The pulling power of chocolate: Effects of approach–avoidance training on approach bias and consumption

Highlights

• Inclusion of both approach and avoidance training does alter action tendencies to chocolate.

• There was no effect on chocolate consumption in a later taste test.

• Proof of concept is not yet established for approach avoidance training as an intervention.

Abstract

Previous research has shown that action tendencies to approach alcohol may be modified using computerized Approach–Avoidance Task (AAT), and that this impacted on subsequent consumption. A recent paper in this journal (Becker, Jostman, Wiers, & Holland, 2015) failed to show significant training effects for food in three studies: Nor did it find effects on subsequent consumption. However, avoidance training to high calorie foods was tested against a control rather than Approach training. The present study used a more comparable paradigm to the alcohol studies. It randomly assigned 90 participants to ‘approach’ or ‘avoid’ chocolate images on the AAT, and then asked them to taste and rate chocolates. A significant interaction of condition and time showed that training to avoid chocolate resulted in faster avoidance responses to chocolate images, compared with training to approach it. Consistent with Becker et al.'s Study 3, no effect was found on amounts of chocolate consumed, although a newly published study in this journal (Schumacher, Kemps, & Tiggemann, 2016) did do so. The collective evidence does not as yet provide solid basis for the application of AAT training to reduction of problematic food consumption, although clinical trials have yet to be conducted.

Introduction

Current psychological strategies to address excessive weight have limited effects (Hartmann-Boyce et al., 2014, Michie et al., 2009), especially over a medium-term follow-up (Barte et al., 2010). A different approach may be required.

A focus of some recent laboratory research has been on the use of cognitive tasks to modify attention or analogue responses to food stimuli. Some have focused on changing attentional bias on a dot-probe task (Kakoschke et al., 2014, Kemps et al., 2014). Others have modified associations of chocolate with words related to approach or avoidance (Kemps, Tiggemann, Martin, & Elliott, 2013). Training go-no go responses to images can not only reduce immediately subsequent consumption of chocolate and other high-calorie foods (Houben and Jansen, 2015, Veling et al., 2013), but there is emerging evidence that it may improve weight loss (Veiling, van Koningsbruggen, Aarts, & Stroebe, 2014).

Even more direct training of behavioural tendencies may be obtained by using the Approach Avoidance Task (AAT; Rinck & Becker, 2007), which typically involves pulling a joystick to make a picture larger (simulating approach) or pushing it away to make it smaller (simulating avoidance). These responses have high face validity as analogues of pulling or pushing away a target such as a bar of chocolate in the natural environment. Action tendencies to approach alcohol are positively associated with both previous consumption of alcohol (Barkby et al., 2012, Peeters et al., 2012), and with alcohol consumption measured on subsequent laboratory-based ‘taste tests’ (e.g. Wiers, Rinck, Kordts, Houben, & Strack, 2010). Changes in approach biases can be elicited by training approach or avoidance of a particular target category (e.g. alcohol vs. soft drinks; Wiers et al., 2010). When repeated trials on the AAT are configured so participants more consistently pull in response to alcohol pictures and push in response to soft drink, this increases approach biases for alcohol, while the reverse contingency decreases them. When heavy drinkers are successfully trained to approach or avoid alcohol pictures, this also affects the amount they consume in a subsequent taste test (Wiers et al., 2010). Importantly, two randomized controlled trials have also demonstrated that multiple AAT sessions increase rates of abstinence in alcohol-dependent patients receiving cognitive-behavioural therapy (Eberl et al., 2013, Wiers et al., 2011).

A recent paper by Becker, Jostmann, Wiers and Holland (2015), published in this journal, described three studies applying the AAT to foods. Unlike the original study on alcohol, avoidance training was compared with sham training, which involved a 50% association of approach and avoidance to both high- and low-calorie foods. In Study 1 there was a trend (p = .058) towards greater avoidance in the experimental condition, but this was not seen in the other two studies. No effects on post-session consumption were seen in Studies 1 and 2, and in Study 3, participants trained to avoid chocolate images actually ate more chocolate in a taste test.

The current study was undertaken independently of the ones by Becker et al. (2015). A key difference in the protocol was that it compared approach versus avoidance training on the AAT, consistent with the initial demonstration study on alcohol by Wiers et al. (2010). Like Becker et al.’s Study 3, we had an undergraduate sample, focused on chocolate, and used a ‘taste test’ of chocolate during the session to examine chocolate consumption. However, in contrast to that study, we did not select participants who intended to reduce chocolate consumption, and we did not elicit chocolate craving before the AAT. We predicted that our AAT procedure would result in differential training effects from avoiding vs. approaching chocolate, and that consumption of chocolate in the taste test would then be greater in the approach than in the avoidance condition. We also examined whether the training altered subjective craving for chocolate, while noting that previous research has sometimes found a disjunction between approach/avoidance training and craving effects (Wiers et al., 2011).