Frontoparietal Hubs Leverage Probabilistic Representations and Integrated Uncertainty to Guide Cognitive Flexibility.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Neuroimaging

Background:

• Adaptive behavior relies on integrating information from internal cognitive states and external sensory cues.
• Integrating stochastic signals with inherent uncertainty presents a significant challenge for neural systems.
• The neural and computational mechanisms underlying this integration process are not well understood.

Purpose of the Study:

• To elucidate how brain systems integrate internally maintained and externally cued stochastic information.
• To reveal the neural basis of integrating probabilistic information for guiding behavior.
• To introduce a computational neuroimaging framework for analyzing cognitive integration.

Main Methods:

• Collected neuroimaging data from healthy adult human participants.
• Developed a computational model to estimate trial-by-trial beliefs about latent states and perceptual cues.
• Integrated these beliefs into a unified joint probability distribution to quantify uncertainty.

Main Results:

• Latent state beliefs are encoded in the anterior middle frontal gyrus, mediodorsal thalamus, and inferior parietal lobule.
• Perceptual beliefs are encoded in distinct regions, including lateral temporo-occipital areas and the intraparietal sulcus.
• Integrated probability and uncertainty converged in frontoparietal hubs (middle frontal gyrus, intraparietal sulcus).

Conclusions:

• Frontoparietal hub areas integrate distributed information and resolve uncertainty to guide behavior flexibly.
• Distinct brain regions encode internal latent states versus external perceptual cues.
• The findings reveal how frontoparietal systems implement cognitive integration of stochastic information.