While I agree in general with Stevan Harnad's symbol grounding proposal, I do not believe "transduction" (or "analog process") PER SE is useful in distinguishing between what might best be described as different "degrees" of grounding and, hence, for determining whether a particular system might be capable of cognition. By 'degrees of grounding' I mean whether the effects of grounding go "all the way through" or not. Why is transduction limited in this regard? Because transduction is a physical process which (...) does not speak to the issue of representation, and, therefore, does not explain HOW the informational aspects of signals impinging on sensory surfaces become embodied as symbols or HOW those symbols subsequently cause behavior, both of which, I believe, are important to grounding and to a system's cognitive capacity. Immunity to Searle's Chinese Room (CR) argument does not ensure that a particular system is cognitive, and whether or not a particular degree of groundedness enables a system to pass the Total Turing Test (TTT) may never be determined. (shrink)

Stevan Harnad and I seem to be thinking about many of the same issues. Sometimes we agree, sometimes we don't; but I always find his reasoning refreshing, his positions sensible, and the problems with which he's concerned to be of central importance to cognitive science. His "Grounding Symbols in the Analog World with Neural Nets" (= GS) is no exception. And GS not only exemplifies Harnad's virtues, it also provides a springboard for diving into Harnad- Bringsjord terrain.

What is the relation between the material, conventional symbol structures that we encounter in the spoken and written word, and human thought? A common assumption, that structures a wide variety of otherwise competing views, is that the way in which these material, conventional symbol-structures do their work is by being translated into some kind of content-matching inner code. One alternative to this view is the tempting but thoroughly elusive idea that we somehow think in some natural language (such as English). (...) In the present treatment I explore a third option, which I shall call the "complementarity" view of language. According to this third view the actual symbol structures of a given language add cognitive value by complementing (without being replicated by) the more basic modes of operation and representation endemic to the biological brain. The "cognitive bonus" that language brings is, on this model, not to be cashed out either via the ultimately mysterious notion of "thinking in a given natural language" or via some process of exhaustive translation into another inner code. Instead, we should try to think in terms of a kind of coordination dynamics in which the forms and structures of a language qua material symbol system play a key and irreducible role. Understanding language as a complementary cognitive resource is, I argue, an important part of the much larger project (sometimes glossed in terms of the "extended mind") of understanding human cognition as essentially and multiply hybrid: as involving a complex interplay between internal biological resources and external non-biological resources. (shrink)

What language allows us to do is to "steal" categories quickly and effortlessly through hearsay instead of having to earn them the hard way, through risky and time-consuming sensorimotor "toil" (trial-and-error learning, guided by corrective feedback from the consequences of miscategorisation). To make such linguistic "theft" possible, however, some, at least, of the denoting symbols of language must first be grounded in categories that have been earned through sensorimotor toil (or else in categories that have already been "prepared" for us (...) through Darwinian theft by the genes of our ancestors); it cannot be linguistic theft all the way down. The symbols that denote categories must be grounded in the capacity to sort, label and interact with the proximal sensorimotor projections of their distal category-members in a way that coheres systematically with their semantic interpretations, both for individual symbols, and for symbols strung together to express truth-value-bearing propositions. (shrink)

"Symbol Grounding" is beginning to mean too many things to too many people. My own construal has always been simple: Cognition cannot be just computation, because computation is just the systematically interpretable manipulation of meaningless symbols, whereas the meanings of my thoughts don't depend on their interpretability or interpretation by someone else. On pain of infinite regress, then, symbol meanings must be grounded in something other than just their interpretability if they are to be candidates for what is going on (...) in our heads. Neural nets may be one way to ground the names of concrete objects and events in the capacity to categorize them (by learning the invariants in their sensorimotor projections). These grounded elementary symbols could then be combined into symbol strings expressing propositions about more abstract categories. Grounding does not equal meaning, however, and does not solve any philosophical problems. (shrink)

Connectionism and computationalism are currently vying for hegemony in cognitive modeling. At first glance the opposition seems incoherent, because connectionism is itself computational, but the form of computationalism that has been the prime candidate for encoding the "language of thought" has been symbolic computationalism (Dietrich 1990, Fodor 1975, Harnad 1990c; Newell 1980; Pylyshyn 1984), whereas connectionism is nonsymbolic (Fodor & Pylyshyn 1988, or, as some have hopefully dubbed it, "subsymbolic" Smolensky 1988). This paper will examine what is and is not (...) a symbol system. A hybrid nonsymbolic/symbolic system will be sketched in which the meanings of the symbols are grounded bottom-up in the system's capacity to discriminate and identify the objects they refer to. Neural nets are one possible mechanism for learning the invariants in the analog sensory projection on which successful categorization is based. "Categorical perception" (Harnad 1987a), in which similarity space is "warped" in the service of categorization, turns out to be exhibited by both people and nets, and may mediate the constraints exerted by the analog world of objects on the formal world of symbols. (shrink)

There has been much discussion recently about the scope and limits of purely symbolic models of the mind and about the proper role of connectionism in cognitive modeling. This paper describes the symbol grounding problem: How can the semantic interpretation of a formal symbol system be made intrinsic to the system, rather than just parasitic on the meanings in our heads? How can the meanings of the meaningless symbol tokens, manipulated solely on the basis of their (arbitrary) shapes, be grounded (...) in anything but other meaningless symbols? The problem is analogous to trying to learn Chinese from a Chinese/Chinese dictionary alone. A candidate solution is sketched: Symbolic representations must be grounded bottom-up in nonsymbolic representations of two kinds: (1) iconic representations, which are analogs of the proximal sensory projections of distal objects and events, and (2) categorical representations, which are learned and innate feature-detectors that pick out the invariant features of object and event categories from their sensory projections. Elementary symbols are the names of these object and event categories, assigned on the basis of their (nonsymbolic) categorical representations. Higher-order (3) symbolic representations, grounded in these elementary symbols, consist of symbol strings describing category membership relations (e.g., An X is a Y that is Z). Connectionism is one natural candidate for the mechanism that learns the invariant features underlying categorical representations, thereby connecting names to the proximal projections of the distal objects they stand for. In this way connectionism can be seen as a complementary component in a hybrid nonsymbolic/symbolic model of the mind, rather than a rival to purely symbolic modeling. Such a hybrid model would not have an autonomous symbolic module, however; the symbolic functions would emerge as an intrinsically dedicated symbol system as a consequence of the bottom-up grounding of categories' names in their sensory representations. Symbol manipulation would be governed not just by the arbitrary shapes of the symbol tokens, but by the nonarbitrary shapes of the icons and category invariants in which they are grounded. (shrink)

Computationalist theories of mind require brain symbols, that is, neural events that represent kinds or instances of kinds. Standard models of computation require multiple inscriptions of symbols with the same representational content. The satisfaction of two conditions makes it easy to see how this requirement is met in computers, but we have no reason to think that these conditions are satisfied in the brain. Thus, if we wish to give computationalist explanations of human cognition, without committing ourselvesa priori to a (...) strong and unsupported claim in neuroscience, we must first either explain how we can provide multiple brain symbols with the same content, or explain how we can abandon standard models of computation. It is argued that both of these alternatives require us to explain the execution of complex tasks that have a cognition-like structure. Circularity or regress are thus threatened, unless noncomputationalist principles can provide the required explanations. But in the latter case, we do not know that noncomputationalist principles might not bear most of the weight of explaining cognition. Four possible types of computationalist theory are discussed; none appears to provide a promising solution to the problem. Thus, despite known difficulties in noncomputationalist investigations, we have every reason to pursue the search for noncomputationalist principles in cognitive theory. (shrink)

This paper provides a theory of the nature of symbols in the language of thought (LOT). My discussion consists in three parts. In part one, I provide three arguments for the individuation of primitive symbols in terms of total computational role. The first of these arguments claims that Classicism requires that primitive symbols be typed in this manner; no other theory of typing will suffice. The second argument contends that without this manner of symbol individuation, there will be computational processes (...) that fail to supervene on syntax, together with the rules of composition and the computational algorithms. The third argument says that cognitive science needs a natural kind that is typed by total computational role. Otherwise, either cognitive science will be incomplete, or its laws will have counterexamples. Then, part two defends this view from a criticism, offered by both Jerry Fodor and Jesse Prinz, who respond to my view with the charge that because the types themselves are individuated. (shrink)

According to the language of thought (LOT) approach and the related computational theory of mind (CTM), thinking is the processing of symbols in an inner mental language that is distinct from any public language. Herein, I explore a deep problem at the heart of the LOT/CTM program—it has yet to provide a plausible conception of a mental symbol.

Symbols should be grounded, as has been argued before. But we insist that they should be grounded not only in subsymbolic activities, but also in the interaction between the agent and the world. The point is that concepts are not formed in isolation (from the world), in abstraction, or "objectively." They are formed in relation to the experience of agents, through their perceptual/motor apparatuses, in their world and linked to their goals and actions. This paper takes a detailed look at (...) this relatively old issue, with a new perspective, aided by our work of computational cognitive model development. To further our understanding, we also go back in time to link up with earlier philosophical theories related to this issue. The result is an account that extends from computational mechanisms to philosophical abstractions. (shrink)

This article is the second step in our research into the Symbol Grounding Problem (SGP). In a previous work, we defined the main condition that must be satisfied by any strategy in order to provide a valid solution to the SGP, namely the zero semantic commitment condition (Z condition). We then showed that all the main strategies proposed so far fail to satisfy the Z condition, although they provide several important lessons to be followed by any new proposal. Here, we (...) develop a new solution of the SGP. It is called praxical in order to stress the key role played by the interactions between the agents and their environment. It is based on a new theory of meaning—Action-based Semantics (AbS)—and on a new kind of artificial agents, called two-machine artificial agents (AM²). Thanks to their architecture, AM2s implement AbS, and this allows them to ground their symbols semantically and to develop some fairly advanced semantic abilities, including the development of semantically grounded communication and the elaboration of representations, while still respecting the Z condition. (shrink)

This paper develops a bridge from AL issues about the symbol–matter relation to AI issues about symbol-grounding by focusing on the concepts of formality and syntactic interpretability. Using the DNA triplet-amino acid specification relation as a paradigm, it is argued that syntactic properties can be grounded as high-level features of the non-syntactic interactions in a physical dynamical system. This argu- ment provides the basis for a rebuttal of John Searle’s recent assertion that syntax is observer-relative (1990, 1992). But the argument (...) as developed also challenges the classic symbol-processing theory of mind against which Searle is arguing, as well as the strong AL thesis that life is realizable in a purely computational medium. Finally, it provides a new line of support for the autonomous systems approach in AL and AI (Varela & Bourgine 1992a, 1992b). © 1997 Academic Press. (shrink)

This chapter provides the teleological foundations for our analysis of guidance to goal. Its objective is to ground goal-directedness genetically. The basic suggestion is this. Organisms are small things, with few energy resources and puny physical means, battling a ruthless physical and biological nature. How do they manage to survive and multiply? CLEVERLY, BY ORGANIZING.

In 1991, I included a brief discussion of the Baldwin effect in my account of the evolution of human consciousness, thinking I was introducing to non-specialist readers a little-appreciated, but no longer controversial, wrinkle in orthodox neo-Darwinism. I had thought that Hinton and Nowlan (1987) and Maynard Smith (1987) had shown clearly and succinctly how and why it worked, and restored the neglected concept to grace. Here is how I put it then.

Darwin differs from Newton and Einstein in that his ideas do not require a complicated or deep mind to understand them, and perhaps did not even require such a mind in order to generate them in the first place. It can be explained to any school-child (as Newtonian mechanics and Einsteinian relativity cannot) that living creatures are just Darwinian survival/reproduction machines. They have whatever structure they have through a combination of chance and its consequences: Chance causes changes in the genetic (...) blueprint from which organisms' bodies are built, and if those changes are more successful in helping their owners survive and reproduce than their predecessors or their rivals, then, by definition, those changes are reproduced, and thereby become more prevalent in succeeding generations: Whatever survives/reproduces better survives/reproduces better. That is the tautological force that shaped us. (shrink)

According to an influential view in contemporary cognitive science, many human cognitive capacities are innate. The primary support for this view comes from ‘poverty of stimulus’ arguments. In general outline, such arguments contrast the meagre informational input to cognitive development with its rich informational output. Consider the ease with which humans acquire languages, become facile at attributing psychological states (‘folk psychology’), gain knowledge of biological kinds (‘folk biology’), or come to understand basic physical processes (‘folk physics’). In all these cases, (...) the evidence available to a growing child is far too thin and noisy for it to be plausible that the underlying principles involved are derived from general learning mechanisms. This only alternative hypothesis seems to be that the child’s grasp of these principles is innate. (Cf. Laurence and Margolis, 2001.). (shrink)

We examine the verification of simple quantifiers in natural language from a computational model perspective. We refer to previous neuropsychological investigations of the same problem and suggest extending their experimental setting. Moreover, we give some direct empirical evidence linking computational complexity predictions with cognitive reality. In the empirical study we compare time needed for understanding different types of quantifiers. We show that the computational distinction between quantifiers recognized by finite-automata and push-down automata is psychologically relevant. Our research improves upon hypothesis (...) and explanatory power of recent neuroimaging studies as well as provides evidence. (shrink)

We compare time needed for understanding different types of quantifiers. We show that the computational distinction between quantifiers recognized by finite-automata and pushdown automata is psychologically relevant. Our research improves upon hypothesis and explanatory power of recent neuroimaging studies as well as provides evidence for the claim that human linguistic abilities are constrained by computational complexity.

Situation theory has been developed over the last decade and various versions of the theory have been applied to a number of linguistic issues. However, not much work has been done in regard to its computational aspects. In this paper, we review the existing approaches towards `computational situation theory' with considerable emphasis on our own research.

This paper continues a strain of intellectual complaint against the presumptions of certain kinds of formal semantics (the qualification is important) and their bad effects on those areas of artificial intelligence concerned with machine understanding of human language. After some discussion of the use of the term epistemology in artificial intelligence, the paper takes as a case study the various positions held by McDermott on these issues and concludes, reluctantly, that, although he has reversed himself on the issue, there was (...) no time at which he was right. (shrink)

The situated cognition movement has emerged in recent decades (although it has roots in psychologists working earlier in the 20^th century including Vygotsky, Bartlett, and Dewey) largely in reaction to an approach to explaining cognition that tended to ignore the context in which cognitive activities typically occur. Fodor’s (1980) account of the research strategy of methodological solipsism, according to which only representational states within the mind are viewed as playing causal roles in producing cognitive activity, is an extreme characterization of (...) this approach. (As Keith Gunderson memorably commented when Fodor first presented this characterization, it amounts to reversing behaviorism by construing the mind as a white box in a black world). Critics as far back as the 1970s and 1980s objected to many experimental paradigms in cognitive psychology as not being ecologically valid; that is, they maintained that the findings only applied to the artificial circumstances created in the laboratory and did not generalize to real world settings (Neisser, 1976; 1987). The situated cognition movement, however, goes much further than demanding ecologically valid experiments—it insists that an agent’s cognitive activities are inherently embedded and supported by dynamic interactions with the agent’s body and features of its environment. (shrink)

Cognitive science, as an interdisciplinary school of thought, may have recently moved beyond the bandwagon stage onto the throne of orthodoxy, but it does not make a favorable first impression on many people. Familiar reactions on first encounters range from revulsion to condescending dismissal--very few faces in the crowd light up with the sense of "Aha! So that's how the mind works! Of course!" Cognitive science leaves something out, it seems; moreover, what it apparently leaves out is important, even precious. (...) Boiled down to its essence, cognitive science proclaims that in one way or another our minds are computers, and this seems so mechanistic, reductionistic, intellectualistic, dry, philistine, unbiological. It leaves out emotion, or what philosophers call qualia, or value, or mattering, or . . . the soul. It doesn't explain what minds are so much as attempt to explain minds away. (shrink)

Much of traditional AI exemplifies the explicit representation paradigm, and during the late 1980''s a heated debate arose between the classical and connectionist camps as to whether beliefs and rules receive an explicit or implicit representation in human cognition. In a recent paper, Kirsh (1990) questions the coherence of the fundamental distinction underlying this debate. He argues that our basic intuitions concerning explicit and implicit representations are not only confused but inconsistent. Ultimately, Kirsh proposes a new formulation of the distinction, (...) based upon the criterion ofconstant time processing.The present paper examines Kirsh''s claims. It is argued that Kirsh fails to demonstrate that our usage of explicit and implicit is seriously confused or inconsistent. Furthermore, it is argued that Kirsh''s new formulation of the explicit-implicit distinction is excessively stringent, in that it banishes virtually all sentences of natural language from the realm of explicit representation. By contrast, the present paper proposes definitions for explicit and implicit which preserve most of our strong intuitions concerning straightforward uses of these terms. It is also argued that the distinction delineated here sustains the meaningfulness of the abovementioned debate between classicists and connectionists. (shrink)

At present, the prevailing Connectionist methodology forrepresenting rules is toimplicitly embody rules in neurally-wired networks. That is, the methodology adopts the stance that rules must either be hard-wired or trained into neural structures, rather than represented via explicit symbolic structures. Even recent attempts to implementproduction systems within connectionist networks have assumed that condition-action rules (or rule schema) are to be embodied in thestructure of individual networks. Such networks must be grown or trained over a significant span of time. However, arguments (...) are presented herein that humanssometimes follow rules which arevery rapidly assignedexplicit internal representations, and that humans possessgeneral mechanisms capable of interpreting and following such rules. In particular, arguments are presented that thespeed with which humans are able to follow rules ofnovel structure demonstrates the existence of general-purpose rule following mechanisms. It is further argued that the existence of general-purpose rule following mechanisms strongly indicates that explicit rule following is not anisolated phenomenon, but may well be a common and important aspect of cognition. The relationship of the foregoing conclusions to Smolensky''s view of explicit rule following is also explored. The arguments presented here are pragmatic in nature, and are contrasted with thekind of arguments developed by Fodor and Pylyshyn in their recent, influential paper. (shrink)

Explicitness has usually been approached from two points of view, labelled by Kirsh the structural and the process view, that hold opposite assumptions to determine when information is explicit. In this paper, we offer an intermediate view that retains intuitions from both of them. We establish three conditions for explicit information that preserve a structural requirement, and a notion of explicitness as a continuous dimension. A problem with the former accounts was their disconnection with psychological work on the issue. We (...) review studies by Karmiloff-Smith, and Shanks and St. John to show that the proposed conditions have psychological grounds. Finally, we examine the problem of explicit rules in connectionist systems in the light of our framework. (shrink)

Unifying traditional cognitive science is the idea that thinking is a process of symbol manipulation, where symbols lead both a syntactic and a semantic life. The syntax of a symbol comprises those properties in virtue of which the symbol undergoes rule-dictated transformations. The semantics of a symbol constitute the symbolsÕ meaning or representational content. Thought consists in the syntactically determined manipulation of symbols, but in a way that respects their semantics. Thus, for instance, a calculating computer sensitive only to the (...) shape of symbols might produce the symbol Ô5Õ in response to the inputs Ô2Õ, Ô+Õ, and Ô3Õ. As far as the computer is concerned, these symbols have no meaning, but because of its program it will produce outputs that, to the user, Òmake senseÓ given the meanings the user attributes to the symbols. (shrink)

I examine whether it is possible for content relevant to a computer''s behavior to be carried without an explicit internal representation. I consider three approaches. First, an example of a chess playing computer carrying emergent content is offered from Dennett. Next I examine Cummins response to this example. Cummins says Dennett''s computer executes a rule which is inexplicitly represented. Cummins describes a process wherein a computer interprets explicit rules in its program, implements them to form a chess-playing device, then this (...) device executes the rules in a way that exhibits them inexplicitly. Though this approach is intriguing, I argue that the chess-playing device cannot exist as imagined. The processes of interpretation and implementation produce explicit representations of the content claimed to be inexplicit. Finally, the Chinese Room argument is examined and shown not to save the notion of inexplicit information. This means the strategy of attributing inexplicit content to a computer which is executing a rule, fails. (shrink)

The self-advertising, at least, suggests that 'situated cognition' involves the most fundamental conceptual re-organization in AI and cognitive science, even appearing to deny that cognition is to be explained by mental representations. In their defence of the orthodox symbolic representational theory, A. Vera and H. Simon (1993) have rebutted many of these claims, but they overlook an important reading of situated arguments which may, after all, involve a revolutionary insight. I show that the whole debate turns on puzzles familiar from (...) the history of philosophy and psychology and these may serve to clarify the current disputes. (shrink)

Does self?knowledge help? A rationalist, presumably, thinks that it does: both that self?knowledge is possible and that, if gained through appropriate channels, it is desirable. Descartes notoriously claimed that, with appropriate methods of enquiry, each of his readers could become an expert on herself or himself. As well as the direct, first?person knowledge of self to which we are led in the Meditationes , we can also seek knowledge of our own bodies, and of the union of our minds and (...) our bodies: the latter forms of self?knowledge are inevitably imperfect, but are no less important in guiding our conduct in the search after truth. (shrink)

Humans have a folk psychology, without question. Paul Churchland used the term to describe “our commonsense conception of psychological phenomena” (Churchland 1981, p. 67), whatever that may be. When we ask the question whether animals have their own folk psychology, we’re asking whether any other species has a commonsense conception of psychological phenomenon as well. Different versions of this question have been discussed over the past 25 years, but no clear answer has emerged. Perhaps one reason for this lack of (...) progress is that we don’t clearly understand the question. In asking whether animals have folk psychology, I hope to help clarify the concept of folk psychology itself, and in the process, to gain a greater understanding of the role of belief and desire attribution in human social interaction. (shrink)

Artificial intelligence research has foundered on the issue of representation. When intelligence is approached in an incremental manner, with strict reliance on interfacing to the real world through perception and action, reliance on representation disappears. In this paper we outline our approach to incrementally building complete intelligent Creatures. The fundamental decomposition of the intelligent system is not into independent information processing units which must interface with each other via representations. Instead, the intelligent system is decomposed into independent and parallel activity (...) producers which all interface directly to the world through perception and action, rather than interface to each other particularly much. The notions of central and peripheral systems evaporateeverything is both central and peripheral. Based on these principles we have built a very successful series of mobile robots which operate without supervision as Creatures in standard office environments. (shrink)

Connectionism and classicism, it generally appears, have at least this much in common: both place some notion of internal representation at the heart of a scientific study of mind. In recent years, however, a much more radical view has gained increasing popularity. This view calls into question the commitment to internal representation itself. More strikingly still, this new wave of anti-representationalism is rooted not in armchair theorizing but in practical attempts to model and understand intelligent, adaptive behavior. In this paper (...) we first present, and then critically assess, a variety of recent anti-representationalist treatments. We suggest that so far, at least, the sceptical rhetoric outpaces both evidence and argument. Some probable causes of this premature scepticism are isolated. Nonetheless, the anti-representationalist challenge is shown to be both important and progressive insofar as it forces us to see beyond the bare representational/non-representational dichotomy and to recognize instead a rich continuum of degrees and types of representationality. (shrink)

The field of Artificial Intelligence has produced so many new concepts--or at least vivid and more structured versions of old concepts--that it would be surprising if none of them turned out to be of value to students of animal behavior. Which will be most valuable? I will resist the temptation to engage in either prophecy or salesmanship; instead of attempting to answer the question: "How might Artificial Intelligence inform the study of animal behavior?" I will concentrate on the obverse: "How (...) might the study of animal behavior inform research in Artificial Intelligence?". (shrink)

A discussion is going on in cognitive science about the use of representations to explain how intelligent behavior is generated. In the traditional view, an organism is thought to incorporate representations. These provide an internal model that is used by the organism to instruct the motor apparatus so that the adaptive and anticipatory characteristics of behavior come about. So-called interactionists claim that this representational specification of behavior raises more problems than it solves. In their view, the notion of internal representational (...) models is to be dispensed with. Instead, behavior is to be explained as the intricate interaction between an embodied organism and the specific make up of an environment. The problem with a non-representational interactive account is that it has severe difficulties with anticipatory, future oriented behavior. The present paper extends the interactionist conceptual framework by drawing on ideas derived from the study of morphogenesis. This extended interactionist framework is based on an analysis of anticipatory behavior as a process which involves multiple spatio-temporal scales of neural, bodily and environmental dynamics. This extended conceptual framework provides the outlines for an explanation of anticipatory behavior without involving a representational specification of future goal states. (shrink)

A startling amount of intelligent activity can be controlled without reasoning or thought. By tuning the perceptual system to task relevant properties a creature can cope with relatively sophisticated environments without concepts. There is a limit, however, to how far a creature without concepts can go. Rod Brooks, like many ecologically oriented scientists, argues that the vast majority of intelligent behaviour is concept-free. To evaluate this position I consider what special benefits accrue to concept-using creatures. Concepts are either necessary for (...) certain types of perception, learning, and control, or they make those processes computationally simpler. Once a creature has concepts its capacities are vastly multiplied. (shrink)