Different Ways of Linking Behavioral and Neural Data via Computational Cognitive Models

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Neuroscience

Background:

• Formal models are used in cognitive neuroscience to understand brain mechanisms underlying cognitive processes.
• These models use latent variables to explain behavioral data, aiming to connect them with brain measurements.
• Linking cognitive models to neural data is crucial but lacks a standardized philosophical approach.

Purpose of the Study:

• To propose a framework for linking formal cognitive models with neural data.
• To introduce a continuum of linking approaches based on explicitness and quantification.
• To illustrate these approaches with examples from decision making, reinforcement learning, and symbolic reasoning.

Main Methods:

• Conceptual framework development.
• Categorization of linking approaches into four types: qualitative structural, qualitative predictive, quantitative predictive, and single model linking.
• Illustrative examples from diverse research domains.

Main Results:

• A continuum of four distinct approaches for linking cognitive models to brain data is defined.
• These approaches vary in their degree of quantitative and explicit hypothesizing.
• Examples demonstrate the application of different linking strategies across various cognitive domains.

Conclusions:

• The proposed continuum offers a structured way to approach the model-brain linking problem in cognitive neuroscience.
• Understanding these different linking strategies can advance both cognitive theory and neuroscientific investigation.
• This framework facilitates clearer communication and methodological choices in the field.