A meta-analysis of the facilitation of arm flexion and extension movements as a function of stimulus valence

Abstract

This article presents a meta-analysis of research on the affective compatibility effect: the relative facilitation of arm flexion and extension movements, in response to positive and negative stimuli, respectively. Across 68 effect sizes (computed on 3169 participants), a small, significant average compatibility effect emerged (ES = .118; 95% CI [.051, .185]). Importantly, analyses also revealed significant heterogeneity in the set of effect sizes. Moderator analyses were conducted to explain this observed heterogeneity with a view to testing between extant theoretical accounts of the compatibility effect. Affective compatibility effects were significantly larger (1) for face stimuli than for words or pictorial stimuli; (2) when the negative stimuli partly comprising the effect were anger-related; (3) for responses made using vertical button press; (4) when situated aspects of the processing task framed flexion as approach and extension as avoidance; and (5) when explicit response labels framed flexion as positive and extension as negative. Significant reverse compatibility effects emerged (1) when aspects of the processing context framed flexion as avoidance and extension as approach and (2) when explicit response labels framed flexion as negative and extension as positive. The results of the meta-analysis provide little support for the strong embodiment, specific muscle activation account of affective compatibility and are broadly consistent with distance regulation, and, in particular, evaluative coding accounts.

Keywords:

• Affective compatibility
• Attitudes
• Embodiment
• Attitude–behaviour link

The research reported in this article was supported by a grant from the Asian Office of Aerospace R&D (Award No. FA2386-09-4093) and a grant from the Australian Research Council (DP1095323) to Yoshihisa Kashima and Simon M. Laham.

The research reported in this article was supported by a grant from the Asian Office of Aerospace R&D (Award No. FA2386-09-4093) and a grant from the Australian Research Council (DP1095323) to Yoshihisa Kashima and Simon M. Laham.

Notes

¹ During the latter stages of manuscript preparation, we discovered a recent meta-analysis, by Phaf, Mohr, Rotteveel & Wicherts (2014), on the affective compatibility effect. The results of our meta-analysis are broadly consistent with those of Phaf et al., although our work draws on a larger sample of studies. We discuss points of overlap and difference between these two meta-analyses in later sections of the Introduction and relevant sections of the Discussion.

² This link may be more aptly termed the stimulus valence–behaviour link, but we term it A–B on the assumption that respondents will typically have positive attitudes to or evaluations of positively valenced stimuli (and negative attitudes towards negatively valenced stimuli). In so naming the link we maintain consistency with other nomenclatures (e.g., Eder & Rothermund, 2008; Laham et al., 2014).

³ We use compatibility effect to refer to the facilitation of positive-flexion and negative-extension stimulus-response pairings relative to the reverse pairings (negative-flexion; positive-extension). We use reverse compatibility effect to refer to the relative facilitation of positive-extension and negative-flexion S-R pairings compared to the reverse pairings. It should be noted that other researchers use the term compatibility in relation to the congruence of motivational systems and stimulus valence (positive-approach; negative-avoid).

⁴ Accounts of the A–B and B–A links make assumptions about the valence implications of approach and avoidance systems. Typically, the assumption is made that approach=positive and avoidance=negative, despite a growing body of research questioning the ubiquity of these mappings (e.g., Carver & Harmon-Jones, 2009).

⁵ A fixed effects model on this set of 68 ES yielded a very small, yet significant average ES of 0.041 (SE = 0.01), z = 6.36, p < .001, 95% CI [.028, .054].

⁶ Results of fixed and random effects analyses on the full set of 69 ES (including the outlier) were as follows: Fixed: ES = 0.042 (SE = 0.01), z =6.50, p < .001, 95% CI [.029, .055]; Random: ES = 0.135 (SE = 0.04), z = 3.89, p < .001, 95% CI [.067,.203].