How should we deliver Approach avoidance trainings?

Approach-Avoidance Trainings (AAT) have been shown to be very beneficial in the relapse-preventing treatment of substance use disorders. These trainings usually employ joysticks, and this is severely limiting their use because joysticks are becoming outdated, and they are not well-suited for online delivery. Therefore, this paper is testing the usefulness of different response devices, particularly their reliability and validity during the assessment of approach-avoidance tendencies. Three different devices were compared to each other, namely joystick, computer mouse, and touchscreen. They were used to measure approach tendencies for chocolate (sorry, there cannot be avoidance of chocolate).


Wittekind, C. E., Blechert, J., Schiebel, T., Lender, A., Kahveci, S., & Kühn, S. (2021). Comparison of different response devices to assess behavioral tendencies towards chocolate in the approach-avoidance task. Appetite, 165, 105294.


Commentary by Victoria Manning
Monash University, Australia

I was excited when I learned of this paper, since interest in measuring and manipulating approach bias in the content of appetitive behaviours is growing at an unprecedent rate. Despite the numerous studies over the past decade, it remains unclear how we best measure approach bias, and indeed with joysticks becoming fast becoming obsolete, there is an urgent need to establish the feasibility and accuracy of alternative methods to do so. As well as addressing a knowledge gap, it also advances a burgeoning line of research in approach bias to chocolate, which is of considerable interest among the general public.
This well-designed study makes an important contribution to the literature by comparing the performance of 45 regular chocolate consumers on three different devices administering the Approach-Avoidance Task (AAT); the traditional joystick, mouse and touchscreen, all of which require an arm flection/extension movement in response to images of chocolate or object-related pictures. Low to medium correlations were found with approach bias demonstrated across all 3 devices, with moderate to good internal consistency, though craving scores correlated only with the joystick version, and only when the incongruent version was administered before the congruent version.
This is an important observation worthy of further exploration because CBM researchers have questioned for years whether explicit versus implicit AAT is better. These findings support the case for explicit AAT, but highlight that this may depend on block order. Other strengths were the assessment of multiple outcomes relating to the target domain and the ad-libitum chocolate consumption (unsupervised), pre-registered analyses, and that the fact that testing was restricted to between meal times so that effects of hunger on approach bias were likely minimised.
Readers should however be cautious about overgeneralising these findings, due to its modest sample size and underpowered exploratory analyses, especially for interested in approach bias in clinical populations, given that the sample were a relatively healthy (normal BMI) population consuming the equivalent of only 1 bar of chocolate a day. This limits generalisability to the populations in whom these types of measures may have more clinical importance and where measuring and manipulating approach biases would be more relevant.
Another consideration is that the poorer psychometrics reported with the mouse version could potentially be explained by the requirement to click and hold the left mouse button to move the image, and to return to the centre, by looking away from the screen and down at the mouse position on a mouse pad in between each trial. The touchscreen and grabbing of chocolate is an interesting and novel approach, though this would not readily translate to phones, to permit ecological momentary assessment which would be advantageous as suggested by the authors.
These considerations do not detract from importance of this study which provides a valuable first-step in informing the development of more sensitive measures of approach bias to ‘unhealthy’ foods like chocolate. Future replication on a larger and more clinically-relevant sample (i.e., where the majority want to reduce chocolate consumption) could pave the way for innovative adjunctive treatments like approach bias modification for obesity and overeating, which stand to have a major public health impact.


Commentary by Reinout Wiers
University of Amsterdam, The Netherlands

A paper assessing the approach bias for chocolate in a relatively small sample of healthy volunteers (N=45), why would this be of importance? The reason is that the reliable and valid measurement of cognitive biases is crucial in this field of study, and an approach bias is an important bias in appetitive disorders, including addictions. Moreover, one of the most used devices to assess an approach bias uses a joystick, which is becoming less fashionable among gamers (and therefore less available), for this reason, several studies have started to use alternative ways to assess an approach bias, but few have validated these alternative approaches. This study directly compared three different devices: the most standard version in research (joystick), with a mouse-based version and a smartphone version. Importantly, the approach-bias assessment that correlated best with craving was the joystick, and then specifically when the incongruent block (avoid chocolate) was provided first. The latter finding is not totally surprising, the same has been observed for the often used Implicit Association Test (IAT, Hofmann, Gawronski, Gschwendner, Le, & Schmitt, 2005). The finding that the joystick assessment appears to be the most valid way to assess the approach bias, is in line with other recent findings in different domains, for example, a study including some of the co-authors in the domain of porn found the same in healthy volunteers (Kahveci, van Bockstaele, Blechert, & Wiers, 2020). This is slightly worrying, given the mentioned foreseeable shortage of joysticks, and calls for either hoarding joysticks or developing other valid ways to assess an approach bias in the appetitive domain (e.g., using virtual reality, Rougier et al., 2018). Almost needless to say here that reliable and valid assessment of cognitive biases is crucial for the progress in cognitive bias modification, because if you cannot assess the change in bias in a relatively valid and reliable way, it becomes hard to tell if your manipulation of the bias worked. So an important paper indeed!


Hofmann, W., Gawronski, B., Gschwendner, T., Le, H., & Schmitt, M. (2005). A meta-analysis on the correlation between the Implicit Association Test and explicit self-report measures. Personality and Social Psychology Bulletin, 31(10), 1369–1385.

Kahveci, S., van Bockstaele, B., Blechert, J., & Wiers, R. W. (2020). Pulling for pleasure? Erotic approach-bias associated with porn use, not problems. Learning and Motivation, 72.

Rougier, M., Muller, D., Ric, F., Alexopoulos, T., Batailler, C., Smeding, A., & Aubé, B. (2018). A new look at sensorimotor aspects in approach/avoidance tendencies: The role of visual whole-body movement information. Journal of Experimental Social Psychology, 76(July 2017), 42–53.

Scroll to Top