Linguistic priors shape categorical perception

ABSTRACT

This article reviews recent literature on the role of top-down feedback processes in semantic representations in the brain. Empirical studies on perception and theoretical models of semantic cognition show that sensory input is filtered and interpreted based on predictions from higher order cognitive areas. Here, we review the present evidence to the proposal that linguistic constructs, in particular, words, could serve as effective priors, facilitating perception and integration of sensory information. We address a number of theoretical questions arising from this assumption. The focus here is if linguistic categories have a direct top-down effect on early stages of perception; or rather interact with later processing stages such as semantic analysis. We discuss experimental approaches that could discriminate between these possibilities. Taken together, this article provides a review on the interaction between language and perception from the predictive perspective, and suggests avenues to investigate the underlying mechanisms from this perspective.

KEYWORDS:

• Naming
• predictive coding
• top-down effects
• multisensory integration
• language and thought

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Although the existence of trimodal convergence zones is questioned by other researchers, see Man et al. (2013).