Commentary on Tsuda
Abstract: 60 words
Main Text: 997 words
References: 147 words
Total Text: 1282 words
The relevance of chaotic itinerancy and other types of exotic dyamical behavior described by Tsuda certainly goes beyond the scope of his target article. These concepts of dynamics may offer a general framework for the understanding of complexity, which could help to restructure the analysis and conceptualization of mental states in novel ways, providing insights for the philosophy of mind.
Most philosophical treatments of the classical concepts of cognition (such as volition, consciousness, intentionality, mental content, rationality) seem to be trapped in an unsafe argument, which goes, by and large, like this. There are two opposing alternatives, naturalism and anti-naturalism, both with unwanted consequences. It is generally accepted that to be a naturalist is to be a physicalist and/or functionalist and to believe in the computational mind and/or physical (computational) brain. To be a non-naturalist is (as Dennett so aptly puts it in a somewhat different context) to play tennis with the net down. If one is a non-naturalist, then one is free from the restrictions of science, and it is easy to justify any concept, from intrinsic intentionality to wonder tissue to qualia, without having to worry how these fit into a description of the universe as a whole. There are other points of view existing somewhere between these two extremes, such as double-aspect theories, where the physical and mental aspects of material are considered distinct, but the manner in which they are distinguished is not addressed. However, this again internalizes the naturalist/non-naturalist dichotomy. So where do we stand? Presently few people are completely comfortable with naturalism-cum-physicalism in a strict form; functionalism is in the defence; accepting the small wonders offered by non-naturalism is not the approach of choice. Are there other options? Maybe there are
The above argument goes too fast. There are several hidden assumptions that underlie it, which together implicitly ensure a harmony of naturalism and reductionism (or eliminativism), or a unity of naturalism and computationalism. At least two of the assumptions are easy to identify. The first is that the mind is confined to a single level of phenomena (and therefore, that all mental events can be treated winthin a single explanatory scheme). The second is that mental states are homogeneous (and therefore, that they are all similar, constituting a natural kind).
One kind of thing makes up the mind, and one functional theory explains how it works: cognitive science has traditionally held this. However, Ryle (1949), perhaps Haugeland (1995), and more recently van Gelder (1998) laid the groundwork for a fundamental challenge to the concept of a homogeneous mind. Their concern is with the difference between mind and cognition, but their idea that the mind is somehow not one thing is far more general. Van Gelder makes a significant move when he combines this philosophical idea of ontological heterogeneity with low-level functional models to form what is known as the "dynamical hypothesis" in cognition (Port 2001). This now allows for a new kind of continuation, through which the other pillar of the traditonally posited unity of the mind, that of a single functional scheme, might also be discredited.
This is the point where dynamical systems of the kind studied by Tsuda enter. We can view the dynamical hypothesis as a step in a progression towards a greater flexibility of the mind: the progression begins with the rigid symbolic systems of "classical" (Fodor-style) cognitivism, and continues through connectionism to dynamical systems, and possibly beyond.
Previous attempts to use dynamical systems as models of cognition have been unsuccessful. Ten or fifteen years ago the study of complex systems consisted of only chaos, bifurcation and self-organization, phenomena that have a technical meaning in hydrodynamics, basic-level brain theory and the like, but apparently are not appropriate in the description of higher cognition. They are hopelessly far from anything in which the philosopher or psychologist is interested. How can we elucidate mental content, for instance? Early-day cybernetics, based on simple dynamical systems, had nothing to offer in this regard, and this is one reason why cybernetics has been ignored in cognitive science. Chaos, although representing a richer class of phenomena, is no different on this score. It appears to possess physiological relevance for the brain (see Freeman (1991), which is now a classic), but it is difficult to see what could be gained by assuming that mental computations are chaotic - computations are just computations. Also, chaos is, by itself, uninteresting: in a chaotic system, most things happen on the attractor, where long-term behavior always remains qualitatively the same.
Chaotic itinerancy (CI), however, is fundamentally different, and I believe it may change our thinking about what can be achieved by dynamical systems and minds. A CI is a chameleon. It can behave like one particular type of system for some time, and then make a transition and behave like some other particular system, and so on, indefinitely. Although generated by a single set of equations, at a phenomenological level the process does not seem to follow any fixed rules. Yet it can be controlled, or at least this is a feasible expectation. This may make CI an ideal tool for the understanding of mental processes that form and break their own rules. Part of this anticipated development is visible in Tsuda's memory model already.
The problem of conceptualizing complexity is another matter. Complex dynamical systems of the type studied by Tsuda and some other authors, including Kaneko (1998)(see also Kaneko and Tsuda 2001), realize a wide spectrum of behaviors. This has far-ranging philosophical implications. Putnam (1988) may be right, after all, in his claim that every natural system can behave like any other. With Tsuda's works we start to understand that for a complex system every model can be justified. What is more, perhaps every physical system is capable of exhibiting chaotic itinerancies and other similarly rich processes. If this is true, every physical system may literally be able to realize every computational behaviour. Opponents of Putnam's idea (such as Chalmers, 1996) tend to assume that a system is basically something perfectly well-defined, and therefore, that the construction of a single system that can behave like many different systems must be seen as a dirty trick. CIs can realize such a situation quite naturally. This is a further example where the benefits of an advanced analysis made possible by the new paradigm of complex systems becomes apparent.
Chalmers, D. (1996). Does a rock implement every finite automaton? Synthese
, 108, 309-33.
http://www.u.arizona.edu/~chalmers/papers/rock.html
Freeman, W.J. (1991). The physiology of perception. Scientific American
, 264, 78-85.
http://sulcus.berkeley.edu/FreemanWWW/manuscripts/IE1/91.html
Haugeland, J. (1995). Mind embodied and embedded. Acta Philosophica
Fennica , 58, 233-67.
Kaneko, K. (1998). On the strength of attractors in a high-dimensional
system: milnor attractor network, robust global attraction, and noise-induced
selection. Physica D, 124, 308-30.
Kaneko, K. & Tsuda, I. (2001). Complex systems: chaos and beyond
– a constructive approach with applications in life sciences. New York:
Springer.
Port, R.F. (2001). Dynamical systems hypothesis in Cognitive Science. In:
Encyclopedia of Cognitive Science. Assoc. Ed. Amy Lockyer. London:
MacMillan, in print.
http://www.cs.indiana.edu/hyplan/port/pap/dynamic.cognition.sglspc.htm
Putnam, H. (1988). Representation and Reality. Cambridge, MA: MIT
Press.
Ryle, G. (1949). The Concept of Mind. London: Hutchinson & Co.
Van Gelder, T. J. (1998): The dynamical hypothesis in cognitive science.
Behavioral and Brain Sciences, 21, 1-14.
http://www.cogsci.soton.ac.uk/bbs/Archive/bbs.vangelder.html
The author wishes to thank the Fujitsu Company and Professor Susumu Kunifuji
for their support that made possible the preparation of this manuscript.