Approach bias modification training to increase physical activity: A pilot randomized controlled trial in healthy volunteers

Abstract

Regular physical activity is associated with better physical and mental health outcomes as well as higher quality of life. This pilot randomized controlled trial examined whether approach bias modification, an economical and easily accessible computerized cognitive training, could increase objectively and subjectively measured physical activity in individuals aiming for more physical activity. Forty healthy volunteers of normal weight were randomly allocated to six sessions of approach bias modification or no treatment. The approach bias modification adopted an implicit learning paradigm that trained participants to show approach behavior in response to visual cues of physical activity. Approach bias modification did not increase objectively and subjectively measured physical activity.

Keywords

approach-avoidance task automatic approach tendencies cognitive bias modification obesity overweight

Introduction

Overweight and obesity constitute a growing global health challenge leading to enormous financial costs for the health care system (Mensink et al., 2013; Withrow and Alter, 2011). Moreover, obesity is associated with a range of severe somatic sequelae (e.g. cardiovascular diseases, certain types of cancer, and diabetes) as well as mental disorders (e.g. depression and anxiety disorders) and leads to higher mortality rates (Ng et al., 2014; Puhl and Heuer, 2010; Wang et al., 2011).

To date, standard first-line treatments of overweight and obesity are multicomponent interventions that incorporate psychological, behavioral, and nutritional interventions that aim to increase physical activity (PA) and decrease energy intake (Altman and Wilfley, 2015; Shaw et al., 2006). Caspersen et al. (1985) defined PA as any bodily movement produced by the skeletal muscles that results in energy expenditure. This includes different types of movements such as walking, cycling, and vigorous-intensity activity (Hallal et al., 2012). There is irrefutable evidence that regular PA is linked to better physical and mental health outcomes as well as higher quality of life (Myers et al., 2015; Penedo and Dahn, 2005; Warburton et al., 2006). In addition, PA was found to have a positive effect on several aspects of cognition and self-esteem in healthy children and adolescents (Rasmussen and Laumann, 2013). As a consequence, Haskell et al. (2009) pointed out that the accumulated data on the general beneﬁts of PA for people of all ages is more than adequate to support strong preventive medicine policies.

In the short term, behavioral interventions that include food intake modification and PA enhancement are effective resulting in significant although little weight reduction (Dombrowski et al., 2010). However, high rates of attrition and non-response pose a substantial problem: for instance, in a study by Grave et al. (2005), 51.7 percent of obese patients who had sought treatment had discontinued treatment 12 months later. Fitzpatrick et al. (2015) identified 23.6 percent of the participants, who attended a behavioral lifestyle treatment to lose weight, as non-responders as they did not meet the behavioral recommendations at baseline and after treatment. Albeit bariatric surgery leads to substantial and rapid weight loss (i.e. 12–17 body mass index (BMI) points on average), this invasive treatment entails the risks of complication, reoperation, and death (Chang et al., 2014). Besides, growth mixture models revealed subgroups that regained weight after bariatric surgery (Courcoulas et al., 2014). Although King et al. (2012) reported increased PA levels for the majority of adult patients after bariatric surgery, most patients continued to be insufficiently active and some even lowered their PA levels. Post-surgical exercise was associated with a greater weight loss of over 4 percent of BMI (Livhits et al., 2010), underscoring the importance of PA even in the context of bariatric surgery.

Long-term effects of behavioral interventions that promote PA are mostly disappointing (Moldovan and David, 2011; Nauta et al., 2001) which emphasizes the need for novel effective interventions that can be easily disseminated and that are cost-effective. Again, a physically active lifestyle is one of the key factors to successful long-term weight loss maintenance (Elfhag and Rössner, 2005; Wing et al., 2008; Wing and Phelan, 2005).

The reflective–impulsive model (RIM; Strack and Deutsch, 2004) proposes that two separate yet interacting information processing systems — the reflective system and the impulsive system — motivate behavior (Rothman et al., 2009). According to Rothman et al. (2009), the two systems differ in their levels of awareness and some further factors that determine behavior. Whereas the reflective system operates on a conscious, deliberate level and rather slowly, the impulsive system initiates automatic processes the individual is not necessarily aware of (Deutsch and Strack, 2006; Krishna and Strack, 2017; Rothman et al., 2009). For example, even with the best conscious intention and highest motivation to work out in the gym we might find ourselves on the couch watching television instead. Conroy et al. (2010) pointed out that an individual’s motivation to be physically active includes both explicit and implicit processes, and that incorporating relevant implicit motivation processes might enhance promotion efforts.

Implicit approach and avoidance motivation can be measured by means of the approach-avoidance task (AAT; Rinck and Becker, 2007). In this joystick-based computer task, pictures of two different categories (e.g. healthy vs unhealthy food) are presented in combination with another stimulus feature (e.g. two different background colors). Participants are asked to pull or push a joystick depending on the stimulus feature that is irrelevant to the stimulus dimension that the task is meant to assess (implicit task). As arm flexion (pulling) is associated with more positive evaluations than extension (pushing; Cacioppo et al., 1993), an approach bias is indicated by faster pulling than pushing of one picture type, whereas faster pushing than pulling is interpreted as an avoidance bias (Lender et al., 2018). Approach and avoidance biases have been observed in various domains, for example, toward alcohol cues in heavy drinkers (Field et al., 2008), and away from spider pictures in people with spider phobia (Rinck and Becker, 2007).

Regarding PA, previous studies (e.g. Cheval et al., 2014; Conroy et al., 2010) found that implicit approach biases toward PA were linked to objective PA, over and beyond (deliberate) PA intentions. In a study by Hannan et al. (2018), higher levels of leisure-time exercise were associated with significantly stronger approach biases for exercise in healthy adults. In addition, approach–avoidance associations explained unique variance in the individuals’ exercise behavior supporting the idea that implicit motivational processes, at least partly, motivate exercise as proposed by the RIM (Hannan et al., 2018).

In the context of alcohol addiction, Wiers et al. (2010, 2011) turned the AAT into a modification task by manipulating the contingencies of push and pull movements in response to alcohol cues. Patients who underwent the described approach bias modification (ABM) by means of the AAT, shifted from an approach bias to an avoidance bias for alcohol. Importantly, these patients also showed better treatment outcomes after 12 months (Wiers et al., 2011). This effect could be replicated in a number of studies (Eberl et al., 2013; Manning et al., 2016; Rinck et al., 2018). As a result of this success story, several researchers have tried to transfer ABM to other domains such as eating behavior and smoking, with mixed findings (e.g. Aulbach et al., 2019; Becker et al., 2015; Boffo et al., 2019; Brockmeyer et al., 2015, 2019; Kakoschke et al., 2017; Lender et al., 2018; Schumacher et al., 2016; Wittekind et al., 2019).

To date, ABM trainings were mainly applied to reduce undesirable/unhealthy behavior (e.g. drinking alcohol and eating lots of high-calorie food). To the best of our knowledge, ABM has not yet been used to increase PA. In this study, we aimed to examine whether PA as a desirable/healthy behavior can be facilitated by ABM in normal-weight individuals aiming for more PA. We expected that six sessions of ABM would lead to a stronger increase in objectively and subjectively measured PA than no treatment, through an increase in automatic approach tendencies toward visual cues of PA.

Materials and methods

Design of the study

This study was a mono-center, pilot randomized controlled superiority trial utilizing a parallel group design. Participants were randomly allocated to one of two groups (ABM vs no treatment control group). The amount of steps as measured by a pedometer during a time interval of 2 weeks served as primary outcome. Secondary outcomes were as follows: approach bias toward PA, self-reported PA, and motivation toward PA. All outcomes were assessed at baseline (t1) and at the end of the treatment (t2). The study was approved by the ethics committee of the Institute of Psychology at the University of Göttingen.

Sample

An a priori power calculation revealed that a total sample size of n = 34 participants had to be included to detect a medium effect size in a 2 × 2 mixed analysis of variance (ANOVA) with two groups (with 80% power and 0.05 two-tailed significance level). We assumed a data loss up to 15 percent due to technical errors, dropouts (i.e. a = 15%), and applied an attrition correction factor of 1/(1 − a). Hence, a sample size of n = 40 participants was needed (i.e. n = 20 per group).

The sample consisted of 40 healthy students (36 women and 4 men; ABM group: 17 women and 3 men, control group: 19 women and 1 man) with a mean age of 22.33 ± 2.60 years, ranging from 19 to 30 years. Participants were recruited via posters that were distributed over the university campus and via social media. Interested individuals were contacted via e-mail to check inclusion criteria. These were as follows: age ⩾18 years, normal weight according to the World Health Organization (WHO) (BMI: 18.5–25 kg/m²), and personal desire to increase PA. Exclusion criteria were as follows: diabetes mellitus and physical impairment that precludes PA. Participants provided written informed consent and could choose between financial compensation (€30) and course credit for their participation in the study.

Measures

Pedometer

The participants’ covered footsteps were measured using the Omron Walking Style 2.1 Model HJ321-E. Worn at the hip, it accurately measures steps made by the individual (Battenberg et al., 2017), estimates the energy consumption as well as the distance in kilometers and stores the data from 7 days. With an accuracy rate of over 90 percent (Battenberg et al., 2017) and step count errors within 1 percent (Huang et al., 2016), this device was successfully validated and previously used in a number of studies (e.g. Phing et al., 2017; Tsianakas et al., 2017; Yusoff et al., 2018).

Approach bias

AAT

The participants’ approach bias toward visual stimuli of PA were assessed by means of the computerized AAT (Rinck and Becker, 2007). Using Inquisit 5 (Millisecond Software), pictures of two categories (PA vs physical inactivity (PIA)) comparable to Cheval et al. (2014) were presented on a computer screen with two different backgrounds (blue vs orange). The participants were asked to pull or push a joystick (counterbalanced across participants) depending on the background’s color and thus irrespective of the picture’s content. Pulling the joystick increased the picture, whereas pushing the joystick decreased it. By combining proprioceptive (arm movement) and exteroceptive (zooming effect) cues, the task imitates respective sensations of approach versus avoidance behavior (Neumann and Strack, 2000). Pictures were presented in a pseudo-randomized order with a maximum of three pictures of the same category or background color in a row.

The visual stimuli consisted of 40 black and white graphics from the Pixabay picture database that were divided into two categories: 20 pictures illustrated PA (e.g. riding a bicycle, jogging, and playing basketball), whereas the other 20 pictures represented PIA (e.g. watching television, playing computer, and reading; see Figure 1).

Figure 1.

Examples of the two categories (a) PA and (b) PIA of visual stimuli that were presented during the approach-avoidance task (AAT) and the stimulus-response-compatibility task (SRCT).

Following common principles in AAT research (Becker et al., 2019; Lender et al., 2018), approach bias for PA was calculated by means of a double difference bias score: (median reaction time when pushing PA cues + median reaction time when pulling PIA cues) − (median reaction time when pulling PA cues + median reaction time when pushing PIA cues). Positive values indicate a specific approach bias toward stimuli of PA, whereas negative values represent an avoidance bias toward these stimuli. Participants completed 10 practice trials in which they had to react correctly to the color of the background (orange/blue — without any picture of PA or PIA) in order to get used to the handling of the joystick. These practice trials were followed by 80 trials. In the context of automatic alcohol action tendencies, Kersbergen et al. (2015) found the reliability of the AAT version used in this study to be poor and unrelated to alcohol consumption and hazardous drinking, questioning its validity as well.

Stimulus-response-compatibility-task

Previous research (Kersbergen et al., 2015; Lender et al., 2018) identified limitations concerning the AAT’s reliability and validity, especially with regard to the version using irrelevant features as in this study. Therefore, we additionally used the stimulus-response-compatibility-task (SRCT) (Houwer et al., 2001) to assess automatic approach biases toward stimuli representing PA. The task was performed using Inquisit 5 (Millisecond Software). Participants were asked to move a manikin on a computer screen via keyboard presses toward or away from a presented stimulus. We displayed the same visual stimuli as in the AAT (see Figure 1). First, the participants saw an empty screen (1000 ms), followed by a stimulus of one of two categories (PA vs PIA) in the center of the screen. The manikin appeared at the side of the picture, over, or under the picture. According to the particular instruction, participants had to move the manikin toward or away from the picture. After the correct answer was given, the correct movement of the manikin was shown per video for 1000 ms. Wrong answers resulted in the presentation of a red cross in the middle of the screen for 1000 ms. Reaction times between the presentation of the pictures and the answers were assessed.

Three different blocks with different instructions were used: (1) approaching pictures of PA and avoiding pictures of PIA, (2) approaching pictures of PIA and avoiding pictures of PA, and (3) moving the manikin to one side (left vs right) when seeing a picture (PA vs PIA). Order of blocks was counterbalanced across participants. Each block consisted of eight practice trials followed by 2 × 28 test trials. Within the test block each picture was presented twice, once with the manikin above and once below the picture. Participants were instructed to answer as fast and as correctly as possible. The bias score was calculated by subtracting the mean of approach PA/avoid PIA trials from the mean of avoid PIA/approach PIA trials. Hence, a positive score represents an approach bias toward PA. With regard to automatic alcohol action tendencies, the SRCT as used in this sample was found to be poorly reliable and related to neither alcohol consumption nor hazardous drinking, challenging whether the SRCT is a valid measure as well (Kersbergen et al., 2015).

International Physical Activity Questionnaire

We used the German short form of the International Physical Activity Questionnaire (IPAQ) in order to assess the participants’ subjective rate of health-related PA. Seven items identify all strenuous and moderate physical as well as walking activities in minutes of the last 7 days. The IPAQ has shown satisfactory retest reliability and construct validity (Craig et al., 2003; Hagströmer et al., 2006).

Behavioral Regulation in Exercise Questionnaire

In order to identify the processes that lead to the initiation and maintenance of PA, the German version of the Behavioral Regulation in Exercise Questionnaire (BREQ-3) was used (Rausch Osthoff, 2017). The questionnaire consists of 24 items rated on a scale from 0 (not true to me) to 4 (very true to me) which correspond to the different motivation types and their regulatory styles (amotivation: non-regulation; extrinsic motivation: external, introjected, identified, and integrated regulation; intrinsic motivation: intrinsic regulation) from the self-determination theory as referred to by Ryan and Deci (2000). These motivation types and their regulatory styles differ in the degree to which they are self-determined (in ascending order). Amotivation lacks both types of motivation (intrinsic and extrinsic) leading to a complete lack of self-determination concerning the target behavior. External regulation is a form of extrinsic motivation in which specific external contingencies (e.g. desired reward or to avoid punishment) control an individual’s behavior. Introjection implies taking in of external regulation but executing the contingent consequences to oneself, for instance, contingent self-worth (pride). Identification entails that underlying values of a behavior are recognized and accepted by the individual, leading to higher commitment and maintenance of behavior. For instance, identifying with the importance of regular PA for one’s own health would lead to more volitional exercise. Integration represents the most complete type of internalization concerning extrinsic motivation and includes both identifying with the importance of behaviors and integrating those identifications with other aspects of the self-values and identity. Intrinsic motivation arises if an individual perceives an activity as rewarding independent of operationally separable consequences (Deci and Ryan, 2000). The psychometric properties of the questionnaire have not been published yet.

Procedure

After participants provided written informed consent, basic and demographic variables (age, gender, self-reported body height and weight) were assessed. Next, the AAT, SRCT, IPAQ, and BREQ-3 were carried out. Thereafter, participants were randomly assigned to one of the two treatment groups (ABM group vs no treatment control group) using a computer-based randomization program (Research Randomizer; Urbaniak and Plous, 2013). Randomization was performed in a 1:1 ratio, results were stored in sealed envelopes to ensure allocation concealment. Subsequently, participants who were allocated to the ABM condition had their first training session. This was followed by further five training sessions during the next 2 weeks. After the sixth training session, the AAT, SRCT, IPAQ, and BREQ-3 were assessed again. Participants in the control group took part in the same assessments but did not receive any treatment in between. During the 2 weeks between pre- and post-assessment, all participants were asked to wear the pedometer during their wake phase. The measured steps were regularly documented and stored by the experimenter.

Intervention: ABM

Participants in the ABM group attended six treatment sessions of 10 minutes each, supervised by the investigator. The practice sessions took place over a minimum period of 10 to a maximum of 21 days (M_days = 14.23, SD_days = 2.31), with a minimum of one session per week. During the training sessions, participants performed a treatment version of the AAT (Rinck and Becker, 2007) which comprised 240 trials. In this treatment version, the contingencies of the pictures representing PA/PIA in relation to the background color (orange/ blue) were manipulated to enable implicit learning. Hence, stimuli showing PA were always to be approached (by pulling the joystick toward oneself), whereas stimuli showing PIA were always to be avoided (by pushing the joystick away).

Statistics

SPSS 25 was used for the statistical analyses. The statistical tests were performed two-tailed with the statistical significance value set to p < 0.05. Moreover, effect sizes (partial eta square, η²) were calculated. According to Cohen (1988), an η² of 0.01 represents a small effect, whereas an η² of 0.06 reflects a medium effect and an η² of 0.14 a large effect. Due to technical difficulties, the pedometry data of one participant and AAT data of three participants had to be excluded (Figure 2). A per protocol analysis was performed. As this is a pilot experimental study and since no participant dropped out from the study, no intent-to-treat analysis was performed. Participants whose data deviated more than 3.29 SD from the group’s mean were defined as outliers and excluded from the particular analysis. Excluded participants are reported for each analysis in the particular results section.

Figure 2.

Consolidated standards of reporting trials (CONSORT) diagram illustrating flow of participants through the study.

In order to evaluate treatment-related changes, 2 × 2 mixed ANOVAs with group (ABM vs control) as between-subjects factor and time (baseline vs post-treatment) as within-subjects factor were calculated for each outcome variable. Means and SDs at baseline are reported and tested for group differences by means of t tests for independent samples.

Results

Comparison of conditions at baseline

Groups did not differ in any variable at baseline (see Table 1).

Table 1.

Demographic characteristics and outcome measures at baseline and post treatment.

Variable	Baseline							Post-treatment
	ABM group			Control group			t-test	ABM group		Control group
	M	SD	n	M	SD	n		M	SD	M	SD
Age (years)	22.60	2.72	20	22.05	2.50	20	t(38) = 0.67, p = 0.51
BMI (kg/m²)	23.59	3.31	20	23.19	2.77	20	t(38) = 0.41, p = 0.68
Steps per day	6498.54	2980.50	20	6735.60	2214.00	18	t(36) = –0.28, p = 0.78	6879.99	3106.38	6705.99	2044.61
Mean (km/day)	5.01	2.32	20	6.29	3.89	18	t(36) = –1.25, p = 0.22	8.44	14.62	5.04	1.56
AAT	−4.47	50.42	17	−30.47	43.91	19	t(34) = 1.64, p = 0.11	−6.89	49.97	−2.79	46.52
SRCT	120.75	312.38	20	322.45	571.79	20	t(38) = –1.38, p = 0.17	12,73	205.48	13.30	449.08
IPAQ
Vigorous activities	127.63	97.84	19	315.00	111.74	19	t(36) = 0.55, p = 0.58	131.58	119.19	64.11	60.99
Moderate activities	193.00	215.38	20	105.53	114.60	19	t(37) = 1.57, p = 0.13	169.90	173.38	123.68	75.75
Walking	150.00	281.68	19	153.37	220.26	19	t(36) = –0.04, p = 0.97	169.47	103.09	151.58	143.72
BREQ-3
Amotivation	0.51	0.66	20	0.54	0.54	20	t(38) = –0.13, p = 0.90	0.40	0.63	0.40	0.55
External regulation	0.63	0.79	20	0.51	0.53	20	t(38) = 0.53, p = 0.60	0.55	0.86	0.46	0.45
Introjected regulation	1.79	0.77	20	1.89	0.89	20	t(38) = –0.38, p = 0.71	1.61	0.85	1.66	0.96
Identified regulation	2.71	0.76	20	2.70	0.60	20	t(38) = 0.06, p = 0.95	2.78	0.85	2.70	0.73
Integrated regulation	1.86	1.09	20	1.58	0.87	20	t(38) = 0.92, p = 0.36	2.09	1.22	1.80	0.93
Intrinsic regulation	2.54	0.86	20	2.38	0.78	20	t(38) = 0.63, p = 0.54	2.68	0.86	2.21	0.77

SD: standard deviation; ABM: approach bias modification; BMI: body mass index; AAT: approach-avoidance task; SRCT: stimulus-response-compatibility task; IPAQ: International Physical Activity Questionnaire, physical activity during the last 7 days in minutes; BREQ-3: Behavioral Regulation in Exercise Questionnaire.

Effects of ABM

Pedometry

In order to assess whether or not ABM increased PA levels, a 2 × 2 mixed ANOVA with group (ABM vs control) as between-subjects factor and time (first week vs second week) as within-subjects factor was calculated for the taken steps. As only 21 participants wore the pedometer longer than the period of 14 days, we refrained from taking the measured steps of the third week into account. The data of two participants had to be excluded (one outlier and one with technical difficulties). Hence, the sample concerning pedometry data was reduced to n = 38.

There was neither a significant main effect of time (F(1, 36) = 0.18, p = 0.68, partial η² < 0.01), nor of group (F(1, 36) = 0.002, p = 0.97, partial η² < 0.01). In addition, there was no significant interaction effect of time × group (F(1, 36) = 0.24, p = 0.62 partial η² < 0.01) indicating that PA levels did not change in general and not differently in the two groups.

Approach bias toward PA (AAT and SRCT)

As four participants had to be excluded (one outlier and three due to technical difficulties), the sample concerning the AAT was reduced to n = 36. There was no significant main effect of time (F(1, 35) = 1.76, p = 0.19, partial η² = 0.05), no significant main effect of group (F(1, 35) = 0.73, p = 0.40, partial η² = 0.02), and no significant interaction effect of time × group (F(1, 35) = 2.49, p = 0.12 partial η² = 0.07).

We observed a significant main effect of time on the SRCT (F(1, 38) = 7.07, p = 0.01, partial η² = 0.16; baseline > post-treatment). However, there was neither a significant main effect of group (F(1,38) = 0.97, p = 0.33, partial η² = 0.03) nor a significant interaction effect of time × group (F(1, 38) = 1.64, p = 0.21, partial η² = 0.04).

IPAQ

With regard to vigorous and walking activities in the IPAQ, two outliers were excluded, reducing the sample to n = 38. One participant had to be excluded as outlier concerning the subscale “moderate physical activities,” which reduced the sample to n = 39.

With respect to the effects of ABM on subjective ratings of vigorous physical activities, mixed ANOVA neither revealed a significant main effect of time (F(1, 36) = 3.91, p = 0.06, partial η² = 0.10), nor of group (F(1, 36) = 2.28, p = 0.14, partial η² = 0.06). There was, however, a significant interaction effect of time × group on IPAQ—vigorous activities (F(1, 36) = 5.46, p = 0.03, partial η² = 0.13). As can be seen in Table 1, this effect was mainly driven by a reduction of vigorous activity in the control group, not by an increase of vigorous activity in the ABM group.

Regarding the effects of ABM on subjective ratings of moderate physical as well as walking activities, there were no main effects of time (moderate physical activities: F(1, 37) = 0.01, p = .93, partial η² < 0.01; walking activities: F(1, 36) = 0.06, p = 0.80, partial η² < 0.01), no main effect of group (moderate physical activities: F(1, 37) = 2.47, p = 0.12, partial η² = 0.06; walking activities: F(1, 36) = 0.02, p = 0.89, partial η² < 0.01), and no interaction effect of time × group time (moderate physical activities: F(1, 37) = 0.63, p = 0.43, partial η² = 0.02; walking activities: F(1, 36) = 0.09, p = 0.76, partial η² < 0.01).

BREQ-3

The effect of ABM on motivational styles toward PA was calculated by means of mixed ANOVAs for the six subscales of the BREQ-3. There was a significant main effect of time with regard to the introjected (F(1, 38) = 6.01, p = 0.02, partial η² = 0.14; baseline > post-treatment) and the integrated modulation style (F(1, 38) = 5.93, p = 0.02, partial η² = 0.14; baseline < post-treatment). Concerning all other subscales of the BREQ-3, there were no significant main effects of time or group and no significant interaction effects of time × group (all p > 0.05).

Discussion

This is the first study examining whether ABM can be used to increase objective and subjective amounts of PA as well as approach bias toward PA in a healthy, normal-weight sample. The results clearly demonstrate that in the current form, ABM is not ready to be used for this purpose.

No effects of ABM on objective and subjective PA

Previous studies found a positive effect of ABM on unhealthy behavior (e.g. Brockmeyer et al., 2015; Schumacher et al., 2016; Wiers et al., 2011) and an influence of implicit processes on PA (e.g. Cheval et al., 2014; Conroy et al., 2010; Hannan et al., 2018). By contrast, we did not find any effect of ABM on objective or subjective PA.

The unexpected absence of an ABM effect on PA needs to be discussed elaborately in light of the differences between the present and previous studies. First, previous studies found stronger ABM treatment effects in clinical samples, for example, alcohol-dependent patients seeking help (Wiers et al., 2011) in contrast to hazardous drinking students participating in a study (mainly to earn money; Wiers et al., 2010), indicating that the participants’ motivation to change might be one crucial factor that enables ABM to be effective. In this study, we only included individuals who declared that they were motivated to increase their PA level. However, the fact that all participants received payment or course credits calls into question whether participants were actually motivated to change their PA levels. Second, the present sample only included individuals of normal weight. In a study by Chevance et al. (2017), implicit attitudes were associated with PA among obese individuals, but not in the general population. As previous studies (e.g. Nederkoorn et al., 2010; Sutin et al., 2011) indicated that individuals with obesity differed in their self-regulative abilities from the general population, and Chevance et al. (2017) concluded that implicit attitudes may wield more influence over practicing PA in obese people. Consequently, this study should be replicated in a sample of obese individuals (BMI > 30 kg/m²).

Third, previous research mainly applied ABM training to reduce undesirable/unhealthy behavior. To our knowledge, there has been very limited research investigating whether desired behavior can be induced by means of ABM (e.g. Hahn et al., 2019; Taylor and Amir, 2012). However, in a very recent study by Hahn et al. (2019), ABM successfully induced positive attitudes toward condoms and increased the use of condoms 3 months later. Further research is needed to verify whether ABM is suitable to increase desired behavior or if other treatment approaches are more fruitful. For example, evaluative conditioning was effective to alter automatic evaluations of exercise and lead to subsequent PA level increases (Antoniewicz and Brand, 2016).

ABM has been shown to be effective in the treatment of alcohol dependence (Eberl et al., 2013; Wiers et al., 2010, 2011). Within that context, behavior changes in response to specific stimuli (visual alcohol cues) were targeted. As PA is a rather broad category including subcategories that differ in their extent to be intrinsically motivating (e.g. riding a bike in contrast to doing push-ups), future studies should investigate whether ABM is effective to increase specifically defined subcategories of PA (e.g. cycling) or intrinsically motivated subcategories of PA.

Based on findings by Eberl et al. (2014) on the optimal dosage for AAT training in alcohol dependence, we used six training sessions in this study. However, in the study by Eberl et al. (2014), still a number of patients improved further with more than six training sessions (Eberl et al., 2014). Future studies should thus investigate whether more sessions are needed to increase PA. In addition, in previous studies that found positive effects of ABM, the intervention was used as an add-on to multimodal inpatient treatment (Eberl et al., 2013; Wiers et al., 2010, 2011). It could be that ABM imparts its beneficial effects best when embedded in a context in which participants strongly focus on altering their behavior/lifestyle. This would again suggest to test the ABM intervention as an add-on in a sample of obese individuals who take part in a lifestyle intervention.

No effect of ABM on automatic approach biases toward PA

In the context of alcohol dependence and unhealthy eating, ABM influenced approach biases toward specific stimuli (Schumacher et al., 2016; Wiers et al., 2011). By contrast, ABM had no effect on automatic approach biases toward PA in this study. This was consistent for both measures of automatic approach biases — the AAT and the SRCT — and accords with the finding that ABM had no effect on PA. The above discussed reasons for this observation apply here as well. Apart from that, current research challenges the reliability and validity of existing versions of the AAT (Kersbergen et al., 2015; Meule et al., 2019). In a recent study by Lender et al. (2018), the AAT could only indicate an approach bias toward food if the participants directly responded to the content of the stimuli but not when they were instructed to respond to task-irrelevant features (e.g. picture outline). In addition, Van Dessel et al. (2018) pointed out that during the AAT, participants do not learn the consequences of their approach-avoidance responses which might hinder stronger effects.

Surprisingly, there was a significant effect of time on the SRCT in the opposite as expected direction, indicating that participants in both groups showed a reduced approach bias for PA post-treatment. Both baseline and post-treatment bias scores were positive (pointing to an approach bias toward PA); however, this bias was smaller post-treatment. Especially with regard to the ABM group, this is utterly against our hypothesis. However, it is possible that participants of the ABM group who did not observe the desired PA increase were disappointed. This frustration could have reduced the participants’ social desirability and might have let them to deliberately manipulate their reactions in the SRCT.

No effect of ABM on motivation

In line with the above-mentioned findings, ABM did not affect subjective motivation to be physically active. However, there was an effect of time on the subscales of introjected and integrated motivation style, irrespective of group. Whereas ratings concerning the introjected motivation style decreased over time, ratings of the integrated motivation style increased. According to self-determination theory (Deci and Ryan, 1985; Ryan and Deci, 2000), introjected regulation describes behavior that occurs due to inner stress, for instance, bad conscience, in order to prevent negative affect. Integrated motivation occurs in relation to an individuals’ goals and norms that the person identifies with and that are integrated in the person’s self-perception, for example, to be a highly performing athlete. In line with previous research that found an increase of motivation to be physically active as well as actual PA by use of pedometers, the fact that the participants took part in a study investigating PA as well as wearing a pedometer every day might have affected these variables (Bravata et al., 2007; Kang et al., 2009; Tudor-Locke, 2002). For example, one might expect that taking part in the study and wearing a pedometer might decrease participants’ potential bad conscience and thus reduces integrated motivation style.

Limitations

As both participants and the experimenter were not blinded, social desirability and demand effects may have confounded the results in this study (Adair, 1984). The sample of healthy, mostly female students of normal weight limits the generalization of the results to other, more heterogeneous populations, for example, individuals with obesity. However, with regard to prevention as an important scope of application, healthy students who are in a special phase of life regularly exposed to a great level of stress, but young and malleable at the same time might represent an important addressee to benefit from such programs (Stowell et al., 2019). Although only individuals took part in the study who declared to be motivated to increase their PA levels, all participants received payment or course credits which questions this information value. The reliability and validity of the outcome measures are restricted, especially with regard to the AAT in order to detect approach bias tendencies (Kersbergen et al., 2015; Lender et al., 2018). In addition, the psychometric properties of the BREQ-3 (Rausch Osthoff, 2017) remain to be assessed. Wearing a pedometer for 2 weeks is a rather extensive requirement, which might elicit a lack of compliance in the participants. Unfortunately, we could not verify whether the participants actually wore the pedometer during all of their waking phase. Daily reminders or keeping a diary might enhance the participants’ compliance. However, such procedures produce additional work for the participants. The validity of a pedometer as a measurement of PA has to be discussed as well. In general, capturing the quantity and quality of PA objectively and precisely poses a challenge. As walking and running are part of most PA patterns, pedometers represent a useful indicator of daily movement (Vanhees et al., 2005). Pedometers are small, cheap, and their application is associated with small efforts and low risks of injury while providing objective and valid information on PA (Bassett and John, 2010). In comparison with subjective ratings of PA (Adams et al., 2005), pedometers reduce biases such as social desirability. However, the accuracy of measurement might vary as a function of the motion speed (Crouter et al., 2003; Giannakidou et al., 2012; Melanson et al., 2004). In addition, movement patterns different from walking and running, such as cycling, swimming, or weight training, are not captured equally correctly (Vanhees et al., 2005). Apart from that, PA thermogenesis can be differentiated in volitional exercise thermogenesis, including purposeful PA such as sport, and non-exercise thermogenesis (NEAT) comprised every PA excluding volitional exercise, such as activities of daily living, fidgeting, sitting, standing, talking, walking to work, or maintaining posture (Levine, 2002). According to Cheval et al. (2014), implicit processes may influence NEAT in a unique way. However, its enormous variety poses a challenge to study NEAT (Levine, 2002) and pedometers are not suitable as they only assess parts of its diversity. Within the limits of this pilot study measuring PA by means of a pedometer can be considered as economic and justified. In future studies, additional parameters, such as heart rate and energy consumption, could be combined with pedometers in order to assess PA as accurately as possible.

Apart from that, potential confounders that might have influenced the participants’ PA levels, such as rainy weather, the individual’s field of study or workload, were not systematically assessed. In future studies, the participants’ physical activity levels prior to the intervention should be assessed as individuals might benefit differentially from ABM as a function of prior activity levels. In line with Eberl et al. (2014), six training sessions were used in this study. However, as some patients improved further after more than six training sessions (Eberl et al., 2014), it is possible that the optimal dosage and frequency of the training need to be adjusted to increase PA levels. In this pilot study, ABM was used as a stand-alone treatment to increase PA levels. However, it is possible that patients who undergo multicomponent interventions would benefit more from ABM as an add-on treatment.

Conclusion

The current pilot study did not find an effect of ABM on objective or subjective PA. Hence, in its current form, ABM as a stand-alone treatment is not ready to increase PA levels in healthy, normal-weight individuals. In light of the discussed limitations of this pilot study and the potential reasons for the null findings, futures studies are needed to shed light on some key factors that might influence ABM effectiveness: modulating effects, such as the participants’ motivation to change, in order to identify appropriate addressees; differential effects of ABM as stand-alone versus add-on treatment; the optimal dosage of ABM with regard to different addressees and contexts; the generalization of previous ABM treatment effects concerning the reduction of unhealthy behavior on the subject of increasing healthy behavior; and the validity of approach bias assessment.

Footnotes

Author contributions

M.A.P. analyzed the data and drafted the first version of the manuscript; M.Z. collected the data and helped with data preparation and analysis; T.B. conceptualized the study and prepared the methodology, acted as project administrator, supervised all parts of the study, and reviewed and edited the first draft of the manuscript. All authors read and approved the final version of the manuscript.

Data availability

The datasets analyzed in the current study are available from the corresponding author on reasonable request.

Declaration of conflicting interests

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Mira A Preis

References

Adair

(1984) The Hawthorne effect: A reconsideration of the methodological artifact. Journal of Applied Psychology 69(2): 334–345.

Adams

Matthews

Ebbeling

, et al. (2005) The effect of social desirability and social approval on self-reports of physical activity. American Journal of Epidemiology 161(4): 389–398.

Altman

Wilfley

(2015) Evidence update on the treatment of overweight and obesity in children and adolescents. Journal of Clinical Child and Adolescent Psychology 44(4): 521–537.

Antoniewicz

Brand

(2016) Learning to like exercising: Evaluative conditioning changes automatic evaluations of exercising and influences subsequent exercising behavior. Journal of Sport & Exercise Psychology 38(2): 138–148.

Aulbach

Knittle

Haukkala

(2019) Implicit process interventions in eating behaviour: A meta-analysis examining mediators and moderators. Health Psychology Review 13(2): 179–208.

Bassett

John

(2010) Use of pedometers and accelerometers in clinical populations: Validity and reliability issues. Physical Therapy Reviews 15(3): 135–142.

Battenberg

Donohoe

Robertson

, et al. (2017) The accuracy of personal activity monitoring devices. Seminars in Arthroplasty 28(2): 71–75.

Becker

Jostmann

Wiers

, et al. (2015) Approach avoidance training in the eating domain: Testing the effectiveness across three single session studies. Appetite 85: 58–65.

Becker

Barth

Smits

JAJ

, et al. (2019) Positivity-approach training for depressive symptoms: A randomized controlled trial. Journal of Affective Disorders 245: 297–304.

10.

Boffo

Zerhouni

Gronau

, et al. (2019) Cognitive bias modification for behavior change in alcohol and smoking addiction: Bayesian meta-analysis of individual participant data. Neuropsychology Review 29(1): 52–78.

11.

Bravata

Smith-Spangler

Sundaram

, et al. (2007) Using pedometers to increase physical activity and improve health: A systematic review. Journal of the American Medical Association 298(19): 2296–2304.

12.

Brockmeyer

Friederich

H-C

Küppers

, et al. (2019) Approach bias modification training in bulimia nervosa and binge-eating disorder: A pilot randomized controlled trial. The International Journal of Eating Disorders 52: 520–529.

13.

Brockmeyer

Hahn

Reetz

, et al. (2015) Approach bias modification in food craving: A proof-of-concept study. European Eating Disorders Review: The Journal of the Eating Disorders Association 23(5): 352–360.

14.

Cacioppo

Priester

Berntson

(1993) Rudimentary determinants of attitudes: II. Arm flexion and extension have differential effects on attitudes. Journal of Personality and Social Psychology 65(1): 5.

15.

Caspersen

Powell

Christenson

(1985) Physical activity, exercise, and physical fitness: Definitions and distinctions for health-related research. Public Health Reports 100(2): 126–131.

16.

Chang

S-H

Stoll

CRT

Song

, et al. (2014) The effectiveness and risks of bariatric surgery: An updated systematic review and meta-analysis, 2003-2012. JAMA Surgery 149(3): 275–287.

17.

Cheval

Sarrazin

Pelletier

(2014) Impulsive approach tendencies towards physical activity and sedentary behaviors, but not reflective intentions, prospectively predict non-exercise activity thermogenesis. PLoS ONE 9(12): e115238.

18.

Chevance

Caudroit

Romain

, et al. (2017) The adoption of physical activity and eating behaviors among persons with obesity and in the general population: The role of implicit attitudes within the Theory of Planned Behavior. Psychology, Health & Medicine 22(3): 319–324.

19.

Cohen

(1988) Statistical Power Analysis for the Behavioral Sciences. Hillsdale, NJ: Erbaum.

20.

Conroy

Hyde

Doerksen

, et al. (2010) Implicit attitudes and explicit motivation prospectively predict physical activity. Annals of Behavioral Medicine 39(2): 112–118.

21.

Courcoulas

Yanovski

Bonds

, et al. (2014) Long-term outcomes of bariatric surgery: A national institutes of health symposium. JAMA Surgery 149(12): 1323–1329.

22.

Craig

Marshall

Sjöström

, et al. (2003) International physical activity questionnaire: 12-country reliability and validity. Medicine and Science in Sports and Exercise 35(8): 1381–1395.

23.

Crouter

Schneider

Karabulut

, et al. (2003) Measuring steps, distance, and energy cost. Medicine and Science in Sports and Exercise 35(8): 1455–1460.

24.

Deci

Ryan

(1985) Intrinsic Motivation and Self-Determination in Human Behavior. New York: Plenum.

25.

Deci

Ryan

(2000) The “what” and “why” of goal pursuits: Human needs and the self-determination of behavior. Psychological Inquiry 11(4): 227–268.

26.

Deutsch

Strack

(2006) Reflective and impulsive determinants of addictive behavior. Handbook of Implicit Cognition and Addiction 16: 45–57.

27.

Dombrowski

Avenell

Sniehott

(2010) Behavioural interventions for obese adults with additional risk factors for morbidity: Systematic review of effects on behaviour, weight and disease risk factors. Obesity Facts 3(6): 377–396.

28.

Eberl

Wiers

Pawelczack

, et al. (2013) Approach bias modification in alcohol dependence: Do clinical effects replicate and for whom does it work best? Developmental Cognitive Neuroscience 4: 38–51.

29.

Eberl

Wiers

Pawelczack

, et al. (2014) Implementation of approach bias re-training in alcoholism—How many sessions are needed? Alcoholism: Clinical and Experimental Research 38(2): 587–594.

30.

Elfhag

Rössner

(2005) Who succeeds in maintaining weight loss? A conceptual review of factors associated with weight loss maintenance and weight regain. Obesity Reviews 6(1): 67–85.

31.

Field

Kiernan

Eastwood

, et al. (2008) Rapid approach responses to alcohol cues in heavy drinkers. Journal of Behavior Therapy and Experimental Psychiatry 39(3): 209–218.

32.

Fitzpatrick

Coughlin

Appel

, et al. (2015) Application of latent class analysis to identify behavioral patterns of response to behavioral lifestyle interventions in overweight and obese adults. International Journal of Behavioral Medicine 22(4): 471–480.

33.

Giannakidou

Kambas

Ageloussis

, et al. (2012) The validity of two Omron pedometers during treadmill walking is speed dependent. European Journal of Applied Physiology 112(1): 49–57.

34.

Grave

Calugi

Molinari

, et al. (2005) Weight loss expectations in obese patients and treatment attrition: An observational multicenter study. Obesity Research 13(11): 1961–1969.

35.

Hagströmer

Oja

Sjöström

(2006) The International Physical Activity Questionnaire (IPAQ): A study of concurrent and construct validity. Public Health Nutrition 9(6): 755–762.

36.

Hahn

Simons

, et al. (2019) Can cognitive bias modification simultaneously target two behaviors? Approach bias retraining for alcohol and condom use. Clinical Psychological Science 7: 1078–1093.

37.

Hallal

Andersen

Bull

, et al. (2012) Global physical activity levels: Surveillance progress, pitfalls, and prospects. The Lancet 380(9838): 247–257.

38.

Hannan

Moffitt

Neumann

, et al. (2018) Implicit approach–avoidance associations predict leisure-time exercise independently of explicit exercise motivation. Sport, Exercise, and Performance Psychology 8: 210–222.

39.

Haskell

Blair

Hill

(2009) Physical activity: Health outcomes and importance for public health policy. Preventive Medicine 49(4): 280–282.

40.

Houwer

de Crombez

Baeyens

, et al. (2001) On the generality of the affective Simon effect. Cognition & Emotion 15(2): 189–206.

41.

Huang

, et al. (2016) Validity of FitBit, Jawbone UP, Nike+ and other wearable devices for level and stair walking. Gait & Posture 48: 36–41.

42.

Kakoschke

Kemps

Tiggemann

(2017) Approach bias modification training and consumption: A review of the literature. Addictive Behaviors 64: 21–28.

43.

Kang

Marshall

Barreira

, et al. (2009) Effect of pedometer-based physical activity interventions. Research Quarterly for Exercise and Sport 80(3): 648–655.

44.

Kersbergen

Woud

Field

(2015) The validity of different measures of automatic alcohol action tendencies. Psychology of Addictive Behaviors: Journal of the Society of Psychologists in Addictive Behaviors 29(1): 225–230.

45.

King

Hsu

Belle

, et al. (2012) Pre- to postoperative changes in physical activity: Report from the longitudinal assessment of bariatric surgery-2 (LABS-2). Surgery for Obesity and Related Diseases: Official Journal of the American Society for Bariatric Surgery 8(5): 522–532.

46.

Krishna

Strack

(2017) Reflection and impulse as determinants of human behavior. In: Meusburger

Werlen

Suarsana

(eds) Knowledge and Action. Cham: Springer, pp. 145–167.

47.

Lender

Meule

Rinck

, et al. (2018) Measurement of food-related approach–avoidance biases: Larger biases when food stimuli are task relevant. Appetite 125: 42–47.

48.

Levine

(2002) Non-exercise activity thermogenesis (NEAT). Best Practice & Research Clinical Endocrinology & Metabolism 16(4): 679–702.

49.

Livhits

Mercado

Yermilov

, et al. (2010) Exercise following bariatric surgery: Systematic review. Obesity Surgery 20(5): 657–665.

50.

Manning

Staiger

Hall

, et al. (2016) Cognitive bias modification training during inpatient alcohol detoxification reduces early relapse: A randomized controlled trial. Alcoholism: Clinical and Experimental Research 40(9): 2011–2019.

51.

Melanson

Knoll

Bell

, et al. (2004) Commercially available pedometers: Considerations for accurate step counting. Preventive Medicine 39(2): 361–368.

52.

Mensink

GBM

Schienkiewitz

Haftenberger

, et al. (2013) Übergewicht und Adipositas in Deutschland: Ergebnisse der Studie zur Gesundheit Erwachsener in Deutschland (DEGS1). Bundesgesundheitsblatt, Gesundheitsforschung, Gesundheitsschutz 56(5–6): 786–794.

53.

Meule

Richard

Lender

, et al. (2019) Measuring approach–avoidance tendencies towards food with touchscreen-based arm movements. Psychological Research. Epub ahead of print 4 May. DOI: 10.1007/s00426-019-01195-1.

54.

Moldovan

David

(2011) Effect of obesity treatments on eating behavior: Psychosocial interventions versus surgical interventions a systematic review. Eating Behaviors 12(3): 161–167.

55.

Myers

McAuley

Lavie

, et al. (2015) Physical activity and cardiorespiratory fitness as major markers of cardiovascular risk: Their independent and interwoven importance to health status. Progress in Cardiovascular Diseases 57(4): 306–314.

56.

Nauta

Hospers

Jansen

(2001) One-year follow-up effects of two obesity treatments on psychological well-being and weight. British Journal of Health Psychology 6(3): 271–284.

57.

Nederkoorn

Houben

Hofmann

, et al. (2010) Control yourself or just eat what you like? Weight gain over a year is predicted by an interactive effect of response inhibition and implicit preference for snack foods. Health Psychology 29(4): 389–393.

58.

Neumann

Strack

(2000) Approach and avoidance: The influence of proprioceptive and exteroceptive cues on encoding of affective information. Journal of Personality and Social Psychology 79(1): 39–48.

59.

Fleming

Robinson

, et al. (2014) Global, regional, and national prevalence of overweight and obesity in children and adults during 1980-2013: A systematic analysis for the Global Burden of Disease Study 2013. The Lancet 384(9945): 766–781.

60.

Penedo

Dahn

(2005) Exercise and well-being: A review of mental and physical health benefits associated with physical activity. Current Opinion in Psychiatry 18(2): 189–193.

61.

Phing

Hoa

Hua

(2017) Overweight and obesity in relation to cardiovascular risk factors among university students in Malaysia. Romanian Journal of Diabetes Nutrition and Metabolic Diseases 24(3): 53.

62.

Puhl

Heuer

(2010) Obesity stigma: Important considerations for public health. American Journal of Public Health 100(6): 1019–1028.

63.

Rasmussen

Laumann

(2013) The academic and psychological benefits of exercise in healthy children and adolescents. European Journal of Psychology of Education 28(3): 945–962.

64.

Rausch Osthoff

A-K

(2017) Behavioural Regulation in Exercise Questionnaire (BREQ-3)-Deutsche Version. Winterthur: Zürcher Hochschule für Angewandte Wissenschaften.

65.

Rinck

Becker

(2007) Approach and avoidance in fear of spiders. Journal of Behavior Therapy and Experimental Psychiatry 38(2): 105–120.

66.

Rinck

Wiers

Becker

, et al. (2018) Relapse prevention in abstinent alcoholics by cognitive bias modification: Clinical effects of combining approach bias modification and attention bias modification. Journal of Consulting and Clinical Psychology 86(12): 1005–1016.

67.

Rothman

Sheeran

Wood

(2009) Reflective and automatic processes in the initiation and maintenance of dietary change. Annals of Behavioral Medicine 38(Suppl. 1): s4–s17.

68.

Ryan

Deci

(2000) Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. American Psychologist 55(1): 68–78.

69.

Schumacher

Kemps

Tiggemann

(2016) Bias modification training can alter approach bias and chocolate consumption. Appetite 96: 219–224.

70.

Shaw

Gennat

O’Rourke

, et al. (2006) Exercise for overweight or obesity. Cochrane Database of Systematic Reviews 4: CD003817.

71.

Stowell

Lewis

Brooks

(2019) Perceived stress, substance use, and mental health issues among college students in the Midwest. Journal of Prevention & Intervention in the Community. Epub ahead of print 20 August. DOI: 10.1080/10852352.2019.1654263.

72.

Strack

Deutsch

(2004) Reflective and impulsive determinants of social behavior. Personality and Social Psychology Review 8(3): 220–247.

73.

Sutin

Ferrucci

Zonderman

, et al. (2011) Personality and obesity across the adult life span. Journal of Personality and Social Psychology 101(3): 579–592.

74.

Taylor

Amir

(2012) Modifying automatic approach action tendencies in individuals with elevated social anxiety symptoms. Behaviour Research and Therapy 50(9): 529–536.

75.

Tsianakas

Harris

Ream

, et al. (2017) CanWalk: A feasibility study with embedded randomised controlled trial pilot of a walking intervention for people with recurrent or metastatic cancer. BMJ Open 7(2): e013719.

76.

Tudor-Locke

(2002) Taking Steps Toward Increased Physical Activity: Using Pedometers to Measure and Motivate. President’s Council on Physical Fitness and Sports Research Digest.

77.

Urbaniak

Plous

(2013) Research randomizer. Available at: http://www.randomizer.org/

78.

van Dessel

Eder

Hughes

(2018) Mechanisms underlying effects of approach-avoidance training on stimulus evaluation. Journal of Experimental Psychology: Learning, Memory, and Cognition 44(8): 1224–1241.

79.

Vanhees

Lefevre

Philippaerts

, et al. (2005) How to assess physical activity? How to assess physical fitness? European Journal of Cardiovascular Prevention & Rehabilitation 12(2): 102–114.

80.

Wang

McPherson

Marsh

, et al. (2011) Health and economic burden of the projected obesity trends in the USA and the UK. The Lancet 378(9793): 815–825.

81.

Warburton

DER

Nicol

Bredin

SSD

(2006) Health benefits of physical activity: The evidence. CMAJ 174(6): 801–809.

82.

Wiers

Eberl

Rinck

, et al. (2011) Retraining automatic action tendencies changes alcoholic patients’ approach bias for alcohol and improves treatment outcome. Psychological Science 22(4): 490–497.

83.

Wiers

Rinck

Kordts

, et al. (2010) Retraining automatic action-tendencies to approach alcohol in hazardous drinkers. Addiction 105(2): 279–287.

84.

Wing

Phelan

(2005) Long-term weight loss maintenance–. The American Journal of Clinical Nutrition 82(1): 222S–225S.

85.

Wing

Papandonatos

Fava

, et al. (2008) Maintaining large weight losses: The role of behavioral and psychological factors. Journal of Consulting and Clinical Psychology 76(6): 1015.

86.

Withrow

Alter

(2011) The economic burden of obesity worldwide: A systematic review of the direct costs of obesity. Obesity Reviews: An Official Journal of the International Association for the Study of Obesity 12(2): 131–141.

87.

Wittekind

Lüdecke

Cludius

(2019) Web-based approach bias modification in smokers: A randomized-controlled study. Behaviour Research and Therapy 116: 52–60.

88.

Yusoff

NAM

Ganeson

Ismail

, et al. (2018) Physical activity level among undergraduate students in Terengganu, Malaysia using pedometer. Journal of Fundamental and Applied Sciences 10(1S): 512–522.