Active Inference, homeostatic regulation and adaptive behavioural control

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Biology

Background:

• Active Inference offers a probabilistic framework for perception and action.
• Associative learning theories describe Pavlovian, habitual, and goal-directed behaviors.
• These two fields are typically studied independently.

Purpose of the Study:

• To connect Active Inference with associative learning theories.
• To provide a unified framework for homeostatic regulation and adaptive behavior control.
• To generalize Active Inference by emphasizing interoceptive inference.

Main Methods:

• Reviewing the theory of homeostatic regulation and adaptive behavioral control within Active Inference.
• Synthesizing classical associative learning processes (Pavlovian, habitual, goal-directed).
• Casting these processes as hierarchical contextualizations of sensorimotor constructs using generative models.

Main Results:

• A unified mechanistic explanation for classical control and learning optimization processes.
• Demonstration that Active Inference can account for associative learning.
• Generalization of Active Inference to include interoceptive inference and homeostatic regulation.

Conclusions:

• Homeostatic regulation and adaptive behavior can be understood within the Active Inference framework.
• Priors in Active Inference act as drives or goals, explaining action control.
• This perspective emphasizes the embodied nature of inference and its role in maintaining homeostasis.