The free energy principle is notoriously difficult to understand. In this paper, we relate the principle to a framework that philosophers of biology are familiar with: Ruth Millikan’s teleosemantics. We argue that: systems that minimise free energy are systems with a proper function; and Karl Friston’s notion of implicit modelling can be understood in terms of Millikan’s notion of mapping relations. Our analysis reveals some surprising formal similarities between the two frameworks, and suggests interesting lines of future research. We hope (...) this will aid further philosophical evaluation of the free energy principle. (shrink)

Over the last fifteen years, an ambitious explanatory framework has been proposed to unify explanations across biology and cognitive science. Active inference, whose most famous tenet is the free energy principle, has inspired excitement and confusion in equal measure. Here, we lay the ground for proper critical analysis of active inference, in three ways. First, we give simplified versions of its core mathematical models. Second, we outline the historical development of active inference and its relationship to other theoretical approaches. Third, (...) we describe three different kinds of claim -- labelled mathematical, empirical and general -- routinely made by proponents of the framework, and suggest dialectical links between them. Overall, we aim to increase philosophical understanding of active inference so that it may be more readily evaluated. -/- This the final submitted version of the Introduction to the Topical Collection "The Free Energy Principle: From Biology to Cognition", forthcoming in Biology & Philosophy. (shrink)

La philosophie économique, dont le développement implique un accroissement accéléré de complexité (ce qui est la définition même de la complexité). Les prémices en ont été le second théorème sur l’équilibre général, et les théorèmes d’impossibilité sur le choix collectif, ou le dilemme des biens publics. Certes, il est possible d’utiliser certains facteurs de complexité, comme des mutations aléatoires, pour définir par exemple des équilibres évolutionnaires, mais ces facteurs sont aussi une source d’instabilité. L’économie expérimentale a mis en lumière la (...) dépendance de nos choix par rapport à leurs évolutions et références antérieures, ce qui est une autre source de complexité. Une exploration rationnelle de ces dépendances met elle-même en jeu la complexité de leurs processus, et implique d’incessantes révisions. Il semble inévitable, puisqu’on ne peut pas réduire les complexités qui tiennent à ces réajustements, qui impliquent un bayésianisme multi-niveaux généralisé, d’admettre que ces révisions n’aient qu’une portée limitée (« minimiser l’énergie libre »). On peut réinterpréter en ce sens les expériences sur des incohérences temporelles dans nos choix, une fois pris en compte l’application de cette minimisation aux processus mêmes d’évaluation des situations de choix. D’autres limitations pourraient être définies en combinant les contraintes neurologiques, psychologiques et économiques. Là encore, une telle combinaison est aussi une source de complexités. Mais dans un autre sens, développer ces relations complexes peut permettre de dépasser les circularités de l’inférence inverse (du neurologique à l’économie expérimentale et inversement). Enfin les simulations computationnelles nous révèlent que des évolutions que l’on croyait asymptotiques sont capables de plus lointaines divergences. Compliquer les modèles computationnels peut limiter leur généralité. Inversement les simplifier n’est utile que si cela révèle aussi quelles simplifications nous font manquer quelle complexité des interactions économiques. L’exploration de ces relations entre cette croissance accélérée de complexité liée non seulement à ces interactions mais à l’entreprise même de leur rationalisation est d’abord la tâche de l’économie dans ses différents champs, mais la philosophie de l’économie peut avoir un rôle dans l’élucidation des rapports entre ces différentes sources de complexité et leurs modalités de limitation les plus raisonnablesCode JEL : B490. (shrink)

Predictive processing (PP) has been repeatedly presented as a unificatory account of perception, action, and cognition. In this paper, we argue that this is premature: As a unifying theory, PP fails to deliver general, simple, homogeneous, and systematic explanations. By examining its current trajectory of development, we conclude that PP remains only loosely connected both to its computational framework and to its hypothetical biological underpinnings, which makes its fundamentals unclear. Instead of offering explanations that refer to the same set of (...) principles, we observe systematic equivocations in PP‐based models, or outright contradictions with its avowed principles. To make matters worse, PP‐based models are seldom empirically validated, and they are frequently offered as mere just‐so stories. The large number of PP‐based models is thus not evidence of theoretical progress in unifying perception, action, and cognition. On the contrary, we maintain that the gap between theory and its biological and computational bases contributes to the arrested development of PP as a unificatory theory. Thus, we urge the defenders of PP to focus on its critical problems instead of offering mere re‐descriptions of known phenomena, and to validate their models against possible alternative explanations that stem from different theoretical assumptions. Otherwise, PP will ultimately fail as a unified theory of cognition. (shrink)

Although research on the mismatch negativity (MMN) has been ongoing for 40 years, the generation process of the MMN remains largely unknown. In this study, we used a single-trial EEG-fMRI coupling method which can analyse neural activity with both high temporal and high spatial resolution and thus assess the generation process of the MMN. We elicited the MMN with an auditory oddball paradigm while recording simultaneous EEG and fMRI. We divided the MMN into five equal-durational phases. Utilizing the single-trial variability (...) of the MMN, we analysed the neural generators of the five phases, thereby determining the spatiotemporal generation process of the MMN. We found two distinct bottom-up prediction error propagations: first from the auditory cortex to the motor areas and then from the auditory cortex to the inferior frontal gyrus (IFG). Our results support the regularity-violation hypothesis of MMN generation. (shrink)

Current diagnostic criteria for schizophrenia place emphasis on delusions and hallucinations, whereas the classical descriptions of schizophrenia by Kraepelin and Bleuler emphasized disorganization and impoverishment of mental activity. Despite the availability of antipsychotic medication for treating delusions and hallucinations, many patients continue to experience persisting disability. Improving treatment requires a better understanding of the processes leading to persisting disability. We recently introduced the term classical schizophrenia to describe cases with disorganized and impoverished mental activity, cognitive impairment and predisposition to persisting (...) disability. Recent evidence reveals that a polygenic score indicating risk for schizophrenia predicts severity of the features of classical schizophrenia: disorganization, and to a lesser extent, impoverishment of mental activity and cognitive impairment. Current understanding of brain function attributes a cardinal role to predictive coding: the process of generating models of the world that are successively updated in light of confirmation or contradiction by subsequent sensory information. It has been proposed that abnormalities of these predictive processes account for delusions and hallucinations. Here we examine the evidence provided by electrophysiology and fMRI indicating that imprecise predictive coding is the core pathological process in classical schizophrenia, accounting for disorganization, psychomotor poverty and cognitive impairment. Functional imaging reveals aberrant brain activity at network hubs engaged during encoding of predictions. We discuss the possibility that frequent prediction errors might promote excess release of the neurotransmitter, dopamine, thereby accounting for the occurrence of episodes of florid psychotic symptoms including delusions and hallucinations in classical schizophrenia. While the predictive coding hypotheses partially accounts for the time-course of classical schizophrenia, the overall body of evidence indicates that environmental factors also contribute. We discuss the evidence that chronic inflammation is a mechanism that might link diverse genetic and environmental etiological factors, and contribute to the proposed imprecision of predictive coding. (shrink)

Recent computational models of perception conceptualize auditory oddball responses as signatures of a learning process, in line with the influential view of the mismatch negativity as a prediction error signal. Novel MMN experimental paradigms have put an emphasis on neurophysiological effects of manipulating regularity and predictability in sound sequences. This raises the question of the contextual adaptation of the learning process itself, which on the computational side speaks to the mechanisms of gain-modulated prediction error. In this study using electrocorticographic signals, (...) we manipulated the predictability of oddball sound sequences with two objectives: Uncovering the computational process underlying trial-by-trial variations of the cortical responses. The fluctuations between trials, generally ignored by approaches based on averaged evoked responses, should reflect the learning involved. We used a general linear model and Bayesian Model Reduction to assess the respective contributions of experimental manipulations and learning mechanisms under probabilistic assumptions. To validate and expand on previous findings regarding the effect of changes in predictability using simultaneous EEG-MEG recordings. Our trial-by-trial analysis revealed only a few stimulus-responsive sensors but the measured effects appear to be consistent over subjects in both time and space. In time, they occur at the typical latency of the MMN. In space, we found a dissociation between time-independent effects in more anterior temporal locations and time-dependent effects in more posterior locations. However, we could not observe any clear and reliable effect of our manipulation of predictability modulation onto the above learning process. Overall, these findings clearly demonstrate the potential of trial-to-trial modeling to unravel perceptual learning processes and their neurophysiological counterparts. (shrink)

What principles and mechanisms allow humans to encode complex 3D information, and how can it be so fast, so accurately and so flexibly transformed into coordinated action? How do these processes work when developed to the limit of human physiological and cognitive capacity—as they are in high-speed sports, such as alpine skiing or motor racing? High-speed sports present not only physical challenges, but present some of the biggest perceptual-cognitive demands for the brain. The skill of these elite athletes is in (...) many ways an attractive model for studying human performance “in the wild”, and its neurocognitive basis. This article presents a framework theory for how these abilities may be realized in high-speed sports. It draws on a careful analysis of the case of the motorsport athlete, as well as theoretical concepts from: (1) cognitive neuroscience of wayfinding, steering, and driving; (2) cognitive psychology of expertise; (3) cognitive modeling and machine learning; (4) human-in-the loop modellling in vehicle system dynamics and human performance engineering; (5) experimental research (in the laboratory and in the field) on human visual guidance. The distinctive contribution is the way these are integrated, and the concept ofchunkingis used in a novel way to analyze a high-speed sport. The mechanisms invoked are domain-general, and not specific to motorsport or the use of a particular type of vehicle (or any vehicle for that matter); theegocentric chunking hypothesisshould therefore apply to any dynamic task that requires similar core skills. It offers a framework for neuroscientists, psychologists, engineers, and computer scientists working in the field of expert sports performance, and may be useful in translating fundamental research into theory-based insight and recommendations for improving real-world elite performance. Specific experimental predictions and applicability of the hypotheses to other sports are discussed. (shrink)

Accounts of mental disorders focusing either on the brain as neurophysiological substrate or on systematic connections between symptoms are insufficient to account for the multifactorial nature of mental illnesses. Recently, multiplexes have been suggested to provide a holistic view of psychopathology that integrates data from different factors, at different scales, or across time. Intuitively, these multi-layered network structures present quite appealing models of mental disorders that can be constructed by powerful computational machinery based on increasing amounts of real-world data. In (...) this paper, I systematically examine what challenges psychopathology models face and to what extent different species of psychopathology models can address them. My analysis highlights that while multiplexes, as they are usually conceived, appear promising, they suffer from the same problems as other approaches. To remedy this, I suggest, we must go a step further and combine different kinds of multiplexes into 4D models. Once we embrace 4D multiplexes and identify appropriate ways to constrain them, we might unlock the true potential of multiplexes for making headway in psychopathology research. (shrink)

According to the free energy principle biological agents resist a tendency to disorder in their interactions with a dynamically changing environment by keeping themselves in sensory and physiological states that are expected given their embodiment and the niche they inhabit :127–138, 2010. doi: 10.1038/nrn2787). Why would a biological agent that aims at minimising uncertainty in its encounters with the world ever be motivated to seek out novelty? Novelty for such an agent would arrive in the form of sensory and physiological (...) states that are unexpected. Such an agent ought therefore to avoid novel and surprising interactions with the world one might think. Yet humans and many other animals find play and other forms of novelty-seeking and exploration hugely rewarding. How can this be understood in frameworks for studying the mind that emphasise prediction error minimisation? This problem has been taken up in recent research concerned with epistemic action—actions an agent engages in to reduce uncertainty. However that work leaves two questions unanswered, which it is the aim of our paper to address. First, no account has been given yet of why it should feel good to the agent to engage the world playfully and with curiosity. Second an appeal is made to precision-estimation to explain epistemic action, yet it remains unclear how precision-weighting works in action more generally, or active inference. We argue that an answer to both questions may lie in the bodily states of an agent that track the rate at which free energy is being reduced. The recent literature on the predictive brain has connected the valence of emotional experiences to the rate of change in the reduction of prediction error :e1003094, 2013. doi: 10.1371/journal.pcbi.1003094; Van de Cruys, in Metzinger and Wiese Philosophy and predictive processing, vol 24, MIND Group, Frankfurt am Main, 2017. doi: 10.15502/9783958573253). In this literature valenced emotional experiences are hypothesised to be identical with changes in the rate at which prediction error is reduced. Experiences are negatively valenced when overall prediction error increases and are positively valenced when the sum of prediction errors decrease. We offer an ecological-enactive interpretation of the concept of valence and its connection to rate of change of prediction error. We show how rate of change should be understood in terms of embodied states of affordance-related action readiness. We then go on to apply this ecological-enactive account of error dynamics to provide an answer to the first question we have raised: It may explain why it should feel good to an agent to be curious and playful. Our ecological-enactive account also allows us to show how error dynamics may provide an answer to the second question we have raised regarding how precision-weighting works in active inference. An agent that is sensitive to rates of error reduction can tune precision on the fly. We show how this ability to tune precision on the go can allow agents to develop skills for adapting better and better to the unexpected, and search out opportunities for resolving uncertainty and progressing in its learning. (shrink)

In addition to active wake, emotions are generated and experienced in a variety of functionally different states such as those of sleep, during which external stimulation and cognitive control are lacking. The neural basis of emotions can be specified by regarding the multitude of emotion-related brain states, as well as the distinct neuro- and psychodynamic stages (generation and regulation) of emotional experience.

This paper examines, for the first time, the relationship between realization relations and the free energy principle in cognitive neuroscience. I argue, firstly, that the free energy principle has ramifications for the wide versus narrow realization distinction: if the free energy principle is correct, then organismic realizers are insufficient for realizing free energy minimization. I argue, secondly, that the free energy principle has implications for synchronic realization relations, because free energy minimization is realized in dynamical agent-environment couplings embedded at multiple (...) time scales. (shrink)

This paper examines the relationship between perceiving and imagining on the basis of predictive processing models in neuroscience. Contrary to the received view in philosophy of mind, which holds that perceiving and imagining are essentially distinct, these models depict perceiving and imagining as deeply unified and overlapping. It is argued that there are two mutually exclusive implications of taking perception and imagination to be fundamentally unified. The view defended is what I dub the ecological–enactive view given that it does not (...) succumb to internalism about the mind-world relation, and allows one to keep a version of the received view in play. (shrink)

Most philosophical accounts of emergence are based on supervenience, with supervenience being an ontologically synchronic relation of determination. This conception of emergence as a relation of supervenience, I will argue, is unable to make sense of the kinds of emergence that are widespread in self-organizing and nonlinear dynamical systems, including distributed cognitive systems. In these dynamical systems, an emergent property is ontological and diachronic.

This paper starts by considering an argument for thinking that predictive processing (PP) is representational. This argument suggests that the Kullback–Leibler (KL)-divergence provides an accessible measure of misrepresentation, and therefore, a measure of representational content in hierarchical Bayesian inference. The paper then argues that while the KL-divergence is a measure of information, it does not establish a sufficient measure of representational content. We argue that this follows from the fact that the KL-divergence is a measure of relative entropy, which can (...) be shown to be the same as covariance (through a set of additional steps). It is well known that facts about covariance do not entail facts about representational content. So there is no reason to think that the KL-divergence is a measure of (mis-)representational content. This paper thus provides an enactive, non-representational account of Bayesian belief optimisation in hierarchical PP. (shrink)

Degeneracy is ubiquitous across biological systems where structurally different elements can yield a similar outcome. Degeneracy is of particular interest in neuroscience too. On the one hand, degeneracy confers robustness to the nervous system and facilitates evolvability: Different elements provide a backup plan for the system in response to any perturbation or disturbance. On the other, a difficulty in the treatment of some neurological disorders such as chronic pain is explained in light of different elements all of which contribute to (...) the pathological behavior of the system. Under these circumstances, targeting a specific element is ineffective because other elements can compensate for this modulation. Understanding degeneracy in the physiological context explains its beneficial role in the robustness of neural circuits. Likewise, understanding degeneracy in the pathological context opens new avenues of discovery to find more effective therapies. (shrink)

The term attention has been used to mean so many different things that some have despaired of it being useful at all. This paper is devoted to bringing a modicum of order to the chaos through the time-honored device of categorization. The chief purpose of this paper is to introduce a comprehensive descriptive taxonomy of the nuanced ways the term attention may be employed. It is presented in table form, followed by elucidations and illustrations of each of its items. But (...) first, I offer reasons why a taxonomy of attention is needed and explore some of its possible benefits. After presenting the taxonomy, I sketch by way of example how it might be applied to two interesting questions: is the umbrella term attention still useful?; and, what is it that ties the subdefinitions of attention together yet distinguishes them from other kinds of cognition? (shrink)

This article is a comparative study between predictive processing (PP, or predictive coding) and cognitive dissonance (CD) theory. The theory of CD, one of the most influential and extensively studied theories in social psychology, is shown to be highly compatible with recent developments in PP. This is particularly evident in the notion that both theories deal with strategies to reduce perceived error signals. However, reasons exist to update the theory of CD to one of “predictive dissonance.” First, the hierarchical PP (...) framework can be helpful in understanding varying nested levels of CD. If dissonance arises from a cascade of downstream and lateral predictions and consequent prediction errors, dissonance can exist at a multitude of scales, all the way up from sensory perception to higher order cognitions. This helps understand the previously problematic dichotomy between “dissonant cognitive relations” and “dissonant psychological states,” which are part of the same perception-action process while still hierarchically distinct. Second, since PP is action-oriented, it can be read to support recent action-based models of CD. Third, PP can potentially help us understand the recently speculated evolutionary origins of CD. Here, the argument is that responses to CD can instill meta-learning which serves to prevent the overfitting of generative models to ephemeral local conditions. This can increase action-oriented ecological rationality and enhanced capabilities to interact with a rich landscape of affordances. The downside is that in today’s world where social institutions such as science a priori separate noise from signal, some reactions to predictive dissonance might propagate ecologically unsound (underfitted, confirmation-biased) mental models such as climate denialism. (shrink)

Baggs and Chemero (2018) propose that certain tensions between enaction and ecological psychology arise due different interpretations about what is meant by the “environment.” In the enactive approach the emphasis is on the umwelt, which describes the environment as the “meaningful, lived surroundings of a given individual.” The ecological approach, on the other hand, emphasises what they refer to as the habitat “the environment as a set of resources for a typical, or ideal, member of a species.” By making this (...) distinction, these authors claim they are able to retain the best of both the ecological and the enactive approaches. Herein I propose an account of the individuation of habits that straddles this distinction, what I call a compatabilist account. This is done in two parts. The first part teases out a host of compatibilities that exist between the enactive account as developed by Di Paolo et al. (2017) and the skilled intentionality framework as developed by Bruineberg and Rietveld (2014) and Rietveld and Kiverstein (2014). In part two these compatibilities are brought together with the philosophy of Gilbert Simondon to develop the notion of enhabiting. Enhabiting describes a set of ongoing processes by which an umwelt emerges from and is reproduced within the relationship between an embodied subject and their habitat. Thus, enhabiting points toward a point of intersection between enaction and ecological psychology. To enhabit is bring forth (to enact), within (to inhabit). (shrink)

The neural correlates of rapid eye movements (REMs) in sleep are extraordinarily robust; including REM-locked activation in the retrosplenial cortex, the supplementary eye field and areas overlapping cholinergic basal nucleus. The phenomenology of REMs speaks to the notion that perceptual experience in both sleep and wakefulness is a constructive process – in which we generate predictions of sensory inputs and then test those predictions through actively sampling the sensorium with eye movements. On this view, REMs during sleep may index an (...) internalized active sampling or ‘scanning’ of self-generated visual constructs that are not constrained by visual input. If this view is correct, it renders REMs an ideal probe to study consciousness as “an exclusively internal affair” (Metzinger 2009). In other words, REMs offer a probe of active inference – in the sense of predictive coding – when the brain is isolated from the sensorium in virtue of the natural blockade of sensory input during REM sleep. REMs are temporally precise events that enable powerful inferences based on timeseries analyses. As a natural, task-free probe, [REMs] could be used in non-compliant subjects, including infants and animals. In short, REM constitutes a promising probe to study the ontogenetic and phylogenetic development of consciousness and perhaps its abnormal development in schizophrenia and autism, which have been considered in terms of aberrant predictive coding. (shrink)

The free energy principle says that any self-organising system that is at nonequilibrium steady-state with its environment must minimize its free energy. It is proposed as a grand unifying principle for cognitive science and biology. The principle can appear cryptic, esoteric, too ambitious, and unfalsifiable—suggesting it would be best to suspend any belief in the principle, and instead focus on individual, more concrete and falsifiable ‘process theories’ for particular biological processes and phenomena like perception, decision and action. Here, I explain (...) the free energy principle, and I argue that it is best understood as offering a conceptual and mathematical analysis of the concept of existence of self-organising systems. This analysis offers a new type of solution to long-standing problems in neurobiology, cognitive science, machine learning and philosophy concerning the possibility of normatively constrained, self-supervised learning and inference. The principle can therefore uniquely serve as a regulatory principle for process theories, to ensure that process theories conforming to it enable self-supervision. This is, at least for those who believe self-supervision is a foundational explanatory task, good reason to believe the free energy principle. (shrink)

Bortolotti’s Delusions and Other Irrational Beliefs defends the view that delusions are beliefs on a continuum with other beliefs. A different view is that delusions are more like illusions, that is, they arise from faulty perception. This view, which is not targeted by the book, makes it easier to explain why delusions are so alien and disabling but needs to appeal to forensic aspects of functioning.

This paper argues that it is possible to embrace the predictive processing framework (PP) without reducing affordances to inferential perception. The cognitivist account of PP contends that it can capture relational perception, such as affordances. The rationale for this claim is that over time, sensory data becomes highly-weighted. This paper, however, will show the inconsistency of this claim in the face of the cognitivist premise that ‘encapsulated’ models can throw away ‘the body, the world, or other people’ [Hohwy 2016: 265]. (...) It is then showed how it is possible to embrace a non-cognitivist reading of PP—one that does not need to reduce affordances to representational content. (shrink)

How do we define religious experiences? And what would be the relationship with spiritual experiences? The author claims that the cognitive turn in science gives us new opportunities to map the territory of religion and spirituality. In line with other authors, he proposes a building block approach of cognitive mechanisms that can deal with questions regarding the specificity, origin, and complexity of religious experiences. Two concepts are presented that bridge the great divide which is presumed to exist between sciences that (...) study the brain and humanities, namely the encultured brain and predictive minds. In section three, six building blocks of the structure of religious experience are formulated. New in his approach is the unexpected possible as distinctive ground between normal experiences and what we consider spiritual and religious experiences. Finally, the author presents a critical reflection on his proposal and challenges for the road ahead. (shrink)

In this review, we deal with two central questions of consciousness how and why, and we outline their possible future development. The question how refers to the empirical endeavor to reveal the neural correlates and mechanisms that form consciousness. On the other hand, the question why generally refers to the “hard problem” of consciousness, which claims that empirical science will always fail to provide a satisfactory answer to the question why is there conscious experience at all. Unfortunately, the hard problem (...) of consciousness will probably never completely disappear because it will always have its most committed supporters. However, there is a good chance that its weight and importance will be highly reduced by empirically tackling consciousness in the near future. We expect that future empirical endeavor of consciousness will be based on a unifying brain theory and will answer the question as to what is the function of conscious experience, which will in turn replace the implications of the hard problem. The candidate of such a unifying brain theory is predictive coding, which will have to explain both perceptual consciousness and conscious mind-wandering in order to become the truly unifying theory of brain functioning. (shrink)

People who habituate to stress show a repetition-induced response attenuation—neuroendocrine, cardiovascular, neuroenergetic, and emotional—when exposed to a threatening environment. But the exact dynamics underlying stress habituation remain obscure. The free energy principle offers a unifying account of self-organising systems such as the human brain. In this paper, we elaborate on how stress habituation can be explained and modelled using the free energy principle. We introduce habituation priors that encode the agent’s tendency for stress habituation and incorporate them in the agent’s (...) decision-making process. Using differently shaped goal priors—that encode the agent’s goal preferences—we illustrate, in two examples, the optimising behaviour of agents. We show that habituation minimises free energy by reducing the precision of goal preferences. Reducing the precision of goal priors means that the agent accepts adverse states. Acceptance or tolerance of adverse outcomes may explain why habituation causes people to exhibit an attenuation of the stress response. Given that stress habituation occurs in brain regions where goal priors are encoded, i.e., in the ventromedial prefrontal cortex and that these priors are encoded as sufficient statistics of probability distributions, our approach seems plausible from an anatomical-functional and neuro-statistical point of view. The ensuing formal and generalisable account—based on the free energy principle—further motivate our novel treatment of stress habituation. Our analysis suggests that stress habituation has far-reaching consequences, protecting against the harmful effects of toxic stress, but on the other hand making the acceptability of precarious living conditions and the development of the obese type 2 diabetes mellitus phenotype more likely. (shrink)

The surprise minimization principle has been applied to explain various cognitive processes in humans. Originally describing perceptual and active inference, the framework has been applied to different types of decision making including long-term policies, utility maximization and exploration. This analysis extends the application of surprise minimization to a multi-agent setup and shows how it can explain the emergence of social rules and cooperation. We further show that in social decision-making and political policy design, surprise minimization is superior in many aspects (...) to the classical approach of maximizing utility. Surprise minimization shows directly what value freedom of choice can have for social agents and why, depending on the context, they enter into cooperation, agree on social rules, or do nothing of the kind. (shrink)

It is often claimed that the greatest value of the Bayesian framework in cognitive science consists in its unifying power. Several Bayesian cognitive scientists assume that unification is obviously linked to explanatory power. But this link is not obvious, as unification in science is a heterogeneous notion, which may have little to do with explanation. While a crucial feature of most adequate explanations in cognitive science is that they reveal aspects of the causal mechanism that produces the phenomenon to be (...) explained, the kind of unification afforded by the Bayesian framework to cognitive science does not necessarily reveal aspects of a mechanism. Bayesian unification, nonetheless, can place fruitful constraints on causal–mechanical explanation. 1 Introduction2 What a Great Many Phenomena Bayesian Decision Theory Can Model3 The Case of Information Integration4 How Do Bayesian Models Unify?5 Bayesian Unification: What Constraints Are There on Mechanistic Explanation?5.1 Unification constrains mechanism discovery5.2 Unification constrains the identification of relevant mechanistic factors5.3 Unification constrains confirmation of competitive mechanistic models6 ConclusionAppendix. (shrink)

Hohwy et al.’s (2008) model of binocular rivalry (BR) is taken as a classic illustration of predictive coding’s explanatory power. I revisit the account and show that it cannot explain the role of reward in BR. I then consider a more recent version of Bayesian model averaging, which recasts the role of reward in (BR) in terms of optimism bias. If we accept this account, however, then we must reconsider our conception of perception. On this latter view, I argue, organisms (...) engage in what amounts to policy-driven, motivated perception. (shrink)