Disentangling predictive processing in the brain: a meta-analytic study in favour of a predictive network

Area of Science:

• Neuroscience
• Cognitive Science

Background:

• Predictive coding (PC) theory posits the brain continuously predicts future states and updates them via error signals.
• Previous research has explored the neural underpinnings of PC, but lacked a systematic, cross-sensory analysis of functional connectivity.
• Defining the neural mechanisms of predictive coding across diverse sensory modalities and studies remains an open challenge.

Purpose of the Study:

• To systematically define the neural mechanisms of predictive coding across studies and sensory channels, with a focus on functional connectivity.
• To investigate the spatial convergence of brain regions involved in prediction error and encoding using meta-analysis.
• To identify the functional network supporting predictive processing and its relationship to attention and execution networks.

Main Methods:

• Coordinate-based meta-analytical approach utilizing the Activation Likelihood Estimation (ALE) algorithm.
• Analysis of spatial convergence across neuroimaging studies related to prediction error and encoding.
• Application of a meta-analytic connectivity method (Seed-Voxel Correlations Consensus) to reveal functional networks.

Main Results:

• ALE results indicate significant roles for the left inferior frontal gyrus and left insula in both prediction error and encoding.
• Meta-analytic connectivity revealed a large, bilateral predictive network.
• This network shows resemblance to large-scale networks involved in task-driven attention and execution.

Conclusions:

• Predictive processing is not uniformly distributed across the brain, with specific regions showing higher involvement across sensory modalities.
• At the network level, there is no discernible distinction between the processing of prediction errors and the processing of predictions themselves.
• The findings suggest a unified network supporting predictive coding, integrating prediction and error-related computations.