Cortical hierarchies perform Bayesian causal inference in multisensory perception.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Perception requires integrating signals from common sources and segregating those from independent sources, known as the causal inference problem.
• Humans solve this problem optimally using Bayesian Causal Inference, but the neural basis remains unclear.

Purpose of the Study:

• To investigate the neural mechanisms underlying Bayesian Causal Inference in human multisensory perception.
• To map the hierarchical processing of sensory signals in the brain during spatial localization tasks.

Main Methods:

• Combined psychophysics, Bayesian modeling, functional magnetic resonance imaging (fMRI), and multivariate decoding.
• Utilized an audiovisual spatial localization task to examine sensory signal integration and segregation.

Main Results:

• A hierarchical processing structure was identified in the human brain for Bayesian Causal Inference.
• Auditory and visual areas initially segregate signals (independent sources).
• Posterior intraparietal sulcus fuses signals (common source), while anterior intraparietal sulcus integrates them based on causal uncertainty, aligning with Bayesian Causal Inference.

Conclusions:

• Bayesian Causal Inference in perception is implemented through a hierarchy of multisensory processes in the human neocortex.
• This hierarchical system effectively manages uncertainty about the world's causal structure to combine sensory information.
• The findings elucidate how the brain performs this fundamental computation for perception and cognition.