The Bayesian-Laplacian brain.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Bayesian Brain Hypothesis

Background:

• The brain's inferential systems are often modeled as Bayesian-Laplacian systems.
• Understanding the nature of priors used in these systems is crucial for refining these models.

Purpose of the Study:

• To propose an improved framework for discussing the brain as a Bayesian-Laplacian system.
• To distinguish between two fundamental classes of priors: biological (β) and artefactual (α).

Main Methods:

• Conceptual analysis distinguishing between biological (β) and artefactual (α) priors.
• Examination of the characteristics and acquisition of each prior type.
• Analysis of the impact of each prior type on the universality of posterior beliefs.

Main Results:

• Biological priors (e.g., color, faces) are innate/rapidly acquired, resistant to change, and universal, leading to high agreement on posteriors.
• Artefactual priors (e.g., man-made objects) are learned post-natally, adaptable, and lead to less universal agreement.
• A spectrum exists where β and α priors merge, influencing posteriors based on both categories.

Conclusions:

• Distinguishing between β and α priors offers a more nuanced understanding of the Bayesian brain.
• The inherent stability and universality of β priors explain widespread agreement on certain percepts.
• The adaptability of α priors supports individual learning and cultural variation.