mechanisms of appetitive behaviour (FWF; KLI762-B). The funder plays no role in the study design; collection, management, analysis, and interpretation of data; writing of the report; and the decision to submit the report for publication. For the purpose of open access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAuthors\u0026apos; contributions\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eConceptualization: MBA, HvA, SK, JB. Funding acquisition: JB. Formal analysis: MBA, MR. Investigation: MBA, MR, HvA, SK. Project administration: MBA. Visualization: MR. Writing - original draft: MBA, MR. Methodology: MBA, HvB, SK, JB. Supervision: JB. Writing\u0026mdash;review and editing: MBA, MR, HvB, SK, JS, JB.\u003c/p\u003e\n\u003cp\u003eAll authors read and approved the final manuscript.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAcknowledgements\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe authors acknowledge the computational resources and services provided by Salzburg Collaborative Computing (SCC), funded by the Federal Ministry of Education, Science and Research (BMBWF) and the State of Salzburg.\u003cbr\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eGlobal Health Observatory. WHO | Overweight and obesity. WHO; 2017.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Health Organization. World Health Statistics. World Health Organ Cardiovasc Dis Factsheet 112; 2015.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Health Organization. (2018) Obesity and Overweight. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-319-33228-4_447\u003c/span\u003e\u003cspan address=\"10.1007/978-3-319-33228-4_447\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Health Organization. Healthy diet. World Health Organization. Regional Office for the Eastern Mediterranean; 2019.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSantos I, Sniehotta FF, Marques MM, Carra\u0026ccedil;a EV, Teixeira PJ. Prevalence of personal weight control attempts in adults: a systematic review and meta-analysis. Obes Rev. 2017;18:32\u0026ndash;50.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHaynes A, Kersbergen I, Sutin A, Daly M, Robinson E. A systematic review of the relationship between weight status perceptions and weight loss attempts, strategies, behaviours and outcomes. Obes Rev. 2018;19:347\u0026ndash;63.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAjzen I. The theory of planned behavior. Organ Behav Hum Decis Process. 1991;50:179\u0026ndash;211.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFishbein M. (1979) A theory of reasoned action: some applications and implications.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFishbein M, Ajzen I. (2010) Predicting and changing behavior: The reasoned action approach. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4324/9780203937082\u003c/span\u003e\u003cspan address=\"10.4324/9780203937082\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHofmann W, Friese M, Wiers RW. Impulsive versus reflective influences on health behavior: a theoretical framework and empirical review. Health Psychol Rev. 2008;2:111\u0026ndash;37.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStrack F, Deutsch R. Reflective and impulsive determinants of social behavior. Personal Soc Psychol Rev Off J Soc Personal Soc Psychol Inc. 2004;8:220\u0026ndash;47.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRinck M, Becker ES. Approach and avoidance in fear of spiders. J Behav Ther Exp Psychiatry. 2007;38:105\u0026ndash;20.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZech HG, Rotteveel M, van Dijk WW, van Dillen LF. A mobile approach-avoidance task. Behav Res Methods. 2020;52:2085\u0026ndash;97.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrockmeyer T, Hahn C, Reetz C, Schmidt U, Friederich H-CC. Approach bias and cue reactivity towards food in people with high versus low levels of food craving. Appetite. 2015;95:197\u0026ndash;202.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMeule A, Lender A, Richard A, Dinic R, Blechert J. Approach\u0026ndash;avoidance tendencies towards food: Measurement on a touchscreen and the role of attention and food craving. Appetite. 2019;137:145\u0026ndash;51.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBooth C, Spronk D, Grol M, Fox E. Uncontrolled eating in adolescents: The role of impulsivity and automatic approach bias for food. Appetite. 2018;120:636\u0026ndash;43.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKakoschke N, Kemps E, Tiggemann M. Differential effects of approach bias and eating style on unhealthy food consumption in overweight and normal weight women. Psychol Health. 2017;32:1371\u0026ndash;85.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKahveci S, Meule A, Lender A, Blechert J. Food approach bias is moderated by the desire to eat specific foods. Appetite. 2020;154:104758.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKahveci S, van Alebeek H, Berking M, Blechert J. Touchscreen-based assessment of food approach biases: Investigating reliability and item-specific preferences. Appetite. 2021;163:105190.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrockmeyer T, Hahn C, Reetz C, Schmidt U, Friederich HC. Approach Bias Modification in Food Craving - A Proof-of-Concept Study. Eur Eat Disord Rev. 2015;23:352\u0026ndash;60.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAulbach MB, Knittle K, Haukkala A. Implicit process interventions in eating behaviour: a meta-analysis examining mediators and moderators. Health Psychol Rev. 2019;13:179\u0026ndash;208.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDickson H, Kavanagh DJ, MacLeod C. The pulling power of chocolate: Effects of approach\u0026ndash;avoidance training on approach bias and consumption. Appetite. 2016;99:46\u0026ndash;51.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKakoschke N, Kemps E, Tiggemann M. Impulsivity moderates the effect of approach bias modification on healthy food consumption. Appetite. 2017;117:117\u0026ndash;25.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKakoschke N, Hawker C, Castine B, de Courten B, Verdejo-Garcia A. Smartphone-based cognitive bias modification training improves healthy food choice in obesity: A pilot study. Eur Eat Disord Rev. 2018;26:526\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVeling H, Verpaalen IAM, Liu H, Mosannenzadeh F, Becker D, Holland RW. (2021) How can food choice best be trained? Approach-avoidance versus go/no-go training. Appetite 105226.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen Z, Veling H, Dijksterhuis A, Holland RW. (2017) Do impulsive individuals benefit more from food go/no-go training? Testing the role of inhibition capacity in the no-go devaluation effect. Appetite. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.appet.2017.04.024\u003c/span\u003e\u003cspan address=\"10.1016/j.appet.2017.04.024\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKahveci S, Rinck M, van Alebeek H, Blechert J. How pre-processing decisions affect the reliability and validity of the approach\u0026ndash;avoidance task: Evidence from simulations and multiverse analyses with six datasets. Behav Res Methods. 2023;56:1551\u0026ndash;82.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLender A, Meule A, Rinck M, Brockmeyer T, Blechert J. Measurement of food-related approach\u0026ndash;avoidance biases: Larger biases when food stimuli are task relevant. Appetite. 2018;125:42\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFishbach A, Shah JY. Self-control in action: implicit dispositions toward goals and away from temptations. J Pers Soc Psychol. 2006;90:820\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMeule A, Richard A, Dinic R, Blechert J. Effects of a Smartphone-Based Approach-Avoidance Intervention on Chocolate Craving and Consumption: Randomized Controlled Trial. JMIR MHealth UHealth. 2019;7:e12298.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchakel L, Veldhuijzen DS, van Middendorp H, Dessel PV, Houwer JD, Bidarra R, Evers AWM. The effects of a gamified approach avoidance training and verbal suggestions on food outcomes. PLoS ONE. 2018;13:e0201309.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAulbach MB, Knittle K, van Beurden SB, Haukkala A, Lawrence NS. App-based food Go/No-Go training: User engagement and dietary intake in an opportunistic observational study. Appetite. 2021;165:105315.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEberl C, Wiers RW, Pawelczack S, Rinck M, Becker ES, Lindenmeyer J. Implementation of Approach Bias Re-Training in Alcoholism\u0026mdash;How Many Sessions are Needed? Alcohol Clin Exp Res. 2014;38:587\u0026ndash;94.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNour M, Sui Z, Grech A, Rangan A, McGeechan K, Allman-Farinelli M. The fruit and vegetable intake of young Australian adults: a population perspective. Public Health Nutr. 2017;20:2499\u0026ndash;512.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAdriaanse MA, van Oosten JMF, de Ridder DTD, de Wit JBF, Evers C. Planning What Not to Eat: Ironic Effects of Implementation Intentions Negating Unhealthy Habits. Pers Soc Psychol Bull. 2011;37:69\u0026ndash;81.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGardner B, Rebar AL. Habit Formation and Behavior Change. Oxf Res Encycl Psychol. 2019. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/acrefore/9780190236557.013.129\u003c/span\u003e\u003cspan address=\"10.1093/acrefore/9780190236557.013.129\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLally P, Wardle J, Gardner B. Experiences of habit formation: A qualitative study. Psychol Health Med. 2011;16:484\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKahveci S, Van Alebeek H, Blechert J. (2024) The Dual-Feature Approach-Avoidance Task: Validity, Training Efficacy, and the Role of Contingency Awareness in Changing Food Preference. Cogn Emot 1\u0026ndash;23.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAulbach MB, Van Alebeek H, Kahveci S, Blechert J. Testing the effectiveness of a mobile approach avoidance intervention and measuring approach biases in an ecological momentary assessment context: study protocol for a randomised-controlled trial. BMJ Open. 2023;13:e070443.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRevelle W. (2022) psych: Procedures for psychological, psychometric, and personality research.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKelley K. (2022) MBESS.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMeule A, Papies EK, K\u0026uuml;bler A. Differentiating between successful and unsuccessful dieters. Validity and reliability of the Perceived Self-Regulatory Success in Dieting Scale. Appetite. 2012;58:822\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan Strien T, Frijters JER, Bergers GPA, Defares PB. The Dutch Eating Behavior Questionnaire (Debq) for Assessment of Restrained, Emotional, and External Eating Behavior Internal Structure and Measurement Invariance of the Dutch Eating Behavior Questionnaire (DEBQ) in a (Nearly) Representative Dutch. Int J Eat Disord. 1986;5:295\u0026ndash;315.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWhiteside SP, Lynam DR, Miller JD, Reynolds SK. Validation of the UPPS impulsive behaviour scale: a four-factor model of impulsivity. Eur J Personal. 2005;19:559\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBlechert J, Lender A, Polk S, Busch NA, Ohla K. Food-Pics_Extended\u0026mdash;An Image Database for Experimental Research on Eating and Appetite: Additional Images, Normative Ratings and an Updated Review. Front Psychol. 2019;10:307.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eToet A, Kaneko D, de Kruijf I, Ushiama S, van Schaik MG, Brouwer A-M, Kallen V, van Erp JBF. CROCUFID: A Cross-Cultural Food Image Database for Research on Food Elicited Affective Responses. Front Psychol. 2019;10:58.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eForoni F, Pergola G, Argiris G, Rumiati RI. The FoodCast research image database (FRIDa). Front Hum Neurosci. 2013;7:51.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMestdagh M, Verdonck S, Piot M, Niemeijer K, Kilani G, Tuerlinckx F, Kuppens P, Dejonckheere E. (2023) m-Path: an easy-to-use and highly tailorable platform for ecological momentary assessment and intervention in behavioral research and clinical practice. Front Digit Health 5.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan Beers JJ, Kaneko D, Stuldreher IV, Zech HG, Brouwer A-M. An Accessible Tool to Measure Implicit Approach-Avoidance Tendencies Towards Food Outside the Lab. Companion Publ. 2020 Int. Conf. Multimodal Interact. Virtual Event Netherlands: ACM; 2020. pp. 307\u0026ndash;11.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRuf A, Neubauer AB, Ebner-Priemer U, Reif A, Matura S. Studying dietary intake in daily life through multilevel two-part modelling: a novel analytical approach and its practical application. Int J Behav Nutr Phys Act. 2021;18:130.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLenth RV. (2023) emmeans: Estimated Marginal Means, aka Least-Squares Means.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAulbach MB, van Alebeek H, Jones CM, Blechert J. Why we don\u0026rsquo;t eat as intended: Moderators of the short-term intention\u0026ndash;behaviour relation in food intake. Br J Health Psychol. 2024. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/bjhp.12714\u003c/span\u003e\u003cspan address=\"10.1111/bjhp.12714\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHofmann W, Van Dillen L. Desire: The New Hot Spot in Self-Control Research. Curr Dir Psychol Sci. 2012;21:317\u0026ndash;22.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKavanagh DJ, Andrade J, May J. Imaginary Relish and Exquisite Torture: The Elaborated Intrusion Theory of Desire. Psychol Rev. 2005;112:446\u0026ndash;67.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMay J, Andrade J, Kavanagh DJ, Hetherington M. Elaborated Intrusion Theory: A Cognitive-Emotional Theory of Food Craving. Curr Obes Rep. 2012;1:114\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan Alebeek H, R\u0026ouml;ttger M, Kahveci S, Blechert J, Aulbach MB. (2024) The Only Constant is Change: Stable vs. Variable Aspects of Food Approach Bias Relate Differently to Food Craving and Intake. Appetite 107726.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStice E, Yokum S, Veling H, Kemps E, Lawrence NS. Pilot test of a novel food response and attention training treatment for obesity: Brain imaging data suggest actions shape valuation. Behav Res Ther. 2017;94:60\u0026ndash;70.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang Y, Morys F, Wu Q, Li J, Chen H. Pilot study of food-specific go/no-go training for overweight individuals: brain imaging data suggest inhibition shapes food evaluation. Soc. Cogn. Affect. Neurosci; 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAdams RC, Button KS, Hickey L, et al. Food-related inhibitory control training reduces food liking but not snacking frequency or weight in a large healthy adult sample. Appetite. 2021;167:105601.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaunders B, Milyavskaya M, More KR, Anderson J. Food cravings are associated with increased self-regulation, even in the face of strong instigation habits: A longitudinal study of the transition to plant-based eating. Appl Psychol Health Well-Being. 2025;17:e12629.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eElliston KG, Ferguson SG, Sch\u0026uuml;z B. Personal and situational predictors of everyday snacking: An application of temporal self-regulation theory. Br J Health Psychol. 2017;22:854\u0026ndash;71.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchumacher SE, Kemps E, Tiggemann M. Bias modification training can alter approach bias and chocolate consumption. Appetite. 2016;96:219\u0026ndash;24.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHouben K, Jansen A. Training inhibitory control. A recipe for resisting sweet temptations. Appetite. 2011;56:345\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVeling H, Aarts H, Stroebe W. Stop signals decrease choices for palatable foods through decreased food evaluation. Front Psychol. 2013;4:1\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHerman CP, Polivy J. Restraint Eating. Philadelphia: In: Obesity. Saunders; 1980. pp. 208\u0026ndash;25.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAdams RC, Chambers CD, Lawrence NS. Do restrained eaters show increased BMI, food craving and disinhibited eating? A comparison of the Restraint Scale and the Restrained Eating scale of the Dutch Eating Behaviour Questionnaire. R Soc Open Sci. 2019;6:190174.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHerman CP, Mack D. Restrained and unrestrained eating. J Pers. 1975;43:647\u0026ndash;60.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRuderman AJ. Dietary restraint: a theoretical and empirical review. Psychol Bull. 1986;99:247.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan Alebeek H, Veling H, Blechert J. Disentangling go/no-go from motivational orientation to foods: Approaching is more than just responding. Food Qual Prefer. 2023;106:104821.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang Y, Shields GS, Wu Q, Liu Y, Chen H, Guo C. Cognitive training on eating behaviour and weight loss: A meta-analysis and systematic review. Obes Rev. 2019. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/obr.12916\u003c/span\u003e\u003cspan address=\"10.1111/obr.12916\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Dessel P, De Houwer J, Gast A. Approach\u0026ndash;Avoidance Training Effects Are Moderated by Awareness of Stimulus\u0026ndash;Action Contingencies. Pers Soc Psychol Bull. 2016;42:81\u0026ndash;93.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Dessel P, Hughes S, De Houwer J. How Do Actions Influence Attitudes? An Inferential Account of the Impact of Action Performance on Stimulus Evaluation. Personal Soc Psychol Rev. 2019;23:267\u0026ndash;84.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Dessel P, Gawronski B, Smith CT, De Houwer J. Mechanisms underlying approach-avoidance instruction effects on implicit evaluation: Results of a preregistered adversarial collaboration. J Exp Soc Psychol. 2017;69:23\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMasterton S, Hardman CA, Jones A. Don\u0026rsquo;t stop believing\u0026rsquo;: The role of training beliefs in cognitive bias modification paradigms. Appetite. 2022;174:106041.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMichie S, Richardson M, Johnston M, Abraham C, Francis J, Hardeman W, Eccles MP, Cane J, Wood CE. The behavior change technique taxonomy (v1) of 93 hierarchically clustered techniques: Building an international consensus for the reporting of behavior change interventions. Ann Behav Med. 2013;46:81\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Footnotes","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003e We had preregistered further analyses and report those in the supplementary materials.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e Due to apparently interchanged data for height and weight of one participant, we excluded this participant from the BMI computation. \u003cem\u003eN\u003c/em\u003e\u0026thinsp;=\u0026thinsp;71\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e Note that the parameter estimates for the hurdle part relate to the probability that craving is zero, that is, the probability of craving being absent.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e As pre-registered, we calculated interaction models with additional participant-level, stimulus-level and training-level variables for the changes from pre- to post training. An overview of effects as well as all model results can be found in the supplements. In this paper, we focus on the variables with consistent effects (i.e., effects for which we found evidence within different outcome variables or different similar moderator variables).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"international-journal-of-behavioral-nutrition-and-physical-activity","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"ijbn","sideBox":"Learn more about [International Journal of Behavioral Nutrition and Physical Activity](http://ijbnpa.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/ijbn/default.aspx","title":"International Journal of Behavioral Nutrition and Physical Activity","twitterHandle":"@IJBNPA","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"cognitive bias modification, approach bias modification, approach-avoidance task, dietary change, eating behavior, mHealth, restraint eating, intervention, food craving, ecological momentary assessment, randomized controlled trial","lastPublishedDoi":"10.21203/rs.3.rs-6751171/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6751171/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eGiven the therapeutic potential of Approach-Avoidance interventions (AAIs) in the alcohol domain, research has increasingly applied them to the food domain. In AAIs, harmful stimuli are avoided while healthy ones are approached, for example by respectively moving a phone away from or towards oneself.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe administered a phone-based AAI six times over two weeks to 156 participants in a pre-registered randomized-controlled trial to reduce intake of six \u0026ldquo;decrease-foods\u0026rdquo; and increase intake of six \u0026ldquo;increase-foods\u0026rdquo;, selected according to each participant\u0026rsquo;s individual dietary goals. The control group received a placebo task in which all stimuli were equally often approached and avoided. Food craving and intake were the outcomes, measured daily during the training period, four days before and after, and once during a follow-up one month after training. Per-food approach bias was recorded before and after training, and at follow-up.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eCompared to placebo, active training reduced the level of decrease-food craving without affecting how often craving occurred. Restrained eaters and those with low past dietary success showed the strongest craving strength reduction. Active training also reduced approach bias for decrease-foods. We found no intervention effects on increase-foods on any outcome. There were no interpretable training effects for food intake and no changes were maintained at follow-up.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eWe find support for the use of AAI against food cravings for goal-incongruent foods, especially for those who struggle with their diet. It remains to future research how this can be effectively translated into reduced food intake.\u003c/p\u003e\u003ch2\u003eTrial registration\u003c/h2\u003e \u003cp\u003eThis study was registered in the German Clinical Trials Register, ID DRKS00030780.\u003c/p\u003e","manuscriptTitle":"Effectiveness of a Smartphone-Delivered Approach-Avoidance Intervention in Dietary Behavior - A Randomized Controlled Trial","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-03 07:19:23","doi":"10.21203/rs.3.rs-6751171/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-08-05T00:03:25+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-30T18:11:25+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"193962555765036066293884777608703805618","date":"2025-07-04T11:32:10+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-06-25T11:42:11+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"84285162060307435593064667645178486670","date":"2025-06-24T07:19:01+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-05-29T08:25:26+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-05-27T04:08:14+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-05-27T04:06:56+00:00","index":"","fulltext":""},{"type":"submitted","content":"International Journal of Behavioral Nutrition and Physical Activity","date":"2025-05-26T13:05:51+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"international-journal-of-behavioral-nutrition-and-physical-activity","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"ijbn","sideBox":"Learn more about [International Journal of Behavioral Nutrition and Physical Activity](http://ijbnpa.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/ijbn/default.aspx","title":"International Journal of Behavioral Nutrition and Physical Activity","twitterHandle":"@IJBNPA","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"1f0450dd-e240-464a-afd3-aeeeb0cf69f9","owner":[],"postedDate":"June 3rd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-12-01T16:06:03+00:00","versionOfRecord":{"articleIdentity":"rs-6751171","link":"https://doi.org/10.1186/s12966-025-01836-2","journal":{"identity":"international-journal-of-behavioral-nutrition-and-physical-activity","isVorOnly":false,"title":"International Journal of Behavioral Nutrition and Physical Activity"},"publishedOn":"2025-11-28 15:57:37","publishedOnDateReadable":"November 28th, 2025"},"versionCreatedAt":"2025-06-03 07:19:23","video":"","vorDoi":"10.1186/s12966-025-01836-2","vorDoiUrl":"https://doi.org/10.1186/s12966-025-01836-2","workflowStages":[]},"version":"v1","identity":"rs-6751171","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6751171","identity":"rs-6751171","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}