Andrew Clark

Much work in economics, the social sciences, and elsewhere takes as its starting point a somewhat unrealistic conception of rationality — a conception that ignores or downplays both the temporal and the situated aspects of human reason. Biological reason, I shall argue, is better conceived as an iterated process of adaptive response made under extreme time pressure and exquisitely keyed to a variety of external structures and circumstances. These external structures and circumstances act as filters and constraints on the spaces of possible real-time responses. Paramount among such structures and circumstances, in the case of human reason, are the cultural artifacts of language and of social and economic institutions. Models of rational decision making need to situate the reasoning agent as just one element in a complex and time-sensitive feedback system in which such external structures play a major role. It is therefore crucial that we understand the complex and mutually modulatory interplay between individual cognition and the extended environmental loops in which it participates. I shall explore a few potential avenues for developing such an understanding including neural network research and multiple time scale..

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 specialpurpose 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.

Kleinberg describes a novel procedure for efficient search in a dense hyper-linked environment, such as the world wide web. The procedure exploits information implicit in the links between pages so as to identify patterns of connectivity indicative of “authorative sources”. At a more general level, the trick is to use this second-order link-structure information to rapidly and cheaply identify the knowledge-structures most likely to be relevant given a specific input. I shall argue that Kleinberg’s procedure is suggestive of a new, viable, and neuroscientifically plausible solution to at least the so-called “Frame Problem” in cognitive science viz the problem of explaining global abductive inference. More accurately, I shall argue that Kleinberg’s procedure suggests a new variety of “fast and frugal heuristic” ) capable of pressing maximum utility from the vast bodies of information and associations commanded by the biological brain. The paper thus takes up the challenge laid down by Fodor ). Fodor depicts the problem of global knowledge-based reason as the point source of many paradigmatic failings of contemporary computational theories of mind. These failings, Fodor goes on to argue, cannot be remedied by any simple appeal to alternative modes of encoding and processing. I shall show, however, that connectionist models can provide for one neurologically plausible incarnation of Kleinberg’s procedure. The paper ends by noting that current commercial applications increasingly confront the kinds of challenge initially posed to biological thought and reason.

The image of the presence of symbols in an inner code pervades recent debates in cognitive science. Classicists worship in the presence. Connectionists revel in the absence. However, the very ideas of code and symbol are ill understood. A major distorting factor in the debates concerns the role of processing in determining the presence or absence of a stuctured inner code. Drawing on work by David Kirsh and David Chambers, the present paper attempts to re-define such notions to begin to reflect the inextrictability of code and presence

Fodor’s theory makes thinking prior to doing. It allows for an inactive agent or pure reflector, and for agents whose actions in various ways seem to float free of their own conceptual repertoires. We show that naturally evolved creatures are not like that. In the real world, thinking is always and everywhere about doing. The point of having a brain is to guide the actions of embodied beings in a complex material world. Some of those actions are, to be sure, more recondite than others. But in every case the contents of thoughts still look to depend, in some non-unique but vitally important way, on the kinds of doings they support.

It is widely appreciated that the difficulty of a particluar computation varies according to how the input data are presented. What is less understood is the effect of this computation/representation tradeoff within familiar learning paradigms. We argue that existing learning algoritms are often poorly equipped to solve problems involving a certain type of important and widespread regularity, which we call 'type-2' regularity. The solution in these cases is to trade achieved representation against computational search. We investigate several ways in which such a trade-off may be pursued including simple incremental learning, modular connectionism, and the developmental hypothesis of 'representational redescription'. In addition, the most distinctive features of human cognition- language and culture- may themselves be viewed as adaptions enabling this representation/computation trade-off to be pursued on an even grander scale.

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 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: “YOU know? What do YOU know. YOU don’t know anything. In 10 minutes time YOU won’t even know you had this conversation” Leonard, however, believes that he does, day by day, come to know new things. But only courtesy of those photos, tattoos, tricks and ploys. Who is right?These are the kinds of question addressed at length in the paper ‘The Extended Mind’. Is the mind contained in the head? Or does the notion of thought allow mental processes to inhere in extended systems of body, brain and aspects of the local environment? The answer, we claimed, was that mental states, including states of believing, could be grounded in physical traces that remained firmly outside the head. As long as a few simple conditions were met, Leonard’s notes and tattoos could indeed count as new additions to his store of long-term knowledge and dispositional belief. In the present treatment I revisit this argument, defending our strong conclusion against a variety of subsequent observations and objections. In particular, I look at objections that rely on a contrast between the intrinsic content of neural symbols and the merely derived content of external inscriptions, at objections concerning the demarcation of scientific domains via natural kinds, and at objections concerning the ultimate locus of agentive control and the nature of perception versus introspection. I also mention a possible alternative interpretation of the argument as a reductio of the very idea of the mind as an object of scientific study. This is an interesting proposal

Markets, companies and various forms of business organizations may all 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.I begin then, by clarifying the nature of the problem. What is it about complex adaptive systems that makes control and intervention so difficult? I then pursue a biological example: the role of genetic factors in influencing mature form. This example yields some lessons that can be applied to business organizations. Section 4 pursues a speculative extension of these lessons, applying them to the task of understanding the role of tagging and explicit models in certain kinds of advanced system. There is a brief conclusion.

Adams and Aizawa, in a series of recent and forthcoming papers ((2001), (In Press), (This Volume)) seek to refute, or perhaps merely to terminally embarrass, the friends of the extended mind. One such paper begins with the following illustration: "Question: Why did the pencil think that 2+2=4? Clark's Answer: Because it was coupled to the mathematician" Adams and Aizawa (this volume) ms p.1 "That" the authors continue "about sums up what is wrong with Clark's extended mind hypothesis". The example of the pencil, they suggest, is just an especially egregious version of a fallacy said to pervade the literature on the extended mind. This fallacy, which they usefully dub the "coupling-constitution fallacy", is attributed , in varying degrees and manners, to Van Gelder and Port (1995), Clark and Chalmers (1998), Haugeland (1998), Dennett (2000), Clark (2001), Gibbs (2001), and Wilson (2004). The fallacy, of course, is to move from the causal coupling of some object or process to some cognitive agent, to the conclusion that the object or process is part of the cognitive agent , or part of the agent's cognitive processing (see e.g. Adams and Aizawa (This volume) ms p.2). Proponents of the extended mind and related theses, Adams and Aizawa repeatedly assert, are prone to this fallacy in part because they either ignore or fail to properly appreciate the importance of " the mark of the cognitive" viz the importance of an account of "what makes something a cognitive agent" (op cit ms p.3). The positive part of Adams and Aizawa's critique then emerges as a combination of the assertion that this "mark of the cognitive" involves the idea that "cognition is constituted by certain sorts of causal process that involve non-derived contents" (e.g. op cit ms p.3) and that these processes look to be characterized by psychological laws that turn out to apply to many internal goings-on but not currently (as a matter of contingent empirical fact) to any processes that take place in non-biological tools and artifacts. In what follows, I shall try to show why these arguments display nothing so much as mutual failures of communication: crossed wires concealing a couple of real, important, but much more subterranean, disagreements. In particular, I try to show why the negative considerations advanced by Adams and Aizawa fail to successfully undermine the argument for the extended mind, and why their more radical positive story, unless supplemented by implausible additional claims, fails to cast doubt on the claim that minds like ours can (without the need for any radically new techniques, technologies or interventions) extend into the world

The first question concerns a fundamental assumption of most researchers who theorize about the brain. Do neural systems exploit classical compositional and systematic representations, distributed representations, or no representations at all? The question is not easily answered. Connectionism, for example, has been criticised for both holding and challenging representational views. The second quesútion concerns the crucial methodological issue of how results emerging from the various brain sciences can help to constrain cognitive scientific models. Finally, the third question focuses attention on a major challenge to contemporary cognitive science: the challenge of understanding the mind as a controller of embodied and environmentally embedded action.

Beer’s (2003) paper is a tour de force of detailed comments on the more general notion of “situated- dynamical modeling, and provides a concrete sample ness”, Beer suggests that “on this view, situated action of the kinds of understanding dynamicists may realis- is the fundamental concern and cognition is … one tically hope to achieve. The analysis is thus, as Beer resource among many that can be brought to bear as an states, a “tool for building intuition”, and in this it suc- agent encounters its world”. In the worked example of ceeds brilliantly. But it is also an attempt to show that an agent that approaches circles and avoids diamonds, dynamical approaches can get a foothold in the expla- we see direct evidence of this in the claim that a certain nation of “minimally cognitive behaviors”; that is to three-dimensional projection provides a potent analytic say, behaviors that seem, on the surface at least, good tool. For this projection happens to be one that involves contenders for more traditional forms of problem de- one environmental state variable (vertical object posi- composition and analysis. In these brief comments, I tion), one body state variable (horizontal position rela- want to focus on one important question that I think tive to the object) and one neuronal state variable remains unanswered, and that bears rather directly on (output of interneuron 9). By dispensing with talk of this enterprise of “scaling up”. representations and their contents, and restricting the The question concerns the notion of an integrated depiction of the inner realm to a depiction of inner state dynamical explanation itself. The point about such alone (section 9.3), the dynamicist makes it easy to..

The last ten years have seen an increasing interest, within cognitive science, in issues concerning the physical body, the local environment, and the complex interplay between neural systems and the wider world in which they function. “Physically embodied, environmentally embedded” approaches thus loom large on the contemporary cognitive scientific scene. Yet many unanswered questions remain, and the shape of a genuinely embodied, embedded science of the mind is still unclear. I begin by sketching a few examples of the approach, and then raise a variety of critical questions concerning its nature and scope. A distinction is drawn between two kinds of appeal to embodiment: ‘simple’ cases, in which bodily and environmental properties merely constrain accounts that retain the focus on inner organization and processing, and more radical appeals, in which attention to bodily and environmental features is meant to transform both the subject matter and the theoretical framework of cognitive science.

1. Throughout the paper, and especially in the section called "LISP vs. DST", I worried that there was not enough focus on EXPLANATION. For the real question, it seems to me, is not whether some dynamical system can implement human cognition, but whether the dynamical description of the system is more explanatorily potent than a computational/representational one. Thus we know, for example, that a purely physical specification can fix a system capable of computing any LISP function. But from this it doesn't follow that the physical description is the one we need to understand the power of the system considered as an information processing device. In the same way, I don't think your demonstration that bifurcating attractor sets can yield the same behavior as a LISP program goes any way towards showing that we should not PREFER the LISP description. To reduce symbolic stories to a subset of DST (as hinted in that section) requires MORE than showing this kind of equivalence: it requires showing that there is explanatory gain, or at the very least, no explanatory loss, at that level. I append an extract from a recent paper of mine that touches on these issues, in case it helps clarify what I am after here

Of course, words aren’t magic. Neither are sextants, compasses, maps, slide rules and all the other paraphenelia which have accreted around the basic biological brains of homo sapiens. In the case of these other tools and props, however, it is transparently clear that they function so as to either carry out or to facilitate computational operations important to various human projects. The slide rule transforms complex mathematical problems (ones that would baffle or tax the unaided subject) into simple tasks of perceptual recognition. The map provides geographical information in a format well-suited to aid complex planning and strategic military operations. The compass gathers and displays a kind of information that (most) unaided human subjects do not seem to command. These various tools and props thus act to generate information, or to store it, or to transform it, or some combination of the three. In so doing, they impact our individual and collective problem- solving capacities in much the same dramatic ways as various software packages impact the performance of a simple pc.

There is a definite challenge in the air regarding the pivotal notion of internal representation. This challenge is explicit in, e.g., van Gelder, 1995; Beer, 1995; Thelen & Smith, 1994; Wheeler, 1994; and elsewhere. We think it is a challenge that can be met and that (importantly) can be met by arguing from within a general framework that accepts many of the basic premises of the work (in new robotics and in dynamical systems theory) that motivates such scepticism in the first place. Our strategy will be as follows. We begin (Section 1) by offering an account (an example and something close to a definition) of what we shall term Minimal Robust Representationalism (MRR). Sections 2 & 3 address some likely worries and questions about this notion. We end (Section 4) by making explicit the conditions under which, on our account, a science (e.g., robot- ics) may claim to be addressing cognitive phenomena

Kleinberg (1999) describes a novel procedure for efficient search in a dense hyper-linked environment, such as the world wide web. The procedure exploits information implicit in the links between pages so as to identify patterns of connectivity indicative of “authorative sources”. At a more general level, the trick is to use this second-order link-structure information to rapidly and cheaply identify the knowledge- structures most likely to be relevant given a specific input. I shall argue that Kleinberg’s procedure is suggestive of a new, viable, and neuroscientifically plausible solution to at least (one incarnation of) the so-called “Frame Problem” in cognitive science viz the problem of explaining global abductive inference. More accurately, I shall argue that.

In _Action in Perception _Alva No develops and presents a sensorimotor account of vision and of visual consciousness. According to such an account seeing (and indeed perceiving more generally) is analysed as a kind of skilful bodily activity. Such a view is consistent with the emerging emphasis, in both philosophy and cognitive science, on the critical role of embodiment in the construction of intelligent agency. I shall argue, however, that the full sensorimotor model faces three important challenges. The first is to negotiate a path between two prima facie unsatisfactory readings of the central claim that conscious perceptual experience is constituted by knowledge of patterns of sensorimotor dependence. The second is to convince us that the sensorimotor contribution, in such cases, is actually constitutive of perceptual experience rather than merely causally implicated in the origination of such experience.² And the third is to respond to the important challenge raised by what I will dub 'sensorimotor summarizing' models of the relation between conscious experience and richly detailed sensorimotor routines. According to such models³ conscious perceptual experience only rather indirectly reflects the rich detail of our actual sensorimotor engagements, which are instead lightly sampled as a coarse guide, optimized for planning and reasoning, and geared and filtered according to current needs and purposes.

Recent work in "embodied, embedded" cognitive science links mental contents to large-scale distributed effects: dynamic patterns implicating elements of (what are traditionally seen as) sensing, reasoning and acting. Central to this approach is an idea of biological cognition as profoundly "action-oriented" - geared not to the creation of rich, passive inner models of the world, but to the cheap and efficient production of real-world action in real-world context. A case in point is Hurley's (1998) account of the profound role of motor output in fixing the contents of conscious visual awareness - an account that also emphasizes distributed vehicles and long-range dynamical loops. Such stories can seem dramatically opposed to accounts, such as Milner and Goodale (1995), that stress relatively local mechanisms and that posit firm divisions between processes of visual awareness and of visuomotor action. But such accounts, I argue, can be deeply complimentary and together illustrate an important lesson. The lesson is that cognition may be embodied and action-oriented in two distinct - but complimentary - ways. There is a way of being embodied and action-oriented that implies being closely geared to the fine-grained control of low level effectors (hands, arms, legs and so on). And there is a way of being embodied and action-oriented that implies being closely geared to gross motor intentions, current goals, and schematic motor plans. Human cognition, I suggest, is embodied and action-oriented in both these ways. But the neural systems involved, and the size and scope of the key dynamic loops, may be quite different in each case

[opening paragraph]: Nicholas Humphrey offers a refreshingly progressive recipe for laying wide the doors of sensation: for understanding the peculiar features of qualitative or sensational experience in terms of the physical or functional facts about brains, bodies and environments. The key move in the treatment is the promotion of a kind of co- ordinated, double-sided tweaking: a careful restatement, with some amendments, of each side of the elusive identity statement ‘sensational property x = brain state y'. Only after such restatements herd the two kinds of facts into a roughly common arena, Humphrey believes, can some kind of identity story be revealed as coherent, plausible, and explanatorily potent