The human claustrum tracks uncertainty and prediction error

Summary

Single-neuron recordings from the human claustrum reveal that this small, poorly understood structure encodes both uncertainty and prediction errors during negative learning. The response profiles of these neurons are functionally distinct from those in the anterior cingulate cortex.

Why it matters

The claustrum has long been an unsolved mystery in neuroscience. Crick and Koch proposed that this structure coordinates distributed cortical activity. However, its computational role in humans has remained largely inferential. This study is one of the first to provide direct single-neuron evidence in humans. It goes beyond anatomy to examine what claustrum neurons actually compute. Furthermore, it clarifies the division of labor between the claustrum and the anterior cingulate cortex, a central question for predictive coding theories of cognition.

Highlights

• First direct single-neuron evidence in humans showing that claustrum neurons encode uncertainty and prediction errors.
• The claustrum and anterior cingulate cortex exhibit distinct response profiles, suggesting they make separate contributions to belief updating.
• The findings are framed within a Bayesian inference model where uncertainty modulates the weight of prediction errors.

Critical take

This is a Research Summary, not the full article. Consequently, critical methodological details such as sample size, electrode coverage, statistical thresholds, and task parameters are missing. Single-neuron recordings in humans almost certainly involve a very small cohort (likely epilepsy patients with intracranial electrodes), which limits generalizability. This makes it difficult to assess how robust the distinction between the claustrum and the anterior cingulate cortex truly is.