Empirically Integrating the Evidence for Different Predictive Coding Components Using Auditory False Perception.

Area of Science:

• Neuroscience
• Cognitive Science
• Psychology

Background:

• Perception involves probabilistic estimation influenced by the brain's internal model.
• Predictive coding models include internal models, priors, prediction error (PE), and perceptual inference.
• Previous research examined components of predictive coding independently.

Purpose of the Study:

• To integrate and investigate components of the predictive coding system.
• To model conditioned hallucination (CH) illusions using a multisensory integration paradigm.
• To compare behavioral and electrophysiological responses in high and low illusion perceivers.

Main Methods:

• Induction of conditioned hallucination (CH) illusions via multisensory integration.
• Utilizing a local-global auditory oddball paradigm to probe predictive coding.
• Comparing behavioral and electrophysiological data between high and low perceivers.

Main Results:

• High perceivers exhibit greater confidence in their internal models compared to low perceivers, who rely more on sensory information.
• High perceivers demonstrate increased sensitivity to prediction errors indicating internal model changes.
• Objective and self-reported illusion likelihoods correlate, but this association diminishes when controlling for perceptual thresholds.

Conclusions:

• Individual differences in perception are linked to biases towards internal models versus sensory information.
• Sensitivity to prediction errors reflects the malleability of the internal model.
• Understanding innate biases is crucial for explaining variations in sensory perception.