Xabier Barandiaran

This paper investigates the ontological characterization of Large Language Models (LLMs) like ChatGPT. Between inflationary and deflationary accounts, we pay special attention to their status as agents. This requires explaining in detail the architecture, processing, and training procedures that enable LLMs to display their capacities, and the extensions used to turn LLMs into agent-like systems. After a systematic analysis we conclude that a LLM fails to meet necessary and sufficient conditions for autonomous agency in the light of embodied theories of mind: the individuality condition (it is not the product of its own activity, it is not even directly affected by it), the normativity condition (it does not generate its own norms or goals), and, partially the interactional asymmetry condition (it is not the origin and sustained source of its interaction with the environment). If not agents, then… What are LLMs? We argue that ChatGPT should be characterized as an interlocutor or linguistic automaton, a library-that-talks, devoid of (autonomous) agency, but capable to engage performatively on non-purposeful yet purpose-structured and purpose-bounded tasks. When interacting with humans, a “ghostly” component of the human-machine interaction makes it possible to enact genuine conversational experiences with LLMs. Despite their lack of sensorimotor and biological embodiment, LLMs textual embodiment (the training corpus), digital extended interface embodiments, and resource-hungry computational embodiment, significantly transform existing forms of machine automatism and human agency. Beyond assisted and extended agency, the LLM-human coupling can produce midtended forms of agency, closer to the production of intentional agency than to the extended instrumentality of any previous technologies.

Teleosemantics explains meaning by appealing to the biological norms that make error possible, but most work in the field still anchors those norms in evolutionary “selected‐effect” functions. We develop an organismic alternative grounded in the self-maintenance of autonomous systems. Building on sensorimotor theory and enactivism, we reconceptualise goals as second-order constraints—attractors in a dynamic sensorimotor field—and show how they are nested into a heterarchy of means–end relations connecting global self‑maintenance of the sensorimotor organization and identity of system with the most basic sensorimotor coordinations. Drawing on this framework, we identify the minimal necessary requirements for genuine teleological behaviour: 1. Initiation that individuates an action in relation to a goal, 2. Modulatory execution that adaptively compensates perturbations or deploys alternative strategies in relation to the goal, and 3. Meaningful termination through successful acknowledgment and transition, goal-sensitive persistence, or coherent abandonment. We locate normativity and error in the dynamic presuppositions among sensorimotor schemes and goals: failure manifests as incoordination within, or across, goal branches. We finally illustrate our account with a detailed analysis of the well known A-not-B error in psychology, showing how infants meet our teleological criteria while revealing incoordination as tension breakdowns inside a heterarchy of goals and sensorimotor schemes. Teleological behaviour, therefore, is the management of tensions and dynamic presuppositions within the goal heterarchy, not a chain of mechanical causes, thus delivering a naturalised notion of purposiveness that opens up an alternative route to teleosemantics beyond evolutionary selection.

This paper introduces the concept of “generative midtended cognition”, that explores the integration of generative AI technologies with human cognitive processes. The term “generative” reflects AI’s ability to iteratively produce structured outputs, while “midtended” captures the potential hybrid (human-AI) nature of the process. It stands between traditional conceptions of _in_tended creation, understood as steered or directed from with_in_, and _ex_tended processes that bring exo-biological processes into the creative process. We examine the working of current generative technologies (based on multimodal transformer architectures typical of large language models like ChatGPT) to explain how they can transform human cognitive agency beyond what the conceptual resources of standard theories of extended cognition can capture. We suggest that the type of cognitive activity typical of the coupling between a human and generative technologies is closer (but not equivalent) to social cognition than to classical extended cognitive paradigms. Yet, it deserves a specific treatment. We provide an explicit definition of _generative midtended cognition_ in which we treat interventions by AI systems as constitutive of the agent’s intentional creative processes. Furthermore, we distinguish two dimensions of generative hybrid creativity: 1. Width: captures the sensitivity of the context of the generative process (from the single letter to the whole historical and surrounding data), 2. Depth: captures the granularity of iteration loops involved in the process. Generative midtended cognition stands in the middle depth between _conversational_ forms of cognition in which complete utterances or creative units are exchanged, and micro-cognitive (e.g. neural) subpersonal processes. Finally, the paper discusses the potential risks and benefits of widespread generative AI adoption, including the challenges of authenticity, generative power asymmetry, and creative boost or atrophy.

In recent years, (autonomy-centered) enactivism has been used to provide an integrative and relational account of mental conditions. A significant advancement lies in its naturalized and pluralistic treatment of normativity, which transcends traditional objectivist and normativist dichotomies. This article explores the varieties of normativity within this paradigm and their implications for understanding mental conditions. We address purported challenges associated with the integration of social normativity into the enactive naturalistic framework of cognition, particularly concerning mental conditions. Drawing upon the distinction between the constitution problem and the status problem, we conceptualize mental conditions as intersubjectively constituted with an intersubjectively negotiated status. Adopting a participatory sense-making perspective, we address three challenges posed by social movements: (1) Hermeneutical dilemmas related to the ontological openness of mental health categories. (2) The difficulties and urgency to mitigate epistemic injustices. (3) The complex attribution of (social) responsibility in psychological wellbeing. In conclusion, this perspective prompts a reevaluation of epistemological assumptions, advocating for a second-person and engaged perspective on mental conditions.

The notion of malfunction is critical to biological explanation. It provides a test-bed for the normative character of functional attribution. Theories of biological functioning must permit traits to operate but, at the same time, be judged as malfunctioning (in some naturalized, non-arbitrary sense). Whereas malfunctioning has attracted most attention and discussion in evolutionary etiological approaches, systemic and organizational ones have been less discussed. The most influential of the organizational approaches (by Saborido, Moreno and Mossio) takes a dual-order approach to malfunctions, as a set of functions that fit first-order constitutive norms but fail to obey second-order regulatory ones. We argue that this conception is unnecessarily complicated (malfunctions do not need to arise as a result of two conflicting orders of norms) and too narrow (it excludes canonical cases of malfunctioning). We provide an alternative organizational account grounded on viability theory. The dynamics of the traits that constitute an organism define the normative field of its viability space: sugar must be replaced at certain rate, blood must be pumped at a certain pace, etc. A trait operates normatively when its effects on the viability space correlate positively with the normative field. Three senses of dysfunctionality might be distinguished: subfunctional operations are those that positively correlate with the normative field but quantitatively fail to match the required speed; malfunctional operations are those that do not positively correlate with the normative field; and, nonfunctional traits either don’t operate at all or operate with null effect on the normative field.

We are currently witnessing the emergence of new forms of collective identities and a redefinition of the old ones through networked digital interactions, and these can be explicitly measured and analyzed. We distinguish between three major trends on the development of the concept of identity in the social realm: (1) an essentialist sense (based on conditions and properties shared by members of a group), (2) a representational or ideational sense (based on the application of categories by oneself or others), and (3) a relational and interactional sense (based on interaction processes between actors and their environments). The interactional approach aligns with current empirical and methodological progress in social network analysis. Moreover, it has been argued that, within the network society, the notion of collective identity (Melucci, 1995) in the political field must be rethought as technologically mediated and interactive. We suggest that collective identities should be understood as recurrent, cohesive, and coordinated communicative interaction networks. We here propose that such identities can be depicted by: (a) mapping and filtering a relevant interaction network, (b) delimiting a set of communities, (c) determining the strongly connected component(s) of such communities (the core identity) in a directed graph, and (d) defining the identity audiences and sources within the community. This technical graph–theoretical characterization is explained and justified in...

The concept of agency is of crucial importance in cognitive science and artificial intelligence, and it is often used as an intuitive and rather uncontroversial term, in contrast to more abstract and theoretically heavy-weighted terms like “intentionality”, “rationality” or “mind”. However, most of the available definitions of agency are either too loose or unspecific to allow for a progressive scientific program. They implicitly and unproblematically assume the features that characterize agents, thus obscuring the full potential and challenge of modeling agency. We identify three conditions that a system must meet in order to be considered as a genuine agent: a) a system must define its own individuality, b) it must be the active source of activity in its environment (interactionalasymmetry) and c) it must regulate this activity in relation to certain norms (normativity). We find that even minimal forms of proto-celular systems can already provide a paradigmatic example of genuine agency. By abstracting away some specific details of minimal models of living agency we define the kind of organization that is capable to meet the required conditions for agency (which is not restricted to living organisms). On this basis, we define agency as an autonomous organization that adaptively regulates its coupling with its environment and contributes to sustaining itself as a consequence. We find that spatiality and temporality are the two fundamental domains in which agency spans at different scales. We conclude by giving an outlook to the road that lies ahead in the pursuit to understand, model and synthesis agents.

Many digital technologies, designed and controlled by intensive data-driven corporate platforms, have become ubiquitous for many of our daily activities. This has raised political and ethical concerns over how they might be threatening our personal autonomy. However, not much philosophical attention has been paid to the specific role that their hyper-designed (sensorimotor) interfaces play in this regard. In this paper, we aim to offer a novel framework that can ground personal autonomy on sensorimotor interaction and, from there, directly address how technological design affects personal autonomy. To do this, we will draw from enactive sensorimotor approaches to cognition, focusing on the central notion of habits, understood as sensorimotor schemes that, in networked relations, give rise to sensorimotor agency. Starting from sensorimotor agency as a basis for more complex forms of personal autonomy, our approach gives us grounds to analyse our relationship with technology (in general) and to distinguish between autonomy-enhancing and autonomy-diminishing technologies. We argue that, by favouring/obstructing the enactment of certain (networks of) habits over others, technologies can directly act upon our personal autonomy, locally and globally. With this in mind, we then discuss how current digital technologies are often being designed to be autonomy-diminishing (as is the case of “dark patterns” in design), and sketch some ideas on how to build more autonomy-enhancing digital technologies.