Online research in philosophy

Computation is interpretable symbol manipulation. Symbols are objects that are manipulated on the basis of rules operating only on theirshapes, which are arbitrary in relation to what they can be interpreted as meaning. Even if one accepts the Church/Turing Thesis that computation is unique, universal and very near omnipotent, not everything is a computer, because not everything can be given a systematic interpretation; and certainly everything can''t be givenevery systematic interpretation. But even after computers and computation have been successfully distinguished from other kinds of things, mental states will not just be the implementations of the right symbol systems, because of the symbol grounding problem: The interpretation of a symbol system is not intrinsic to the system; it is projected onto it by the interpreter. This is not true of our thoughts. We must accordingly be more than just computers. My guess is that the meanings of our symbols are grounded in the substrate of our robotic capacity to interact with that real world of objects, events and states of affairs that our symbols are systematically interpretable as being about.

This is a paperback reissue of a 1988 special issue of Cognition - dated but still of interest. The book consists of three chapters, each making one major negative point about connectionism. Fodor & Pylyshyn (F&P) argue that connectionist networks (henceforth 'nets') are not good models for cognition because they lack 'systematicity', Pinker & Price (P&P) argue that nets are not good substitutes for rule-based models of linguistic ability, and Lachter & Bever (L&B) argue that nets can only model the associative relations between cognitive structures, not the structures themselves.

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.

"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.

Harnad's main argument can be roughly summarised as follows: due to Searle's Chinese Room argument, symbol systems by themselves are insufficient to exhibit cognition, because the symbols are not grounded in the real world, hence without meaning. However, a symbol system that is connected to the real world through transducers receiving sensory data, with neural nets translating these data into sensory categories, would not be subject to the Chinese Room argument. Harnad's article is not only the starting point for the present debate, but is also a contribution to a longlasting discussion about such questions as: Can a computer think? If yes, would this be solely by virtue of its program? Is the Turing Test appropriate for deciding whether a computer thinks?

After people learn to sort objects into categories they see them differently. Members of the same category look more alike and members of different categories look more different. This phenomenon of within-category compression and between-category separation in similarity space is called categorical perception (CP). It is exhibited by human subjects, animals and neural net models. In backpropagation nets trained first to auto-associate 12 stimuli varying along a onedimensional continuum and then to sort them into 3 categories, CP arises as a natural side-effect because of four factors: (1) Maximal interstimulus separation in hidden-unit space during autoassociation learning, (2) movement toward linear separability during categorization learning, (3) inverse-distance repulsive force exerted by the between-category boundary, and (4) the modulating effects of input iconicity, especially in interpolating CP to untrained regions of the continuum. Once similarity space has been "warped" in this way, the compressed and separated "chunks" have symbolic labels which could then be combined into symbol strings that constitute propositions about objects. The meanings of such symbolic representations would be "grounded" in the system's capacity to pick out from their sensory projections the object categories that the propositions were about.

A provisional model is presented in which categorical perception (CP) provides our basic or elementary categories. In acquiring a category we learn to label or identify positive and negative instances from a sample of confusable alternatives. Two kinds of internal representation are built up in this learning by "acquaintance": (1) an iconic representation that subserves our similarity judgments and (2) an analog/digital feature-filter that picks out the invariant information allowing us to categorize the instances correctly. This second, categorical representation is associated with the category name. Category names then serve as the atomic symbols for a third representational system, the (3) symbolic representations that underlie language and that make it possible for us to learn by "description." Connectionism is one possible mechainsm for learning the sensory invariants underlying categorization and naming. Among the implications of the model are (a) the "cognitive identity of (current) indiscriminables": Categories and their representations can only be provisional and approximate, relative to the alternatives encountered to date, rather than "exact." There is also (b) no such thing as an absolute "feature," only those features that are invariant within a particular context of confusable alternatives. Contrary to prevailing "prototype" views, however, (c) such provisionally invariant features must underlie successful categorization, and must be "sufficient" (at least in the "satisficing" sense) to subserve reliable performance with all-or-none, bounded categories, as in CP. Finally, the model brings out some basic limitations of the "symbol-manipulative" approach to modeling cognition, showing how (d) symbol meanings must be functionally grounded in nonsymbolic, "shape-preserving" representations -- iconic and categorical ones. Otherwise, all symbol interpretations are ungrounded and indeterminate. This amounts to a principled call for a psychophysical (rather than a neural) "bottom-up" approach to cognition.

According to "computationalism" (Newell, 1980; Pylyshyn 1984; Dietrich 1990), mental states are computational states, so if one wishes to build a mind, one is actually looking for the right program to run on a digital computer. A computer program is a semantically interpretable formal symbol system consisting of rules for manipulating symbols on the basis of their shapes, which are arbitrary in relation to what they can be systematically interpreted as meaning. According to computationalism, every physical implementation of the right symbol system will have mental states.

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.

1.1 The predominant approach to cognitive modeling is still what has come to be called "computationalism" (Dietrich 1990, Harnad 1990b), the hypothesis that cognition is computation. The more recent rival approach is "connectionism" (Hanson & Burr 1990, McClelland & Rumelhart 1986), the hypothesis that cognition is a dynamic pattern of connections and activations in a "neural net." Are computationalism and connectionism really deeply different from one another, and if so, should they compete for cognitive hegemony, or should they collaborate? These questions will be addressed here, in the context of an obstacle that is faced by computationalism (as well as by connectionism if it is either computational or seeks cognitive hegemony on its own): The symbol grounding problem (Harnad 1990).