Predictive coding

How much does stimulus input shape perception? The common-sense view is that our
perceptions are representations of objects and their features and that the stimulus structures the perceptual object. The problem for this view concerns perceptual biases as responsible for distortions and the subjectivity of perceptual experience. These biases are increasingly studied as constitutive factors of brain processes in recent neuroscience. In neural network models the brain is said to cope with the plethora of sensory information by predicting stimulus regularities on the basis of previous experiences. Drawing on this development, this chapter analyses perceptions as processes. Looking at olfaction as a model system, it argues for the need to abandon a stimulus- centred perspective, where smells are thought of as stable percepts, computationally linked to external objects such as odorous molecules. Perception here is presented as a measure of changing signal ratios in an environment informed by expectancy effects from top-down processes.

More and more, developers use reusable components like li- braries to produce high quality software systems. These sys tems need to satisfy not only the initial demands of their stakeho lders, but they need to also offer support for future, changing requ ire- ments. While several studies have looked at the cost of modif ying systems, there exists no work

While the notion of the brain as a prediction machine has been extremely influential and productive in cognitive science, there are competing accounts of how best to model and understand the predictive capabilities of brains. One prominent framework is of a "Bayesian brain" that explicitly generates predictions and uses resultant errors to guide adaptation. We suggest that the prediction-generation component of this framework may involve little more than a pattern completion process. We first describe pattern completion in the domain of visual perception, highlighting its temporal extension, and show how this can entail a form of prediction in time. Next, we describe the forward momentum of entrained dynamical systems as a model for the emergence of predictive processing in non-predictive systems. Then, we apply this reasoning to the domain of language, where explicitly predictive models are perhaps most popular. Here, we demonstrate how a connectionist model, TRACE, exhibits hallmarks of predictive processing without any representations of predictions or errors. Finally, we present a novel neural network model, inspired by reservoir computing models, that is entirely unsupervised and memoryless, but nonetheless exhibits prediction-like behavior in its pursuit of homeostasis. These explorations demonstrate that brain-like systems can get prediction "for free," without the need to posit formal logical representations with Bayesian probabilities or an inference machine that holds them in working memory.

Přednáška základně představuje v současnosti velmi prominentní teorii, která si klade za cíl být sjednocenou teorií funkce mysli a mozku. Hlavním principem je, že se mozek neustále snaží předvídat následující vstupy a díky tomu si tvoří model světa. Teorie se na základě toho snaží vysvětlit takové jevy jako je vnímání, pozornost, učení, tvorba přesvědčení, jednání, ale i např. jáství, schizofrenii či autismus. Stojí na konceptech jako je Bayesovský mozek, prediktivní kódování a tzv. free energy principle. Tento přístup srovnám s přístupy jako je kognitivismus, vtělená kognice a pod.

Résumé  Le traitement de la parole a connu ces dernières années un formidable développement lié aux avancées technologiques des composants de traitement numérique des signaux et à la numérisation grandissante des réseaux. Cet article fournit une analyse des principales techniques qui se sont imposées récemment dans les domaines du codage, de la reconnaissance et de la synthèse de la parole. En

This paper is in Dutch, if requested (& time permitting) we'll translate it into English.

The HIPPEA theory of autism not only created a lot of interest in academic circles but also got a lot of response in the autistic community. This contribution is specifically inspired by reactions of the (Flemish) Reading and Advice Group Adults with Autism where we had an opportunity to present HIPPEA on a 'member day'. These reactions were positive because HIPPEA was seen to put the cause of autism in a different way of dealing with (uncertainty in) information. This corresponds to the shared lived experience of autism of those present as this atypical way of information processing is directly connected to the most immediate autistic

Published version: http://wta.swptijdschriften.nl/Magazine/Article/3698.

We distinguish between three philosophical views on the neuroscience of predictive models: predictive coding (associated with internal Bayesian models and prediction error minimization), predictive processing (associated with radical con-nectionism and 'simple' embodiment) and predictive engagement (associated with enactivist approaches to cognition). We examine the concept of active inference under each model and then ask how this concept informs discussions of social cognition. In this context we consider Frith and Friston's proposal for a neural hermeneutics, and we explore the alternative model of enactivist hermeneutics.

This paper argues that the still-emerging paradigm of situated cognition requires a more systematic perspective on media to capture the enculturation of the human mind. By virtue of being media, cultural artifacts present central experiential models of the world for our embodied minds to latch onto. The paper identifies references to external media within embodied, extended, enactive, and predictive approaches to cognition, which remain underdeveloped in terms of the profound impact that media have on our mind. To grasp this impact, I propose an enactive account of media that is based on expansive habits as media-structured, embodied ways of bringing forth meaning and new domains of values. We apply such habits, for instance, when seeing a picture or perceiving a movie. They become established through a process of
reciprocal adaptation between media artifacts and organisms and define the range of
viable actions within such a media ecology. Within an artifactual habit, we then become attuned to a specific media work (e.g., a TV series, a picture, a text, or even a city) that engages us. Both the plurality of habits and the dynamical adjustments within a habit require a more flexible neural architecture than is addressed by classical cognitive neuroscience. To detail how neural and media processes interlock, I will introduce the concept of neuromediality and discuss radical predictive processing accounts that could contribute to the externalization of the mind by treating media themselves as generative models of the world. After a short primer on general media theory, I discuss media examples in three domains: pictures and moving images; digital media; architecture and the built environment. This discussion demonstrates the need for a new cognitive media theory based on enactive artifactual habits—one that will help us gain perspective on the continuous re-mediation of our mind.

The relationship between the states of awareness that emerge in hypnosis and meditation or even between different meditation practices is of fundamental importance to understanding the range of potentials open to human consciousness. The recent emergence of cognitive neuroscience provides a powerful set of tools for researchers of human conscious states to probe not only their effects but the underlying causal dynamics of their operation. Within cognitive neuroscience the Bayesian framework of predictive coding has generated deep insights into the fundamental unity of the ordinary psychological processes of perception, learning, attention, memory and action. In what follows key concepts from predictive coding, active inference, generative models and interoceptive predictive coding are applied to our understanding of hypnosis and meditation states to provide a unified theory of these diverse states with clear implications for researchers and clinicians alike.

This presentation gives a broad introduction to the exciting concept of predictive processing as the unifying theory of the brain and mind, in particular the Bayesian brain, predictive coding and free energy principle. I show how these theories aim to explain cognitive phenomena like perception, learning, belief formation, attention, motor control, the self or schizophrenia. Finally, I attempt to compare these new ideas with behaviorism, cognitivism, embodied and enactive cognition.

The imagination is central to esoteric practices, but so far scholars have shown little interest in exploring cognitive theories of how the imagination works. The only exception is Tanya Luhrmann's interpretive drift theory and related research on mental imagery cultivation, which has been used to explain the subjective persuasiveness of modern ritual magic. This article draws on recent work in the neuroscience of perception in order to develop a general theory of kataphatic (that is, imagery based) practice that goes beyond the interpretive drift theory. Mental imagery is intimately linked with perception. Drawing on " predictive coding " theory, I argue that kataphatic practices exploit the probabilistic, expectation-based way that the brain processes sensory information and creates models (perceptions) of the world. This view throws light on a wide range of features of kataphatic practices, from their contemplative and cognitive aspects, to their social organization and demographic make-up, over their pageantry and material culture. By connecting readily observable features of kataphatic practice to specific neurocognitive mechanisms related to perceptual learning and cognitive processing of mental imagery, the predictive coding paradigm also creates opportunities for combining historical research with experimental approaches within the study of religion. I illustrate how this framework may enrich the study of Western esotericism in particular by applying it to the paradigmatic case of " astral travel " as it has developed from the Golden Dawn tradition of ritual magic, especially by Aleister Crowley.

How do cultural artifacts influence the ways we experience and act? In this paper I propose that habits provide a central link between human organisms and the socio- cultural environment and that cognition is cultural tout court. I will defend an enactive account of habits that sees them as expansive in the temporal (they relate us to a history of socio-cultural interactions) and spatial sense (they are co-constituted by cultural artifacts and our enculturated bodies). This account is based on Dewey’s pragmatist- organicist concept of habits that rigorously anchors experience in culture. I will trace cultural factors in Dewey’s philosophy and 4E (embodied, embedded, enacted, extended) cognitive science with a focus on pervasive artifacts – such as architecture, pictures, and moving images – which have become part of our habits of perceiving. I finally situate such habits within recent predictive coding theories that claim to provide the corresponding neuroscientific theory to our cultural minds.

This commentary is structured according to the following discussion points: (1) the compatibility of the recently popular predictive processing framework with 4E approaches, (2) the possible constraints that phenomenological characterizations of cognitive phenomena put on modeling and experimentation, illustrated by the debate on social cognition, (3) the putative replacement of the classical representationalist approach to cognition by a rigorously dynamic and embodied approach to cognition, and (4) the contrast between brainbound and extended or integrated cognitive systems.

Hallucinations possess two main components: (i) a sensory content; and (ii) a sense that the sensory content is real. Influential models of schizophrenic hallucination claim that both the sensory content and the sense of reality can be explained in terms of metacognitive dysfunction. This chapter assesses whether such a claim holds for schizophrenic and drug-induced hallucinations; it further attempts to determine the actual role of metacognition in hallucination and how this role is liable to vary across cultures. It is first argued that the notion of sense of reality is heterogeneous and should therefore be divided into distinct kinds. Next, some monitoring-based models of hallucination are presented. After having briefly distinguished between different levels of metacognitive processing, I show that these monitoring-based models are metacognitive only to a limited extent and that they fail to explain important aspects of the content and sense of reality of hallucinations. I subsequently suggest that the main mechanisms of serotoninergic-hallucinogens are not metacognitive whereas those of anticholinergic-hallucinogens importantly tap into subpersonal metacognitive processes. Looking specifically at the use of ayahuasca across different Amazonian indigenous groups, I put forward the idea that the metacognitive properties of hallucinogenic experiences can be variously exploited or ignored depending on cultural expectations. Finally, I examine how anthropological and linguistic evidence of the ritualized use of hallucinogens supports the existence of multiple metacognitive norms in religion.

A large body of research in cognitive psychology and neuroscience draws on Bayesian statistics to model information processing within the brain. Many theorists have noted that this research seems to be in tension with a large body of experimental results purportedly documenting systematic deviations from Bayesian updating in human belief formation. In response, proponents of the Bayesian brain hypothesis contend that Bayesian models can accommodate such results by making suitable assumptions about model parameters (for example, priors, likelihoods, and utility functions). To make progress in this debate, I argue that it is fruitful to focus not on specific experimental results but rather on what I call the 'sources of epistemic irrationality' in human cognition. I identify four such sources and I explore whether and, if so, how Bayesian models can be reconciled with them: (1) processing costs; (2) evolutionary suboptimality; (3) motivated cognition; and (4) error management.

Friston, K., Fortier, M. & Friedman, D. A. (2018). Of woodlice and men: A Bayesian account of cognition, life and consciousness. An interview with Kark Friston. ALIUS Bulletin, 2,
17-43.

The entire issue of the ALIUS Bulletin can be downloaded for free here: https://www.aliusresearch.org/bulletin.html

We argue that one important aspect of the "cognitive neuroscience revolution" identified by Boone and Piccinini (2015) is a dramatic shift away from thinking of cognitive representations as arbitrary symbols towards thinking of them as icons that replicate structural characteristics of their targets. We argue that this shift has been driven both "from below" and "from above" — that is, from a greater appreciation of what mechanistic explanation of information-processing systems involves ("from below"), and from a greater appreciation of the problems solved by bio-cognitive systems, chiefly regulation and prediction ("from above"). We illustrate these arguments by reference to examples from cognitive neuroscience, principally representational similarity analysis and the emergence of (predictive) dynamical models as a central postulate in neurocognitive research.

Recently, a number of neuroscientists and philosophers have taken the so-called predictive coding approach to support a form of radical neuro-representationalism, according to which the content of our conscious experiences is a neural construct, a brain-generated simulation. There is remarkable similarity between this account and ideas found in and developed by German neo-Kantians in the mid-19th century. Some of the neo-Kantians eventually came to have doubts about the cogency and internal consistency of the representationalist framework they were operating within. In this paper, I will first argue that some of these concerns ought to be taken seriously by contemporary proponents of predictive coding. After having done so, I will turn to phenomenology. As we shall see, Husserl's endorsement of transcendental idealism was partially motivated by his rejection of representationalism and phenomenalism and by his attempt to safeguard the objectivity of the world of experience. This confronts us with an intriguing question. Which position is best able to accommodate our natural inclination for realism: Contemporary neuro-representationalism or Husserl's transcendental idealism?

ABSTRACT: The free-energy principle states that all systems that minimize their free energy resist a tendency to physical disintegration. Originally proposed to account for perception , learning, and action, the free-energy principle has been applied to the evolution, development, morphology, anatomy and function of the brain, and has been called a postulate, an unfalsifiable principle, a natural law, and an imperative. While it might afford a theoretical foundation for understanding the relationship between environment , life, and mind, its epistemic status is unclear. Also unclear is how the free-energy principle relates to prominent theoretical approaches to life science phenomena, such as organicism and mechanism. This paper clarifies both issues, and identifies limits and prospects for the free-energy principle as a first principle in the life sciences.

Predictive processing (PP) is a paradigm in computational and cognitive neuroscience that has recently attracted significant attention across domains, including psychology, robotics, artificial intelligence and philosophy. It is often regarded as a fresh and possibly revolutionary paradigm shift, yet a handful of authors have remarked that aspects of PP seem reminiscent of the work of 18th century philosopher Immanuel Kant. To date there have not been any substantive discussions of how exactly PP links back to Kant. In this article, I argue that several core aspects of PP were anticipated by Kant (1996/1787) in his works on perception and cognition. Themes from Kant active in PP include: (1) the emphasis on " top-down " generation of percepts; (2) the role of " hyperpriors " ; (3) the general function of " generative models " ; (4) the process of " analysis-by-synthesis " and (5) the crucial role of imagination in perception. In addition to these, I also point out that PP echoes Kant's general project in that it aims to explain how minds track causal structure in the world using only sensory data, and that it uses a reverse-engineer or " top-down " method of analysis. I then locate a possible source of Kant's influence on PP by tracing the paradigm back to Hermann von Helmholtz, who saw himself as providing a scientific implementation of Kant's conclusions. I conclude by arguing that PP should not be regarded as a new paradigm, but is more appropriately understood as the latest incarnation of an approach to perception and cognition initiated by Kant and refined by Helmholtz.

Predictive processing has recently been advanced as a global cognitive architecture for the brain. I argue that its commitments concerning the nature and format of cognitive representation are unable to account for two basic characteristics of conceptual thought: first, its generality--the fact that we can think and flexibly reason about phenomena at any level of spatiotemporal scale and abstraction; second, its rich compositionality--the specific way in which concepts productively combine to yield our thoughts. I consider two strategies for avoiding these objections and I argue that both confront formidable challenges.

I identify three lessons from Kenneth Craik's landmark book "The Nature of Explanation" for contemporary debates surrounding the existence, importance, and nature of mental representation: first, an account of mental representations as neural structures that function analogously to public models; second, an appreciation of prediction as the central component of intelligence in demand of such models; and third, a metaphor for understanding the brain as an engineer, not a scientist. I then relate these insights to discussions surrounding the representational status of predictive processing—which, I argue, provides a contemporary vindication of Craik's extremely prescient "hypothesis on the nature of thought."

Researchers in the field of computational psychiatry have recently sought to model the formation and retention of
delusions in terms of dysfunctions in a process of hierarchical Bayesian inference. I present a systematic review of
such models and raise two challenges that have not received sufficient attention in the literature. First, the
characteristic that is supposed to most sharply distinguish hierarchical Bayesian models from their competitors is their
abandonment of the distinction between perception and cognition in favour of a unified inferential hierarchy. Standard
ways of characterising this hierarchy, however, are inconsistent with the range of phenomena that delusions can
represent. Second, there is little evidence that belief fixation in the healthy population is Bayesian, and an abundance of evidence that it is not. Given this, attempts to model delusions in terms of dysfunctions in a process of
Bayesian inference are of dubious theoretical value.

The discovery of mirror neurons in the ventral premotor cortex (area F5) and inferior parietal cortex (area PFG) in the macaque monkey brain has provided the physiological evidence for direct matching of the intrinsic motor representations of the self and the visual image of the actions of others. The existence of mirror neurons implies that the brain has mechanisms reflecting shared self and other action representations. This may further imply that the neural basis self-body representations may also incorporate components that are shared with other-body representations. It is likely that such a mechanism is also involved in predicting other's touch sensations and emotions. However, the neural basis of shared body representations has remained unclear. Here, we propose a neural basis of body representation of the self and of others in both human and non-human primates. We review a series of behavioral and physiological findings which together paint a picture that the systems underlying such shared representations require integration of conscious exteroception and interoception subserved by a cortical sensory-motor network involving parieto-inner perisylvian circuits (the ventral intraparietal area [VIP]/inferior parietal area [PFG]-secondary somatosensory cortex [SII]/posterior insular cortex [pIC]/anterior insular cortex [aIC]). Based on these findings, we propose a computational mechanism of the shared body representation in the predictive coding (PC) framework. Our mechanism proposes that processes emerging from generative models embedded in these specific neuronal circuits play a pivotal role in distinguishing a self-specific body representation from a shared one. The model successfully accounts for normal and abnormal shared body phenomena such as mirror-touch synesthesia and somatoparaphrenia. In addition, it generates a set of testable experimental predictions.

Recent research in cognitive and computational neuroscience portrays the neocortex as a hierarchically structured prediction machine. Several theorists have drawn on this research to challenge the traditional distinction between perception and cognition—specifically, to challenge the very idea that perception and cognition constitute useful kinds from the perspective of cognitive neuroscience. In place of this traditional taxonomy, such theorists advocate a unified inferential hierarchy subject to substantial bi-directional message passing. I outline the nature of this challenge and then raise two objections to the cognitive architecture it proposes as a replacement: first, standard ways of characterising this inferential hierarchy are in tension with the representational reach of conceptual thought; second, there is compelling evidence that commonsense reasoning is structured around highly domain-specific intuitive theories that are difficult to situate within a single hierarchy.

Motor behavior is a phenomenon where the components making up a biological system are organized so as to ensure the coordination of a purposeful movement. According to computational theories, behavior is defined as a motor problem the solution of which can be found by systems divided hierarchically. The components process and communicate information representing the relevant variables of the motor problem (positions, trajectories, velocities, forces, etc.) which are, in turn, assumed to be organized as a hierarchy of increasing abstraction and complexity. The challenge is to tackle the four core problems of behavior: a) The high number of degrees of freedom and their interactions; b) The redundancy of degrees of freedom; c) The anticipation of the effects of movement; d) The uncertainty in information. Classical computational theories advance explanatory schemas made of structured sets of internal models (forward and inverse). More recently, the Bayesian approach advances a more homogeneous hierarchical schema which is supposed to account for uncertainty in information. This research shows that computational theories, including the Bayesian approach, are crippled by an unsolvable dilemma: The first horn is that if the models can scale up while staying computationally tractable, i.e. the computations can be carried out in a reasonable amount of time, then they fail to reproduce the versatility which characterizes the behavior of living beings. The second horn is that if the models aspire to reproduce biological versatility, then they are intractable.

The general principles of perceptuo-motor processing and memory give rise to the Now-or-Never bottleneck constraint imposed on the organization of the language processing system. In particular, the Now-or-Never bottleneck demands for an appropriate structure of linguistic input and rapid incorporation of both linguistic and multisensory contextual information in a progressive, integrative manner. I argue that the emerging predictive processing framework is well-suited for the task of providing a comprehensive account of language processing under the Now-or-Never constraint. Moreover, this framework presents a stronger alternative to the Chunk-and-Pass account proposed by Christiansen and Chater (2016) as it better accommodates the available evidence concerning the role of context (both in the narrow and wider senses) in language comprehension at various levels of linguistic representation. Furthermore, the predictive processing approach allows for treating language as a special case of domain-general processing strategies, suggesting deep parallels with other cognitive processes such as vision.

According to the predictive coding theory of cognition (PCT), brains are predictive machines that use perception and action to minimize prediction error, i.e. the discrepancy between bottom–up, externally-generated sensory signals and top–down, internally-generated sensory predictions. Many consider PCT to have an explanatory scope that is unparalleled in contemporary cognitive science and see in it a framework that could potentially provide us with a unified account of cognition. It is also commonly assumed that PCT is a representational theory of sorts, in the sense that it postulates that our cognitive contact with the world is mediated by internal representations. However, the exact sense in which PCT is representational remains unclear; neither is it clear that it deserves such status—that is, whether it really invokes structures that are truly and nontrivially representational in nature. In the present article, I argue that the representational pretensions of PCT are completely justified. This is because the theory postulates cognitive structures—namely action-guiding, detachable, structural models that afford representational error detection—that play genuinely representational functions within the cognitive system.

The contrast between self- and other-produced tickles, as a special case of sensory attenuation for self-produced actions, has long been a target of empirical research. While in standard wake states it is nearly impossible to tickle oneself, there are interesting exceptions. Notably, participants awakened from REM (rapid eye movement-) sleep dreams are able to tickle themselves. So far, however, the question of whether it is possible to tickle oneself and be tickled by another in the dream state has not been investigated empirically or addressed from a theoretical perspective. Here, we report the results of an explorative web-based study in which participants were asked to rate their sensations during self-tickling and being tickled during wakefulness, imagination, and lucid dreaming. Our results, though highly preliminary, indicate that in the special case of lucid control dreams, the difference between self-tickling and being tickled by another is obliterated, with both self- and other produced tickles receiving similar ratings as self-tickling during wakefulness. This leads us to the speculative conclusion that in lucid control dreams, sensory attenuation for self-produced tickles spreads to those produced by non-self dream characters. These preliminary results provide the backdrop for a more general theoretical and metatheoretical discussion of tickling in lucid dreams in a predictive processing framework. We argue that the primary value of our study lies not so much in our results, which are subject to important limitations, but rather in the fact that they enable a new theoretical perspective on the relationship between sensory attenuation, the self-other distinction and agency, as well as suggest new questions for future research. In particular, the example of tickling during lucid dreaming raises the question of whether sensory attenuation and the self-other distinction can be simulated largely independently of external sensory input.