A tale of two densities: active inference is enactive inference

Abstract

The aim of this article is to clarify how best to interpret some of the central constructs that underwrite the free-energy principle (FEP) -and its corollary, active inference -in theoretical neuroscience and biology: namely, the role that generative models and variational densities play in this theory. We argue that these constructs have been systematically misrepresented in the literature, because of the conflation between the FEP and active inference, on the one hand, and distinct (albeit closely related) Bayesian formulations, centred on the brain -variously known as predictive processing, predictive coding or the prediction error minimisation framework. More specifically, we examine two contrasting interpretations of these models: a structural representationalist interpretation and an enactive interpretation. We argue that the structural representationalist interpretation of generative and recognition models does not do justice to the role that these constructs play in active inference under the FEP. We propose an enactive interpretation of active inferencewhat might be called enactive inference. In active inference under the FEP, the generative and recognition models are best cast as realising inference and control -the self-organising, belief-guided selection of action policies -and do not have the properties ascribed by structural representationalists.

Conscious Agency from Fundamentally Embodied Minds: From Free Energy to Free Will and Back Again

“…However, it may be the case that processes only contribute to consciousness to the degree they couple with dynamic cores on timescales at which information is integrated into global meta-stable states, with coherence potentially being enhanced by mechanisms for stabilizing and coordinating synchronous activity (Buzsáki, 2010;Buzsáki & Watson, 2012). In this way, while minds are certainly extended (Maxwell J. D. Ramstead et al, 2019), consciousness may be a more spatiotemporally limited phenomenon.…”

Is the Free-Energy Principle a Formal Theory of Semantics? From Variational Density Dynamics to Neural and Phenotypic Representations

“…We find approaches in the biological sciences and in the neuroscience less committed to the view of cognition as a representational process taking place within the boundaries of the brain [ 54 , 55 , 56 , 57 , 58 ]. These views include perceptual and motor control theory [ 59 , 60 ]; robotics [ 61 ]; cybernetics [ 62 , 63 , 64 , 65 ]; and, arguably, the free-energy principle and active inference [ 56 , 57 , 66 ]. These accounts, which often hail from embodied and enactive approaches in cognitive science [ 14 , 67 , 68 , 69 , 70 ], converge on the idea that the primary aim of cognition is not internally reconstructing proxies for the structure of a hidden world, but rather to adapt to and act in an environment.…”

“…Several recent papers have discussed whether a realist interpretation of neural representations is warranted under the free-energy principle [ 8 , 32 , 34 , 35 , 56 ]. Besides a few notable exceptions [ 76 ], few papers have sought to evaluate the variety of non-realist arguments in light of the free-energy principle.…”

Entropy, Free Energy, and Symbolization: Free Association at the Intersection of Psychoanalysis and Neuroscience

“…The brain thus constantly responds to interactions with the environment in an enactive way (Ramstead et al, 2019) 7 and these interactions lead to the development of a "generative model" that allows one to make predictions about the environment. The more reliable these predictions are -or the more the brain limits the gap between the internal world and the external world -the lower is the entropy generated in the brain 8 and the fewer are the effects of surprise.…”