method
active
method:gradient-descent-on-variational-free-energyGradient Descent on Variational Free Energy
Process by which neuronal dynamics minimize free energy; produces empirically observable neural phenomena.
Neighborhood — ranked by edge-count
Frameworks (1)
framework
- Active InferenceimplementsFoundational framework by Karl Friston; the paper extends it to three hierarchical levels for modeling meta-awareness.
Findings (8)
finding
- Mismatch NegativitysupportsERP component reproduced by active inference: neural response to prediction violations.
- Theta-Gamma CouplingsupportsHippocampal oscillatory phenomenon reproduced by active inference; phase-amplitude coupling.
- Evidence AccumulationsupportsDecision-making phenomenon reproduced by active inference in parietal/prefrontal cortex.
- Phase PrecessionsupportsHippocampal neural coding phenomenon reproduced by active inference.
- Place Cell ActivitysupportsHippocampal phenomenon reproduced by active inference model.
- Race-to-Bound DynamicssupportsDecision-making neural dynamics reproduced by active inference; threshold crossing.
- Repetition SuppressionsupportsNeural phenomenon reproduced by active inference model: reduced response to repeated stimuli.
- Theta SequencessupportsHippocampal sequential activity pattern reproduced by active inference.
Concepts (3)
concept
- free energyimplementsThermodynamic potential ΔF = ΔE − TΔS; domain walls form if ΔF < 0
- Hamilton's Principle of Least Actionassociated_withNeuronal dynamics conform to Hamilton's principle via free energy minimization; connects to physics.
- Prediction Errorassociated_withRole in optimizing sensory states; unified treatment shows value-learning and perception share error-minimization principle.
Methods (1)
method
- Gradient Descent on Free Energyrelated_toOptimization procedure for simultaneously updating action selection and perception; uses step size ζ (default 4).
Hypotheses (1)
hypothesis
- Paper's core methodological hypothesis: gap between normative and process-level theories can be bridged.
Claims (1)
claim
- Fundamental assertion: single imperative (free energy minimization) explains diverse cognitive and neural phenomena.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Used for updating hidden state expectations; provides dynamical process theory testable against neuronal data
- Central claim: gradient descent on free energy is a valid process-level description of neural activity.
- Central research question: whether process-level neural dynamics conform to free energy minimization.
- Minimizing variational free energy for perceptual inference and learning of model parameters.
- Key theoretical result: gradient descent formulation validates free energy as fundamental principle.
- The dynamics of synaptic plasticity follow a descent on the gradient of variational free energy.claim0.783Learning as free energy gradient descent, Section 8.
- Upper bound on surprisal minimised by any persisting agent; decomposes into noise and insufficient learning in the qFEP
- Optimization technique that computes weight changes by following the gradient of an error function; contrasted with evolutionary stochastic search.