Department of Philosophy
University of California at Berkeley
CA 94720-3200 USA
This is a commentary written by Walter Freeman and Hubert Dreyfus at Berkeley, on the two-part multibook review published by John Searle, in the New York Review of Books 2-18 November, entitled "The Mystery of Consciousness". The Editor saw fit not to publish the commentary, despite Searle's request that he do so, with Searle's response. We are making it available on e-mail, with the hope that John will respond to it likewise.
Editors, New York Review submitted 30 November 1995 250 West 57th Street revised 30 December 1995 New York NY 10107 submitted to B&BS 15 February 1995
John Searle has selected an anomalous array of books to review, through which to address "The Mystery of Consciousness" (NYR 2-18 Nov 1995), in that none of the authors was trained as an experimental neurobiologist. Searle praises the strengths of the books and of the author's contribution to their fields (physics, cognitive science, immunology and molecular biology), then takes each to task for failing to answer his core question: How do neurons cause consciousness? He concludes that no one now has a valid answer, but that, when it comes, it will come from neurobiologists. We write to re-phrase that question, from a neurobiological point of view.
There is a severe limitation in Searle's formulation of the "mystery of consciousness", which may preclude its solvability in the terms in which he presents it. Searle and his authors believe that objects in the environment cause selected receptors to fire, which in turn cause neurons to fire. When the firing of a single neuron first became accessible with a microelectrode, neurobiologists gave a neutral description of its "receptor field", meaning the stimulus configuration that served to drive the firing maximally. Jerry Lettvin, Horace Barlow, and others, however, transformed the interpretation by specifying a property, such as a line, color or tone, which they concluded was represented by the firing of the neuron. That is, the neuron is not merely a dynamical element; it is a symbol generator.
This interpretation has led to the intractable "binding problem": how are representations of features combined so as to form representations of objects, and how are they combined with memories? Various solutions to the binding problem have been proposed, such as through quantum coherence (Penrose), reentrant signaling (Edelman), or binding by dendritic networks (Crick), but none is generally accepted.
In spite of these contradictory proposals, Searle accepts uncritically this general research strategy. He asks, "Now what exactly happens between the assault of the stimuli on our receptors and the experience of consciousness, and how exactly do the intermediate processes cause the conscious states?" And he answers, "As far as we know the relevant processes take place at the micro levels of synapses, neurons, neuron columns, and cell assemblies. ... All of our conscious life is caused by these lower-level processes, but we have only the foggiest idea of how it all works." This is not in itself a philosophical mistake, but Searle's naive trust in the current representational research program grows out of his philosophical position. His ontology is limited to two levels of causal explanation, neuronal and representational, so that the only sensible research question seems to be, how we get "over the hump" from one to the other.
This representationalist view is seductive, because it seems to be supported by data from neurobiology. However, Searle and his target authors seem to be unaware of a subtle circularity in their appeal to empirical evidence.. About 50 years ago, with great developments in electronics and computer science, there began an invasion of researchers and ideas from the physical, engineering, and cognitive sciences, which grew to a flood that transformed neurobiology. Experimental designs and the interpretations of data were reformulated in terms of information, memory storage, analog comparators, networks, filters, integrators, logical gates, etc. In other words, to the extent that neurobiology is identified with computational neuroscience, it becomes indistinguishable from artificial intelligence.
Physicists, philosophers, molecular biologists, and immunologists coming to this recent literature cannot see that its current contents have already been formulated in terms of the concepts for which they then claim to find evidence. That is, what they are looking for in their research and how they interpret their findings have already been determined by these concepts of information, feature detection and representation. The perspective needed to see this circularity can only be gained through a detailed understanding of the neurobiological literature of the preceding half century, which they do not have, nor does the younger generation of neurobiologists.
Fortunately, there is an alternative research program that is not influenced by this imported information processing view and which, therefore, does not have to solve the intractable binding problem. It is not representation because, on this account, the stimulus input only acts to trigger the interaction of masses of neurons whose interconnection has been determined by previous experience. The brain then constructs the significance of the input for the organism rather than merely representing the features of the object that is the source of the stimulation. The key new concept that is needed is interaction of neurons with each other to form assemblies, of assemblies to form brains, of brains to cause muscles to move the organism into the shared environment, thereby perturbing the sensory receptors.
The mechanisms and kinds of interaction in physical, chemical, biological and social systems have been most deeply explored in recent decades by Aharon Katzir-Katchalsky in the nonequilibrium thermodynamics of cells; by Nobelist Ilya Prigogine in studies of dissipative structures and chaotic transitions; by Hermann Haken in the "synergetics" of lasers, with circular causality between macroscopic and microscopic phenomena; and by Michel Foucault in his descriptions of the "power-knowledge duo" in social systems. Models of the relational dynamics of interactions are implemented with the tools of nonlinear dynamics, and the approach has been popularized in theories of chaos and complexity.
Applications to brains of these theories of interaction are at four hierarchical levels. First, large numbers of neurons interacting through innumerable synapses under the influence of stimuli and endogenous neurohormones form macroscopic populations. These are not the neuronal groups of Edelman, which are genetically pre-formed netlets that are selected by Darwinian mechanisms, as with lymphocytes, and that no one has found (good immunology, bad neurobiology). They correspond more closely to Donald Hebb's "nerve cell assemblies", which are formed and modulated through experience, but still, networks in assemblies are not populations. The relationships of the neurons to the mass can be explained by Haken's synergetic theory, whereby the microscopic neurons contribute to the macroscopic order and then are "enslaved" by that order, similarly to particles in lasers and soap bubbles.
At the second level, populations of neurons interact with each other in extended regions of the brain by large bundles and tracts of axons. Each part of cortex and basal ganglia maintains its own "soap bubble" dynamics, with specializations based in its history and input, and it is pushed into creating new patterns within itself that reflect and contribute to an ever-shifting global pattern involving the entire forebrain. These are not Edelman's reentrant excitations, which are "mappings" that correspond to transfers of information in computational neural networks. They are dynamical flows with continuous distributions and trajectories, comparable to hurricanes and tornadoes. The mathematics needed to describe them is still in development.
At the third level, perception begins with an emergent pattern of activity in the brain. From that pattern, firings go into the motor systems that induce search movements. Firings from that same pattern also go as "corollary discharges" (Sperry) to all of the sensory cortices, to prepare them for the consequences of the intended actions, and to specify the classes of stimuli that are sought. (This aspect was discovered by Helmholtz in the 1870's in his studies of patients with paralysis of the muscles controlling the position of the eyes. When asked to look in the direction that they could not, the patients reported that the world seemed to move in the opposite direction. Helmholtz called this the manifestation of the "effort of will"). The closure that is required for interaction between brain and environment comes with the arrival of the stimulus and the resulting perturbation of the central structures, to which the stimulus-evoked activity is transmitted.
In Freeman's book, "Societies of Brains", he made a case that directed, flexible comportment is an aspect of the dynamical interplay of motor output and corollary discharge with proprioceptive and exteroceptive feedback, and with repeated update of the limbic mechanisms for orientation of action in time and space. The internal updating and restructuring of its past, as the basis for constructing each next step into its future, is the essence of the function of each brain. The availability of that structuring for the guidance of actions by each individual, uniquely expressed in an evolutionary unfolding, has its subjective aspect in the experience of that individual. All available parts of the forebrain participate, and the entire body of past experience, in the form of synaptic modifications and neurohormonal modulations, is brought to bear in varying degree at each moment, waking or sleeping. This interpretation is borne out by EEG analyses. On this account, we get a form of explanation that is apart from top-down representational causality and bottom-up brain causality. It corresponds to Merleau-Ponty's description of how we experience everyday meaningful activity.
Whether this answers the mystery of consciousness depends upon what one thinks science can and should be able to explain. One might well claim that, just as Newton, with his differential equations of motion, did not need to explain why there was gravitational attraction, so, once we have a form of brain explanation that mirrors the phenomenon of meaningful perception and action as we experience it, there will be nothing more to explain. The important job, then, is to explain how brains extract what is significant for organisms, and the research we have been describing gives suggestions, unavailable to the dominant paradigm, as to how the brain does it. Freeman's account of the brain requires its patterns of neuron firings be understood in terms of the whole organism coping flexibly with its world. There only seems to be a mystery of consciousness if one thinks one has to explain, as Searle seems to, how intrinsically meaningful mental states are caused by the meaningless firing of neurons.