Born to act: Deferred action and desire as active inference

Area of Science:

• Computational Neuroscience
• Cognitive Science
• Philosophy of Mind

Background:

• The active inference framework (AIF) models the brain as a generative system minimizing variational free energy via hierarchical Bayesian inference.
• AIF has been criticized for inadequately explaining conative agents motivated by desire, as desire is seen as distinct from belief.

Purpose of the Study:

• To integrate the concept of desire into the active inference framework.
• To propose a novel computational model of desire and deferred action.
• To explore the implications of this model for understanding psychopathology.

Main Methods:

• Defining desire as a hierarchical inference on affective dynamics and action policy precision.
• Linking this hierarchy to coordinated activity in intrinsic brain networks (default mode, action, executive, salient).
• Proposing deferred action as crucial for affect-dependent awareness across timescales.

Main Results:

• Demonstrated that desire can be understood as a form of hierarchical inference within AIF, challenging the belief-desire dichotomy.
• Highlighted the role of deferred action in modulating desire and awareness.
• Framed psychopathology, including OCD and depression, as malfunctions in desire and deferred action processes.

Conclusions:

• Desire is reconcilable with active inference through a hierarchical affective inference model.
• Deferred action is integral to desire and temporal awareness.
• This framework offers new perspectives on understanding and potentially treating affective and motivational disorders.