The neural dynamics of hierarchical Bayesian causal inference in multisensory perception

Tim Rohe,Uta Noppeney,Ann-Christine Ehlis

doi:10.1038/s41467-019-09664-2

Tim Rohe, Uta Noppeney + Show 1 more

Open Access

https://doi.org/10.1038/s41467-019-09664-2

Copy DOI

Abstract

Transforming the barrage of sensory signals into a coherent multisensory percept relies on solving the binding problem – deciding whether signals come from a common cause and should be integrated or, instead, segregated. Human observers typically arbitrate between integration and segregation consistent with Bayesian Causal Inference, but the neural mechanisms remain poorly understood. Here, we presented people with audiovisual sequences that varied in the number of flashes and beeps, then combined Bayesian modelling and EEG representational similarity analyses. Our data suggest that the brain initially represents the number of flashes and beeps independently. Later, it computes their numbers by averaging the forced-fusion and segregation estimates weighted by the probabilities of common and independent cause models (i.e. model averaging). Crucially, prestimulus oscillatory alpha power and phase correlate with observers’ prior beliefs about the world’s causal structure that guide their arbitration between sensory integration and segregation.

Highlights

Transforming the barrage of sensory signals into a coherent multisensory percept relies on solving the binding problem – deciding whether signals come from a common cause and should be integrated or, instead, segregated
To form a coherent percept of the world, the human brain needs to integrate signals arising from a common cause, but segregate signals from independent causes
To characterize the neural dynamics of how the brain solves this binding problem, we presented participants with sequences of beeps and flashes that varied in their numeric disparity

Summary

Introduction

Transforming the barrage of sensory signals into a coherent multisensory percept relies on solving the binding problem – deciding whether signals come from a common cause and should be integrated or, instead, segregated. Causal Inference provides a rational strategy to arbitrate between information integration and segregation by explicitly modelling the underlying potential causal structures, i.e. whether visual impacts and splash sounds are caused by common or independent events[1,2]. Some splash sounds are caused by a stone hitting the water surface out of the observer’s sight (e.g. another person throwing a stone), audition and vision will provide conflicting information In this segregation case, the brain needs to estimate the number of events independently for vision and audition. Accumulating evidence suggests that human and animal observers arbitrate between sensory integration and segregation approximately in line with Bayesian Causal Inference[1,5,6,7]. For large intersensory conflicts such as temporal asynchrony, spatial disparity or numeric disparity, multisensory integration breaks down and crossmodal biases are attenuated[5,9]

Methods

Results

Conclusion