The Size-Weight Illusion is not anti-Bayesian after all: a unifying Bayesian account

PeerJ. 2016 Jun 16:4:e2124. doi: 10.7717/peerj.2124. eCollection 2016.

Abstract

When we lift two differently-sized but equally-weighted objects, we expect the larger to be heavier, but the smaller feels heavier. However, traditional Bayesian approaches with "larger is heavier" priors predict the smaller object should feel lighter; this Size-Weight Illusion (SWI) has thus been labeled "anti-Bayesian" and has stymied psychologists for generations. We propose that previous Bayesian approaches neglect the brain's inference process about density. In our Bayesian model, objects' perceived heaviness relationship is based on both their size and inferred density relationship: observers evaluate competing, categorical hypotheses about objects' relative densities, the inference about which is then used to produce the final estimate of weight. The model can qualitatively and quantitatively reproduce the SWI and explain other researchers' findings, and also makes a novel prediction, which we confirmed. This same computational mechanism accounts for other multisensory phenomena and illusions; that the SWI follows the same process suggests that competitive-prior Bayesian inference can explain human perception across many domains.

Keywords: Bayesian inference; Heaviness perception; Hierarchical causal inference; Size-Weight Illusion.

Grants and funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. MAKP was supported by the National Science Foundation Graduate Research Fellowship Program. LS was supported by National Science Foundation grant BCS-1057625. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.