The approach and avoidance task (AAT) has evolved as a promising treatment add-on in the realm of psychology. Certain psychiatric diseases, such as behavioural addictions, social anxiety disorder, and arachnophobia, are characterized by a dysfunctional tendency to either approach or avoid disease-specific objects. This tendency can be measured by means of the approach and avoidance task. More precisely, by a diagnostic AAT, in which participants are instructed to react upon the format or the frame colour of a picture. For instance, pictures have to be pushed away if they are presented in landscape format and pulled towards oneself if they are presented in portrait format (or vice versa). Hence, the format (or the frame colour) becomes the task-relevant dimension, whereas the content of the picture becomes irrelevant for task completion. However, what generally becomes obvious in the psychiatric diseases mentioned above is that the task-irrelevant dimension (picture content) exerts an influence on reaction times. For instance, smokers are generally faster to respond to smoking-related pictures, when presented in a format requiring them to pull towards themselves, and slower to respond, if smoking pictures are shown in the format requiring them to push away a joystick (Wiers et al., 2013). This behavioural tendency has been termed an approach bias for cigarettes or smoking.

In order to counteract these dysfunctional approach or avoidance tendencies, an AAT-training has been developed. In this training participants/patients learn to either avoid or approach disease-specific objects. Smokers, for instance, learn to avoid smoking-related pictures by pushing or swiping the image away. It has been shown that these trainings can lower cigarette consumption among current smokers (Machulska, Zlomuzica, Rinck, Assian, & Margraf, 2016). The aim of the current study is to test whether the avoidance gesture is as important as suggested by the AAT's name or whether inhibiting the urge to approach smoking-related content might be enough to bring about the effect. Furthermore, possible changes in general and domain-specific (i.e. smoking-related) inhibition capacity, that might mediate the effect, will be assessed. Another focus of study will be on functional as well as structural neuronal changes, emerging as a consequence of the AAT-training.

The approach and avoidance task (AAT) has turned out as both a promising diagnostic tool as well as treatment add-on in psychological science. The AAT constitutes one form of cognitive bias modification (CBM), which has been shown to be particularly effective in the field of behavioural addictions (Eberl et al., 2013; Wiers, Eberl, Rinck, Becker, & Lindenmeyer, 2011). The general logic underlying the AAT is to carry out actions that are either compatible or incompatible with an individual's action tendencies. For instance, nicotine addicted individuals tend to approach tobacco-related stimuli faster than control pictures, when they are instructed to react upon the format of a picture and not to its' content. This tendency of comparatively faster approaching and slower avoiding tobacco-related content has been termed an approach bias nicotine/tobacco. The AAT as a therapeutic tool tries to counteract or at least to attenuate approach or avoidance biases by instructing patients to carry out approach or avoidance gestures that are in conflict with an individual's acquired action tendencies.

Whereas the general effectiveness of the AAT as a clinical intervention has been demonstrated several times, little is known about possible mechanisms that might subserve these effects. Therefore, the current study is dedicated to shed some light on one such potential mechanism, i.e. the role of the avoidance gesture within the alcohol-AAT.

As already suggested by the name of the AAT, the avoidance gesture seems to be a key ingredient in bringing about therapeutic effects. However, recent empirical evidence has brought about some interesting findings, giving rise to an alternative explanation.

A study by Kühn et al. (2017), contrary to common-held beliefs, indicated that inhibition capacity can be trained. Inhibition, in turn, consistently has been linked to psychopathology and all kinds of behavioural addictions (Smith, Mattick, Jamadar, & Iredale, 2014). The game by Kühn et al. (2017), used to train inhibition, resembled the AAT in several ways, e.g. certain stimuli appearing on a treadmill had to be collected by swiping towards oneself and others had to be ignored and the objects slowly disappeared. The latter element contrasts with the AAT, since the ignored objects don't have to be pushed away. However, it resembles the AAT in the sense that in both cases stimuli slowly fade out of the screen and eventually disappear. These parallel let to the assumption that a new form of the alcohol AAT training might be equally effective in lowering relapse rates among alcoholic patients. More precisely, within the newly conceptualized AAT training, patients are instructed to inhibit the urge to respond in response to alcohol-related content and to observe the stimuli fading out of the screen. In contrast to the classical AAT training, this zooming out of alcoholic stimuli is not conditional on the avoidance gesture, i.e. swiping/pushing away the stimulus.

It is hypothesized that compared to a control group, in which tobacco and control stimuli have to be swiped to the left and right, both the classical AAT-and the inhibition group will be more successful in stopping or reducing smoking. The intervention includes a training period of four weeks. No intergroup differences in terms of smoking cessation and tobacco-related approach bias are expected for the classical AAT group and the inhibition group after the intervention.

Furthermore, it will be explored whether inhibition capacity changes as a result of the intervention, the assumption being that inhibition capacity increases for both experimental groups, whereas no changes are expected for the control group. An interesting question concerns the domain-specificity of potential effects. In other words, does inhibition capacity improve only for the to be trained domain (i.e. tobacco) or does the general inhibition capacity improve, irrespective of the to be trained domain? Predictions on this issue are not straightforward, but it is assumed that both the general and domain-specific inhibition capacity improve for the experimental groups, when compared to the control group. Slightly more pronounced effects are expected for the domain-specific task, since the training calls for comparable skills.

Hypotheses concerning experimental changes in brain structure are hard to make, as, to the best of our knowledge, no research has been done on this topic up until now. Therefore, whole-brain analyses will be run on all voxels of the brain in order to explore any experimental changes both on the within and between subject level.

The same goes for functional changes in a Stop Signal Task (SST) and a cue reactivity paradigm. It will be explored whether the classical AAT group and the inhibition group show less reward-related activity in response to addiction-related stimuli than participants in the control group. Furthermore, it will be explored whether brain regions commonly associated with inhibition capacity show more activity in response to smoking-related stimuli, since this has been learned throughout the training period.

All secondary outcome measures (e.g. smoking-related questionnaires, Beck's Depression Inventory, Barratt's Impulsivity Scale, AUDIT) assess whether the experimental manipulations lead to a reduction in psychopathological symptoms. Therefore, both experimental groups are expected to show post-interventional declines in nicotine-dependence measures and other measures associated with dependence, such as mental well-being and impulsivity. In contrast, no changes across time are expected for the control group.

In order to test prior mentioned hypotheses, a randomized controlled trial will be conducted. Each of the three groups (inhibition group, classical AAT group, and control group) will consist out of 25 mentally healthy participants, who just stopped smoking or just start an attempt to quit smoking.

In total the study consists out of three points of measurement. After the baseline testing, which assesses participants' approach bias for tobacco, their tobacco history, tobacco consumption within the last months, a multitude of smoking-related questionnaires (see secondary outcome measures), the two inhibition tasks, brain structure and brain function, participants are provided with a tablet on which the AAT app is installed. The participants are asked to train with the app for at least 15 minutes daily within a four-week period. Following this training, the second point of measurement will take place. The general procedure is identical to the first point of measurement. Another four weeks later, the third point of measurement marks the end of the study. Once again, the measures are the same except for the fact that no MRI scan is realized on this last occasion, since the study is primarily interested in post-interventional brain changes.

Inclusion Criteria:

• being mentally healthy
• just having stopped smoking or having the urge to do so
• adequate language skills
• being right-handed

Exclusion Criteria:

• MRI contra indications, such as claustrophobia, tinnitus, and having any non-removable ferromagnetic artefacts within the body
• somatic or neurological disease that precludes the proper execution of the study

Participants do not know that there exists more than one condition.