Inferring What to Do (And What Not to).

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Behavioral Science

Background:

• The
• planning as inference
• paradigm formalizes motivation as a prior belief about future actions.
• This framework integrates principles from optimal experimental design, information theory, and statistical decision-making.
• Understanding the functional architecture of motivated behavior is crucial.

Purpose of the Study:

• To provide an overview of factors contributing to the "planning as inference" prior.
• To explore how this functional architecture can inform the understanding of neurobiological structures.
• To establish a method for functionally interpreting neurobiological findings through behavioral perturbations.

Main Methods:

• Reviewing theoretical underpinnings of "planning as inference" from optimal experimental design, information theory, and statistical decision-making.
• Drawing parallels between message passing in inferential processes and perturbations of the nervous system (e.g., lesions, optogenetics).
• Analyzing how specific perturbations functionally mimic brain lesions to provide anatomical grounding for theoretical results.

Main Results:

• Identified key factors (optimal design, information theory, decision-making) that shape the "planning as inference" prior.
• Proposed that message passing in inferential processes can be perturbed, analogous to nervous system perturbations.
• Demonstrated how matching behavioral effects of perturbations to brain lesions offers functional interpretations and anatomical grounding.

Conclusions:

• The "planning as inference" paradigm offers a formal approach to understanding motivation and behavior.
• Perturbing inferential message passing provides a novel method to link theoretical models with neurobiological structures.
• This approach yields functional interpretations of empirical findings across goal-directed behavior, active learning, and decision-making.