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 inference – what 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.

Key words: active inference, free-energy principle, representationalism, enactivism, structural representations

Open peer commentary on the target article “How and Why the Brain Lays the Foundations for a Conscious Self” by Martin V. Butz. Excerpt: Surprisingly, the paper entirely neglects the issue of the dynamic properties of our environment. Focusing on (static, inanimate) objects only, it fails to acknowledge that anticipation becomes especially relevant when things around us change without being under our control: this is when we are forced to adapt quickly to new circumstances. To estimate as precisely as necessary what will when be where is of vital meaning. Surprisingly though, this issue is not addressed at all. Although it is acknowledged that the environment contains dynamics that are to be predicted somehow (§3) and that have to be represented in the brain somehow (§28), concepts about and empirical data on the prediction of external events are not dealt with further.

The concept of the brain as a prediction machine has enjoyed a resurgence in the context of the Bayesian brain and predictive coding approaches within cognitive science. To date, this perspective has been applied primarily to exteroceptive perception (e.g., vision, audition), and action. Here, I describe a predictive, inferential perspective on interoception: ‘interoceptive inference’ conceives of subjective feeling states (emotions) as arising from actively-inferred generative (predictive) models of the causes of interoceptive afferents. The model generalizes ‘appraisal’ theories that view emotions as emerging from cognitive evaluations of physiological changes, and it sheds new light on the neurocognitive mechanisms that underlie the experience of body ownership and conscious selfhood in health and in neuropsychiatric illness.

Key words: interoception, predictive coding, emotion, experience of body ownership, rubber hand illusion, active inference

We examine Dubois’s (2003) distinction between weak anticipation and strong anticipation. Anticipation is weak if it arises from a model of the system via internal simulations. Anticipation is strong if it arises from the system itself via lawful regularities embedded in the system’s ordinary mode of functioning. The assumption of weak anticipation dominates cognitive science and neuroscience and in particular the study of perception and action. The assumption of strong anticipation, however, seems to be required by anticipation’s ubiquity. It is, for example, characteristic of homeostatic processes at the level of the organism, organs, and cells. We develop the formal distinction between strong and weak anticipation by elaboration of anticipating synchronization, a phenomenon arising from time delays in appropriately coupled dynamical systems. The elaboration is conducted in respect to (a) strictly physical systems, (b) the defining features of circadian rhythms, often viewed as paradigmatic of biological behavior based in internal models, (c) Pavlovian learning, and (d) forward models in motor control. We identify the common thread of strongly anticipatory systems and argue for its significance in furthering understanding of notions such as “internal”, “model” and “prediction”.

Key words: anticipation, synchronization, delayed feedback, homeostasis.

The premise of the standard regulatory model, “homeostasis”, is flawed: the goal of regulation is not to preserve constancy of the internal milieu. Rather, it is to continually adjust the milieu to promote survival and reproduction. Regulatory mechanisms need to be efficient, but homeostasis (error-correction by feedback) is inherently inefficient. Thus, although feedbacks are certainly ubiquitous, they could not possibly serve as the primary regulatory mechanism. A newer model, “allostasis”, proposes that efficient regulation requires anticipating needs and preparing to satisfy them before they arise. The advantages: (i) errors are reduced in magnitude and frequency; (ii) response capacities of different components are matched – to prevent bottlenecks and reduce safety factors; (iii) resources are shared between systems to minimize reserve capacities; (iv) errors are remembered and used to reduce future errors. This regulatory strategy requires a dedicated organ, the brain. The brain tracks multitudinous variables and integrates their values with prior knowledge to predict needs and set priorities. The brain coordinates effectors to mobilize resources from modest bodily stores and enforces a system of flexible trade-offs: from each organ according to its ability, to each organ according to its need. The brain also helps regulate the internal milieu by governing anticipatory behavior. Thus, an animal conserves energy by moving to a warmer place – before it cools, and it conserves salt and water by moving to a cooler one before it sweats. The behavioral strategy requires continuously updating a set of specific “shopping lists” that document the growing need for each key component (warmth, food, salt, water). These appetites funnel into a common pathway that employs a “stick” to drive the organism toward filling the need, plus a “carrot” to relax the organism when the need is satisfied. The stick corresponds broadly to the sense of anxiety, and the carrot broadly to the sense of pleasure. This design constrains anxieties to be non-adapting and pleasures to be brief – fast-adapting – to make way for the next anxiety. The stick/carrot mechanisms evolved early and expanded so that in humans they govern higher level learning and social organization. Correspondingly, the “funnel” widened to allow innumerable activities and experiences to each provide non-adapting anxieties and brief pleasures, their reward values depending partly on the effort expended. But modern life narrows the variety of small pleasures and reduces effort, thereby reducing their reward value and requiring larger portions for equivalent satisfaction – a cycle that generates addictive behaviors. Homeostasis and allostasis locate pathology at different levels. Homeostasis identifies proximate causes; for example, it attributes essential hypertension to excess salt water in too small a vascular reservoir. Thus it directs pharmacotherapy toward reducing salt and water, expanding the reservoir, and blocking feedbacks that would counteract these measures. Allostasis attributes essential hypertension to the brain. Chronically anticipating a need for higher pressure, the brain mobilizes all the low level mechanisms in concert: kidney to retain salt and water, vascular system to tighten, and salt appetite to rise. Correspondingly, allostasis would direct therapy toward higher levels – to reduce demand and increase sense of control – so that the brain can down-shift its prediction and relax all the low-level mechanisms in concert. For disorders of addiction homeostasis pursues pharmacological treatments: drugs to treat drug addiction, obesity, and other compulsive behaviors. Allostasis suggests broader approaches – such as re-expanding the range of possible pleasures and providing opportunities to expend effort in their pursuit.

Key words: homeostasis, allostasis, predictive regulation, hypertension, addiction, obesity.

This article explores the possibility that the tendency to believe in an objective, mind-independent external world traces to innate aspects of the human mind. The aspects of mind in question are, first, the capacity to distinguish mental states that have objective referents from those that do not (e.g., perceptual states versus mental imagery), and second, the capacity to mentally represent the continued existence of parts of the world that are beyond the reach of the senses. It is proposed that the evolutionary function of these cognitive abilities relates to the production of novel but adaptive voluntary behaviour. Evidence and arguments are provided in support of the innateness hypothesis. Among these is a Chomskyan-style poverty-of-the-stimulus argument derived from the philosophical literature. The evolutionary account of the subjective-objective distinction leads to the prediction that, in conditions of uncertainty, people will tend to err on the side of assuming the objectivity of their perceptions and other judgements.

Key words: Evolutionary psychology, external world, innate ideas, object permanence, objective-subjective distinction, philosophy.

Open peer commentary on the target article “How and Why the Brain Lays the Foundations for a Conscious Self” by Martin V. Butz. Excerpt: This paper is full of stimulating and creative ideas. It posits that an anticipatory drive is what guides the development in the brain of a set of internal motor models, specifically a set of inverse and forward models. Through these models becoming increasingly complex, a conscious self develops. This is a simple and important thesis, if true. But is it? As my title suggests, it may be so for sportsmen, with their emphasis on ever more refined motor responses. However, those of a more cerebral nature may find themselves burdened by all those coupled internal motor models and not able to think as clearly as they would like. This is not to say that prediction isn’t a useful property to possess, both for finance (especially now) and in one’s general living patterns. But the question I wish to consider is: What sort of predictive model can lead to thinking?

Excerpt: A self-fulfilling prophecy is an assumption or prediction that, purely as a result of having been made, causes the expected or predicted event to occur and thus confirms its own “accuracy.” For example, if someone assumes, for whatever reason, that he is not respected, he will, because of this assumption, act in such a hostile, overly sensitive, sus-picious manner that he brings about that very contempt in others which “proves” again and again his firmly entrenched conviction. This mechanism may be commonplace and well known, but it is based upon a number of facts that are by no means part of our everyday thinking and which have a profound significance for our view of reality.

The article introduces the problematics of the classical two-valued logic on which Western thought is generally based, outlining that under the conditions of its logical assumptions the subject I is situated in a world that it cannot address. In this context, the article outlines a short history of cybernetics and the shift from first- to second-order cybernetics. The basic principles of Gordon Pask’s 1976 Conversation Theory are introduced. It is argued that this second-order theory grants agency to others through a re-conception of living beings as You logically transcending the I. The key principles of Conversation Theory are set in relation to the poetic forms of discourse that played a key role in art as well as philosophical thinking in China in the past. Second-order thinking, the article argues, is essentially poetic. It foregoes prediction in favour of the potentiality of encountering tomorrow’s delights.

Key words: china, conversation theory, aesthetics, art, poetics, second-order cybernetics

Predictive processing and its apparent commitment to explaining cognition in terms of Bayesian inference over hierarchical generative models seems to flatly contradict the pragmatist conception of mind and experience. Against this, I argue that this appearance results from philosophical overlays at odd with the science itself, and that the two frameworks are in fact well-poised for mutually beneficial theoretical exchange. Specifically, I argue: first, that predictive processing illuminates pragmatism’s commitment to both the primacy of pragmatic coping in accounts of the mind and the profound organism-relativity of experience; second, that this pragmatic, “narcissistic” character of prediction error minimization undermines its ability to explain the distinctive normativity of intentionality; and third, that predictive processing therefore mandates an extra-neural account of intentional content of exactly the sort that pragmatism’s communitarian vision of human thought can provide.