The visual system's internal model of the world.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Cognitive Science

Background:

• The Bayesian paradigm offers a theoretical framework for perceptual computation.
• Neural mechanisms of Bayesian inference in the brain are under active investigation.
• Recent work highlights computational principles and representational architectures.

Purpose of the Study:

• To analyze representational schemes of internal models in the brain.
• To explore how these models facilitate learning and inference.
• To propose a unified theoretical framework connecting internal models to visual cortex neural phenomena.

Main Methods:

• Conceptual analysis of Bayesian inference in perception.
• Review of computational and neurophysiological studies.
• Theoretical modeling of internal world models.

Main Results:

• The visual system employs a modular hierarchy to build internal world models.
• Perception is achieved through statistical inference using these internal models.
• Analysis of diverse representational schemes for internal models.

Conclusions:

• A unified theoretical framework can relate internal models to neural mechanisms.
• Understanding these models is key to deciphering visual cortex function.
• Bayesian inference provides a powerful lens for perceptual computation.