A meta-analysis of the object-based compatibility effect

Abstract

The object-based compatibility effect (CE) describes, in the context of two-choice keypress tasks, the facilitation of response times (RTs) by the correspondence between participants’ response hand and the task-irrelevant orientation of a viewed object’s handle. Object-based CEs are often attributed to affordance perception. Although the object-based CE paradigm is the major RT task used to study affordances, failures to replicate the effect have raised questions about its robustness. Moreover, it remains unclear whether the object-based CE is indeed indicative of affordances, or whether it is merely an example of spatial CEs brought about by the object’s protruding handle. We present a meta-analysis of object-based CEs to (1) obtain a point estimate of the overall effect and (2) test for moderation consistent with either affordance or spatial compatibility accounts. From 88 independent effects (computed on 2359 participants), we estimated a small but significant compatibility effect (ES = 0.106, z = 5.44, p < .001 95% CI:[0.068, 0.145]), although evidence of publication bias suggests that the true effect is smaller in magnitude. Further, we found significant heterogeneity in effect sizes, indicating between-study variation beyond sampling variability. Moderator analyses indicated that CEs were larger when (1) task-relevant decisions were not about the function of objects, (2) when stimuli were silhouettes as opposed to photographs, and (3) when objects were centered on-screen according to their base or pixel distribution. Response mode (within vs between-hand) did not moderate CEs, nor did the critical interaction between stimulus type (photograph vs silhouette/illustration) and response mode. In all, results are mostly consistent with a spatial compatibility account of object-based CEs. Finally, analyses revealed moderation by trial and task structure, providing implications for study design.