Contextual inference in learning and memory.

Area of Science:

• Cognitive Science
• Neuroscience
• Machine Learning

Background:

• Context is crucial for learning and memory across diverse fields like conditioning, episodic memory, and motor skills.
• Current research lacks a unified framework to connect studies on contextual learning.
• A formalized understanding of context's role in learning is needed.

Purpose of the Study:

• To develop a unified vernacular for comparing contextual learning across different domains.
• To propose a Bayesian model for understanding the role of context in learning and memory.
• To formalize the relationship between contextual inference and memory processes.

Main Methods:

• Development of a unified theoretical framework for contextual learning.
• Formulation of a Bayesian model where context is an unobserved variable to be inferred.
• Analysis of how contextual inference impacts memory creation, expression, and updating.

Main Results:

• The proposed model distinguishes between proper learning (memory creation/updating) and apparent learning (expression adjustments).
• Contextual inference is posited as the key mechanism controlling memory dynamics.
• The framework allows for direct comparisons of contextual learning across disparate research areas.

Conclusions:

• A unified Bayesian approach provides a formal framework for contextual learning.
• Inferring context is fundamental to memory formation, expression, and adaptation.
• The model offers a basis for integrating research on learning and memory across different domains.