Andrew Clark

Recent research has relied on the use of fine-tuning techniques to incorporate philosophical knowledge into Large Language Models (LLMs). The present paper outlines an alternative approach to the development of such systems—one that is rooted in a technique known as Retrieval-Augmented Generation (RAG). In contrast to fine-tuning, RAG does not seek to adjust the internal parameters (or internal memory) of an LLM. Instead, RAG relies on the retrieval of information from an externally-situated store, which functions as a form of non-parametric (or external) memory. Applying this technique to the works of the contemporary philosopher Andy Clark yields Digital Andy: an LLM that is able to respond to questions about the extended mind. This serves as a practical demonstration of RAG-based techniques, highlighting how philosophical knowledge can be ‘incorporated’ into an LLM without the need for additional machine learning. But Digital Andy’s reliance on extra-systemic resources also raises questions about the scope of active externalist theorizing, encouraging us to consider Digital Andy’s status as an extended cognitive/computational system. Addressing these questions reveals some interesting points of convergence between the philosophical effort to understand the extended mind and the technological effort to build the next generation of LLMs.

This article applies the thesis of the extended mind to ambient smart environments. These systems are characterised by an environment, such as a home or classroom, infused with multiple, highly networked streams of smart technology working in the background, learning about the user and operating without an explicit interface or any intentional sensorimotor engagement from the user. We analyse these systems in the context of work on the “classical” extended mind, characterised by conditions such as “trust and glue” and phenomenal transparency, and find that these conditions are ill-suited to describing our engagement with ambient smart environments. We then draw from the active inference framework, a theory of brain function which casts cognition as a process of embodied uncertainty minimisation, to develop a version of the extended mind grounded in a process ontology, where the boundaries of mind are understood to be multiple and always shifting. Given this more fluid account of the extended mind, we argue that ambient smart environments should be thought of as extended allostatic control systems, operating more or less invisibly to support an agent’s biological capacity for minimising uncertainty over multiple, interlocking timescales. Thus, we account for the functionality of ambient smart environments as extended systems, and in so doing, utilise a markedly different version of the classical thesis of extended mind.

In the movie, Memento, the hero, Leonard, suffers from a form of anterograde amnesia that results in an inability to lay down new memories. Nonetheless, he sets out on a quest to find his wife’s killer, aided by the use of notes, annotated polaroids, and (for the most important pieces of information obtained) body tattoos. Using these resources he attempts to build up a stock of new beliefs and to thus piece together the puzzle of his wife’s death. At one point in the movie, a character exasperated by Leonard’s lack of biological recall, shouts.

[Andy Clark] What is the relation between perceptual experience and the suite of sensorimotor skills that enable us to act in the very world we perceive? The relation, according to 'sensorimotor models' is tight indeed. Perceptual experience, on these accounts, is enacted via skilled sensorimotor activity, and gains its content and character courtesy of our knowledge of the relations between movement and sensory stimulation. I shall argue that this formulation is too extreme, and that it fails to accommodate the substantial firewalls, dis-integrations, and special-purpose streamings that form the massed strata of human cognition. In particular, such strong sensorimotor models threaten to obscure the computationally potent insensitivity of key information-processing events to the full subtleties of embodied cycles of sensing and moving. /// [Naomi Eilan] The strong sensorimotor account of perception gives self-induced movements two constitutive roles in explaining visual consciousness. The first says that self-induced movements are vehicles of visual awareness, and for this reason consciousness 'does not happen in the brain only'. The second says that the phenomenal nature of visual experiences is consists in the action-directing content of vision. In response I suggest, first, that the sense in which visual awareness is active should be explained by appeal to the role of attention in visual consciousness, rather than self-induced movements; and second, that the sense in which perceptual consciousness does not happen in the brain only should be explained by appeal to the relational nature of perceptual consciousness, appeal to which also shows why links with action cannot exhaust phenomenal content.

Advanced biological brains are by nature open-ended opportunistic controllers. Such controllers compute, pretty much on a moment-to-moment basis, what problem-solving resources are readily available and recruit them into temporary problem-solving wholes. Neural plasticity, exaggerated in our own species, makes it possible for such resources to become factored deep into both our cognitive and physical problem-solving routines. One way to think about this is to depict the biological brain as a master of what I shall dub ‘ecological control’. Ecological control is the kind of top-level control that does not micro-manage every detail, but rather encourages substantial devolvement of power and responsibility. This kind of control allows much of our skill at walking to reside in the linkages and elastic properties of muscles and tendons. And it allows much of our prowess at thought and reason to depend upon the robust and reliable operation, often in dense brain-involving loops, of a variety of non-biological problem-solving resources spread throughout our social and technological surround. Are the complex distributed systems that result in some sense ‘out of control’, beyond the reach of useful governance? I shall argue that they are not, although understanding them requires us to re-think some key ideas about control and the nature of the self. To make this case, I shall first examine some strategies for efficient, external opportunity exploiting control in simple systems. I shall then argue that many of the same lessons apply to the case of higher-level human problem-solving.

Thought happens. Here I sit, sipping coffee, scribbling on paper, accessing files, reading and re-reading those four wonderful, challenging, yet immaculately constructive reviews. And somewhere, and to my eternal surprise, thought happens. But where, amidst the whirl of organization, should we locate the cognitive process? One possibility is that everything worth counting as (all or part) of any genuinely cognitive process hereabouts is firmly located inside the head, safe behind the ancient fortress of skin and skull. All the rest, according to this surgically neat view, is scene setting: preparing and maintaining the pitch upon which the great thinking organ performs.

Enquiries into the possible nature and scope of innate knowledge never proceed in an empirical vaccuum. Instead, such conjectures are informed by a theory concerning probable representational form. Classical approaches to the nativism debate often assume a quasi-linguistic form of knowledge representation and deliniate a space of options accordingly. Recent connectionist theorizing posits a different kind of represenational form, and thus determines a different picture of the space of possible nativisms.

What do linguistic symbols do for minds like ours, and how (if at all) can basic embodied, dynamical and situated approaches do justice to high-level human thought and reason? These two questions are best addressed together, since our answers to the first may inform the second. The key move in ‘scaling-up’ simple embodied cognitive science is, I argue, to take very seriously the potent role of human-built structures in transforming the spaces of human learning and reason. In particular, in this paper I look at a range of cases involving what I dub ‘surrogate situations’. Here, we actively create restricted artificial environments that allow us to deploy basic perception-action- reason routines in the absence of their proper objects. Examples include the use of real-world models, diagrams and other concrete external symbols to support dense looping interactions with a variety of stable external structures that stand in for the absent states of affairs.

First, a theoretical background to the volume’s topic, extended epistemology, is provided by a brief outline of its cross-disciplinary theoretical lineage and some key themes. In particular, it is shown how and why the emergence of recent and more egalitarian thinking in the cognitive sciences about the nature of human cognizing and its bounds—viz., the so-called ‘extended cognition’ program, and the related idea of an ‘extended mind’—has important and interesting ramifications in epistemology. Second, an overview is provided of the papers included as chapters in the volume. The sixteen contributions are divided into two categories: those that engage with foundational issues to do with extended epistemology, and those that pursue applications of extended epistemology to new areas of research.

Markets, companies and various forms of business organizations may all (we have argued) be usefully viewed through the lens of CAS -- the theory of complex adaptive systems. In this chapter, I address one fundamental issue that confronts both the theoretician and the business manager: the nature and opportunities for control and intervention in complex adaptive regimes. The problem is obvious enough. A complex adaptive system, as we have defined it, is soft assembled and largely self-organizing. This means that it is the emergent product of multiple, and often very heterogeneous, interacting factors and forces, and that the crucial interactions are not controlled and orchestrated by an overseeing executive, detailed program, or any other source of strict hierarchical control. There is thus a pressing problem -- are such systems strictly out of control and beyond the reach of useful governance? I shall argue that they are not. Such systems, though initially unfamiliar, can nonetheless be led, influenced and enabled in a variety of ways.

We seem, or so it seems to some theorists, to experience a rich stream of highly detailed information concerning an extensive part of our current visual surroundings. But this appearance, it has been suggested, is in some way illusory. Our brains do not command richly detailed internal models of the current scene. Our seeings, it seems, are not all that they seem. This, then, is the Grand Illusion. We think we see much more than we actually do. In this paper I shall (briefly) rehearse the empirical evidence for this rather startling claim, and then critically examine a variety of responses. One especially interesting response is a development of the so-called ‘skill theory’, according to which there is no illusion after all. Instead, so the theory goes, we establish the required visual contact with our world by an ongoing process of active exploration, in which the world acts as a kind of reliable, interrogable, external memory (Noe, Pessoa and.

Sensory substitution devices are a type of sensory prosthesis that (typically) convert visual stimuli transduced by a camera into tactile or auditory stimulation. They are designed to be used by people with impaired vision so that they can recover some of the functions normally subserved by vision. In this chapter we will consider what philosophers might learn about the nature of the senses from the neuroscience of sensory substitution. We will show how sensory substitution devices work by exploiting the cross-modal plasticity of sensory cortex: the ability of sensory cortex to pick up some types of information about the external environment irrespective of the nature of the sensory inputs it is processing. We explore the implications of cross-modal plasticity for theories of the senses that attempt to make distinctions between the senses on the basis of neurobiology.