Open peer commentary on the article “Decentering the Brain: Embodied Cognition and the Critique of Neurocentrism and Narrow-Minded Philosophy of Mind” by Shaun Gallagher. Abstract: Simulation theory and theory theory are interaction accounts of theory of mind that have been neurocentrically characterized. A hybrid of these theories approximates the interaction theory of social cognition, and can be described in an indexical-symbolic processing framework.

The first part of this paper examines the differences between Piaget’s constructivism, what Papert refers to as“constructionism,” and the socio-constructivist approach as portrayed by Vygotsky. All these views are developmental, and they share the notion that people actively contribute to the construction of their knowledge, by transforming their world. Yet the views also differ, each highlighting on some aspects of how children learn and grow, while leaving other questions unanswered. Attempts at integrating these views [learning through experience, through media, and through others] helps shed light on how people of different ages and venues come to make sense of their experience, and find their place – and voice – in the world. Tools, media, and cutural artifacts are the tangible forms, or mediational means, through which we make sense of our world and negociate meaning with others. In the second part of this paper, I speak to the articulations between make-believe activities and creative symbol-use as a guiding connection to rethink the aims of representations. Simulacrum and simulation, I show, play a key role besides language in helping children ground and mediate their experience in new ways. From computer-based microworlds for constructive learning (Papert’s turtle geometry, TERC’s body-syntonic graphing), to social virtual environments (MUDing). In each case, I discuss the roles of symbolic recreation, and imaginary projection (people’s abilities to build and dwell in their creations) as two powerful heuristic to keep in touch with situations, to bring what’s unknown to mind’s reach, and to explore risky ideas on safe grounds. I draw implications for education.

The concept of autonomy is of crucial importance for understanding life and cognition. Whereas cellular and organismic autonomy is based in the self-production of the material infrastructure sustaining the existence of living beings as such, we are interested in how biological autonomy can be expanded into forms of autonomous agency, where autonomy as a form of organization is extended into the behaviour of an agent in interaction with its environment (and not its material self-production) In this thesis, we focus on the development of operational models of sensorimotor agency, exploring the construction of a domain of interactions creating a dynamical interface between agent and environment. We present two main contributions to the study of autonomous agency: First, we contribute to the development of a modelling route for testing, comparing and validating hypotheses about neurocognitive autonomy. Through the design and analysis of specific neurodynamical models embedded in robotic agents, we explore how an agent is constituted in a sensorimotor space as an autonomous entity able to adaptively sustain its own organization. Using two simulation models and different dynamical analysis and measurement of complex patterns in their behaviour, we are able to tackle some theoretical obstacles preventing the understanding of sensorimotor autonomy, and to generate new predictions about the nature of autonomous agency in the neurocognitive domain. Second, we explore the extension of sensorimotor forms of autonomy into the social realm. We analyse two cases from an experimental perspective: the constitution of a collective subject in a sensorimotor social interactive task, and the emergence of an autonomous social identity in a large-scale technologically-mediated social system. Through the analysis of coordination mechanisms and emergent complex patterns, we are able to gather experimental evidence indicating that in some cases social autonomy might emerge based on mechanisms of coordinated sensorimotor activity and interaction, constituting forms of collective autonomous agency.

The article considers the complexities of thinking about the computer as a model of the mind. It examines the computer as being a model of the brain in several very different senses of “model‘. On the one hand the basic architecture of the first modern stored-program computers was „modeled on“ the brain by John von Neumann. Von Neumann also sought to build a mathematical model of the biological brain as a complex system. A similar but different approach to modeling the brain was taken by Alan Turing, who on the one hand believed that the mind simply was a universal computer, and who sought to show how brain-like networks could self-organize into Universal Turing Machines. And on the other hand, Turing saw the computer as the universal machine that could simulate any other machine, and thus any particular human skill and thereby could simulate human intelligence. This leads to a discussion of the nature of “simulation” and its relation to models and modeling. The article applies this analysis to a written correspondence between Ashby and Turing in which Turing urges Ashby to simulate his cybernetic Homeostat device on the ACE computer, rather than build a special machine.

Excerpt: The purpose of this essay is to clarify some of the important senses in which the relationship between the brain and the computer might be considered as one of “modeling.” It also considers the meaning of “simulation” in the relationships between models, computers and brains. While there has been a fairly broad literature emerging on models and simulations in science, these have primarily focused on the physical sciences, rather than the mind and brain. And while the cognitive sciences have often invoked concepts of modeling and simulation, they have been frustratingly inconsistent in their use of these terms, and the implicit relations to their scientific roles. My approach is to consider the early convolution of brain models and computational models in cybernetics, with the aim of clarifying their significance for more current debates in the cognitive sciences. It is my belief that clarifying the historical senses in which the brain and computer serve as models of each other in the historical period prior to the birth of AI and cognitive science is a crucial task for an archeology of AI and the history of cognitive science.

The concept of “autonomy,” once at the core of the original enactivist proposal in The Embodied Mind (Varela et al. in The embodied mind: cognitive science and human experience. MIT Press, Cambridge, 1991), is nowadays ignored or neglected by some of the most prominent contemporary enactivists approaches. Theories of autonomy, however, come to fill a theoretical gap that sensorimotor accounts of cognition cannot ignore: they provide a naturalized account of normativity and the resources to ground the identity of a cognitive subject in its specific mode of organization. There are, however, good reasons for the contemporary neglect of autonomy as a relevant concept for enactivism. On the one hand, the concept of autonomy has too often been assimilated into autopoiesis (or basic autonomy in the molecular or biological realm) and the implications are not always clear for a dynamical sensorimotor approach to cognitive science. On the other hand, the foundational enactivist proposal displays a metaphysical tension between the concept of operational closure (autonomy), deployed as constitutive, and that of structural coupling (sensorimotor dynamics); making it hard to reconcile with the claim that experience is sensorimotorly constituted. This tension is particularly apparent when Varela et al. propose Bittorio (a 1D cellular automata) as a model of the operational closure of the nervous system as it fails to satisfy the required conditions for a sensorimotor constitution of experience. It is, however, possible to solve these problems by re-considering autonomy at the level of sensorimotor neurodynamics. Two recent robotic simulation models are used for this task, illustrating the notion of strong sensorimotor dependency of neurodynamic patterns, and their networked intertwinement. The concept of habit is proposed as an enactivist building block for cognitive theorizing, re-conceptualizing mental life as a habit ecology, tied within an agent’s behaviour generating mechanism in coordination with its environment. Norms can be naturalized in terms of dynamic, interactively self-sustaining, coherentism. This conception of autonomous sensorimotor agency is put in contrast with those enactive approaches that reject autonomy or neglect the theoretical resources it has to offer for the project of naturalizing minds.

Key words: Autonomy, enactivism, operational closure of the nervous system, sensorimotor constitution of experience, mental life, habit, sensorimotor autonomous agency, autonomist sensorimotor enactivism.

Dynamicism has provided cognitive science with important tools to understand some aspects of “how cognitive agents work” but the issue of “what makes something cognitive” has not been sufficiently addressed yet and, we argue, the former will never be complete without the latter. Behavioristic characterizations of cognitive properties are criticized in favor of an organizational approach focused on the internal dynamic relationships that constitute cognitive systems. A definition of cognition as adaptive-autonomy in the embodied and situated neurodynamic domain is provided: the compensatory regulation of a web of stability dependencies between sensorimotor structures is created and pre served during a historical/developmental process. We highlight the functional role of emotional embodiment: internal bioregulatory processes coupled to the formation and adaptive regulation of neurodynamic autonomy. Finally, we discuss a “minimally cognitive behavior program” in evolutionary simulation modeling suggesting that much is to be learned from a complementary “minimally cognitive organization program”

Key words: minimal cognition, adaptive autonomy, neurodynamic organization, cognition-as-it-could-be, emotional embodiment

Increasing evidence suggests that mundane knowledge about objects, people, and events is grounded in the brain’s modality-specific systems. The modality-specific representations that become active to represent these entities in actual experience are later used to simulate them in their absence. In particular, simulations of perception, action, and mental states often appear to underlie the representation of knowledge, making it embodied and situated. Findings that support this conclusion are briefly reviewed from cognitive psychology, social psychology, and cognitive neuroscience. A similar representational process may underlie religious knowledge. In support of this conjecture, embodied knowledge appears central to three aspects of religious experience: religious visions, religious beliefs, and religious rituals. In religious visions, the process of simulation offers a natural account of how these experiences are produced. In religious beliefs, knowledge about the body and the environment are typically central in religious frameworks, and are likely to affect the perception of daily experience. In religious rituals, embodiments appear central to conveying religious ideas metaphorically and to establishing them in memory. To the extent that religious knowledge is like non-religious knowledge, embodiment is likely to play central roles.

Ten years ago, a group of researchers, led by Francisco Varela, were proposing an alternative vision of the immune system main behavior and function. I was part of this group. This new vision saw the immune system not as behaving distinctively with self and non-self or according to any dichotomy imposed a priori and from outside (the self-recognition vision), but rather as behaving in a unique way. From this indifferent behavior, any external impact would progressively been treated in two different ways, reactive and tolerant, but now, consequently and from inside the system (the self-assertion view). This paper will recall, through a very artificial simulation, the difference existing between these two visions. Also at that time, we believed that, from an engineering perspective, this new vision, emphasizing more the adaptability and the need for endogenous constraints than the recognition and the defensive ability, although less obvious to accept than the classical defensive one, should be more beneficial. These last ten years proved that we haven’t been convincing enough, and in this paper I resume the crusade.

Embodied cognition (EC) views propose that cognition is shaped by the kind of body that organisms possess. We give an overview of recent literature on EC, highlighting the differences between stronger and weaker versions of the theory. We also illustrate the debates on the notions of simulation, of representation, and on the role of the motor system for cognition, and we address some of the most important research topics. Future challenges concern the understanding of how abstract concepts and words are represented, and the relationship between EC and other promising approaches, the distributional views of meaning and the extended mind views.

Key words: affordances, cognitivism, distributional views, embodied cognition, emotion, enactivism, extended mind, grounded cognition, language comprehension, mental representation, mirror neurons, radical embodied cognition, representation, simulation, situated cog