Predictive coding: a more cognitive process than we thought?

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Predictive coding (PC) theory posits that higher-order brain regions generate predictions sent to lower-order sensory areas.
• Mismatches between top-down predictions and bottom-up sensory data generate prediction errors, crucial for learning and perception.
• Traditionally, prediction errors in PC are thought to be encoded in sensory cortex pyramidal neurons.

Purpose of the Study:

• To investigate the precise neural locus of prediction error generation.
• To evaluate the validity of the predictive coding (PC) model versus an alternative predictive routing (PR) framework.
• To determine if prediction errors are primarily sensory or cognitive in origin.

Main Methods:

• Utilized neuronal spiking data analysis.
• Employed the global-local oddball paradigm across multiple recording modalities.
• Examined neural activity in sensory and prefrontal cortex (PFC) areas.

Main Results:

• Neuroimaging studies consistently show prediction error responses in both sensory and higher-order areas, aligning with PC.
• Recent neuronal spiking studies suggest that genuine prediction errors predominantly emerge in the prefrontal cortex (PFC).
• This finding challenges the sensory-centric view of PC.

Conclusions:

• The localization of prediction errors in the PFC suggests predictive processing is more cognitive than sensory-based.
• This challenges core assumptions of the predictive coding (PC) model.
• Findings better support an alternative framework termed predictive routing (PR).