in Homuphobia in the Cartesian Theatre, Keith Sutherland writes:
Metzinger draws on the research of von der Malsburg, Engel, Gray, Singer, Llinas, Crick, Koch and others to construct what he calls a truly natural-science, bottom-up theory to resolve the binding problem. Perceptual binding is held to result from phase-locked oscillations in the feature-detection neurons involved in the perception of a particular object. This binding is both spatial in nature (across distributed areas in the brain) and also temporal, as illustrated by Llinas (1996). The consequence of this process of perceptual binding is that the brain constructs models of objects, scenes and the world. According to Metzinger the phenomenal self is just another model that we forget is a model and take to be ontologically real.I would agree that any attempt to describe the phenomenal self in terms of feature binding and object recognition alone can't help but be inadequate to the task. And while I consider the 40 Hz hypothesis to have promise -- I don't think it has so much "evaporated" as become the focus of controversy, like so much else in csness studies -- saying that 40 Hz oscillations are THE mechanism by which consciousness is generated is, indeed, overly reductionistic. As an analogy, this would be like stating that speech is "nothing but" the patterns of neural activity generated in the cortex by the sound waves. While sound waves and the neural activity they generate are surely necessary to a scientific theory of speech encoding, they hardly provide a comprehensive explanation of language processing.
I have not read Metzinger's chapter, so I don't even know if it is fair to attribute the criticisms in Sutherland's posting to him. But I spent an entire section in Newman & Baars (1993) applying von der Malsburg's and Singer's (1988) "principles of self-organization in cortical networks" to Baars' (1998) Global Workspace theory of consciousness; and in that context, I would like to offer a defense (at least) of vdM & Singer's (1988), and my views (Newman & Baars, 1993, Newman,1995a&b, Newman, Baars & Cho, in press) on the importance of coherent neural oscillations in generating conscious contents. First of all, oscillatory phenomena are an absolutely ubiquitous function of neural tissue. They are produced wherever pools of excitatory and inhibitory neurons feedback upon each other, which is just about everywhere in the CNS. These oscillatory loops are not dependent upon external stimulation, although they are constantly being influenced by inputs form other neurons. Because there are several frequency bands (delta, theta, alpha, beta) and hundreds of millions of such pools of oscillating neurons, isolating the "generators" of these EEG waves has proved a very challenging task for neuroscience. It took over twenty years of research for workers like Steriade, Jones and Llinas (1990) to isolate the nucleus reticularis thalami (nRt) as the definitive generator of the alpha (8 to 12 Hz) "spindles" which signal the onset of sleep. To this day, there is still controversy about the role of waking alpha in cognition, although any encephalographer knows that the "alpha range frequencies are synchonous over widespread cortical regions" in the awake human subject (Knight, 1985, p. 329). Introductory textbooks tell us that alpha is associated with a resting or "idling" brain, while asynchronous activity in the beta range (20 - 60 Hz) is characteristic of a cognitively active cortex.
"40-Hz oscillations" (sometimes called "gamma" waves) are a fairly new concept (ergo controversial) and refer to pools of often widely separated neurons that show "phase-locked" oscillations, which several labs have shown to be closely correlated with features of an attended object. An example would be neurons separated by several millimeters in primary visual cortex, but all biased to fire preferentially when a vertical bar of light passes across the retina. If the bar is split into two bars moving in tandum, the phase-locked oscillations attenuate, but remain. But if the two halves are moved in opposite directions, phased-locked activity ceases. Now this is a pretty interesting phenomenon, and has been demonstrated for all primary cortical modalities, even inter-hemispherically (see review by Gray, 1994). But can such binding of simple features serve as an adequate model "to describe the phenomenal self in exclusively bottom-up terms."?
Well no, but this is only one aspect of vdM's and Singer's (1988) theorizing. The first point is that the "self-organizing cortical networks" they describe are known to exist virtually everywhere in the cortex. So it seems reasonable to infer that such "coherency matching" of phase-locked pools of neurons involved in the binding of features of an attended object, ALSO operate in V4 and MT and superior parietal and prefrontal cortex. Now these areas are known to mediate highly complex cognitive functions, including "executive", or "top-down", processing (see e.g. Damasio, 1994; Fuster, 1980; Newman & Baars, 1993 for reviews). Thus, von der Malsburg & Singer (1988) write,
it will become a fascinating challenge to both theoreticians and experimentalists to examine the functional properties that emerge if these processes of self-organization are iterated beyond primary sensory areas and applied to architectonics in which "dimensional" or "conceptual" vicinity is ever more emphasized ... ( p. 95).Nor do the authors stop there. They point out that such local, yet ubiquitous, oscillatory couplings, are highly susceptible to being,
gated by central core systems and hence can, in principle, be made dependent upon global states such as arousal, attention and motivation; .... (ibid., p. 94)Since the central contention of my modeling of the neural substrates of Baars' (1998) Global Workspace (GW) theory is that the reticulo-thalamic core of the brain is responsible for "gating" a winner-take-all competition between local cortical processors for access to the GW (and, thus, consciousness), I find this extension of experimental and anatomical findings quite compatible with GW theory (see Newman, Baars, & Cho, 1995, in press).
Llinas (1996) has extended this idea still further by claiming that the great swath of looping circuits projectioning between the thalamus and cortex (and "gated" by nRt) generate a "scanning wave" of 40-Hz activation traveling across the entire cortical mantle every 12-or-so milliseconds. This finding is highly controversial, and has been difficult for other MEG labs to replicate. But to suggest that such recent findings, using equally recent technologies, are "spurious" strikes me as throwing out the baby with the bathwater. It would be like someone calling the patterns of neural activity generated by sound waves "spurious" because there is no clearly articulated theory about how they are tranformed into phonemes in auditory cortex. As for the criticism that models such as these cannot distinguish whether,
the oscillations are the cause or the consequence of perceptual binding... Put more directly: how does the brain "know" that the output of certain feature- detecting neurons is part of any particular object without a prior (top-down) model of that object?I have to answer mea culpa to the first charge; but I can blame the brain: it perniciously refuses to function in a linear, causal fashion. The hundreds of millions of feedback loops it employs made a travesty of our efforts to establish Newtonian causal sequences. But if "top-down" can be defined in terms of cortical influences upon the thalamic gating of information flow in the CNS, then I have no problem answering that criticism. Llinas has noted that there are in the neighborhood of 10x's as many projections from the cortex back to the thalamus than in the other direction. And as Scheibel (1980), Taylor (1992 ) and I (Newman, 1995; Newman et al., in press) have theorized: prefrontal cortex exercises a variety of executive functions related to "self-like" activities (e.g. voluntary attention & recall, monitoring goals, planning) via feedback control over thalamic, limbic and striatal systems.
So I don't know about Metzinger's or Crick & Koch's models, but the more global models I have cited seem to me to accomodate most of the concerns Keith Sutherland articulates. Scheibel (1980) even argues from animal experiments for the existence of a "spatial envelope" in the midbrain - - a multimodal sensory map -- which initially orients the animal in extra-personal. So it could turn out that our brains "'know' that the output of certain feature-detecting neurons is part of any particular object [via exchanges of maps with].... a prior [bottom-up] model of the object". This may sound bizarre, but it looks increasingly like some sort of exchange between visual association cortex and subcortical areas mediates blindsight preformance (re Larry Weiscrantz's Tuscon II presentation on the "tecto-fugal system"). Of course, it will take years of challenging empirical work to see if such theories actually pan out. Sutherland continues,
There is a wide- spread belief within the cognitive sciences that the phenomenal self is a constructed illusion. This is partly due to a robust homuphobia but in some cases also seems to result from a misunderstanding of certain [Eastern] teachings ...Ironically, these statements strike me as dogmatic. If one's own experience were "the only starting point", then ANY sort of 3d person, scientific theory of csness must turn out to be "desperate and flawed". On the other hand, if anyone asserted that all 1st person experience is reducible to 3d person explanations, I would consider that sort of thinking "desperate and flawed". This is the real crux of Chalmers' hard problem: not whether our actual brains can generate qualia, but whether science can accomplish the dubious feat of "doing away" with common-sense descriptions of experience like: "I see a red ball" or "I'm feeling uneasy about that." or "Isn't this a glorious sunset!" , not to mention such "minor" areas of human endeavor as morality, aesthetics, and spirituality. I see no serious obstacles to science's continuing successs in "mapping" consciousness, in all its labyrinth complexity, onto the nervous system (although our maps will always be limited and imperfect), but the simple fact is that,I don't actually have any a priori objection to incorporating religious insights to help unravel the problem of consciousness.... But, I would maintain, in the field of human consciousness it is better to start with our own experience rather than dogma. One's own experience, which is perhaps the only starting point, seems to be centred around a phenomenal experiencer. But if you look inside the brain you can't find any little green men and this has given rise to a fear of homunculi, agents and Cartesian theatres. All this has resulted in some desperate and flawed attemps to build a bottom-up theory.
The map is not the territory. - Gregory BatesonAnd I, for one, do not find this a "problem". Indeed, it is my 1st person experiences of my world and the rich other-worlds of "2d persons" that makes modeling consciousness in 3d-person terms so fascinating (obviously I don't buy us being zombies). So besides enjoying the experience of sunsets and eating out with friends, I love trying to imagine ways of relating the profound reality that is our conscious (and unconscious) exisitence to scientific theories of mind & brain. In that vein, let me end with just such an imagining from a forthcoming chapter in the John Benjamins' Advances in Consciousness Series:
As an example of how [Global Workspace theory] can be fruitfully applied to central philosophical problems in consciousness studies, take the perennial conundrums of the homunculus and Cartesian theater. The two are, of course, related. The metaphor these images conjure up is of a "little man in our head" observing and manipulating the play of conscious images passing across the theater of our mind. That image, while beguiling, is absurd: for who is this strange being lodged in my (or your) mind? And who controls him?From: Jim Newman[GW] theory suggests a more complex and dynamic scenario: the single homunculus is replaced by a large audience. The theater becomes a workspace to which the entire audience of "experts" has potential access, both to "look at" others's inputs and contribute their own. Awareness, at any moment, corresponds to the pattern of activity produced by the then most active coalition of experts, or modular processors. Thus, there is not some fixed, superordinate observer. Individual modules can pay as much or as little attention as suits them, based upon their particular expertise and proclivities. At any one moment, some may be dozing in their seats, others busy on stage. Thus, a crucial difference between these local experts and a theater audience is that each can potentially contribute to the direction the play takes. In this sense the global workspace resembles more a deliberative body than an audience. Each expert has a "vote", and by forming coalitions with other experts can contribute to deciding which inputs receive immediate attention and which are "sent back to committee". Most of the work of this deliberative body is done outside the workspace (i.e., non- consciously). Only matters of central import gain access to center stage.
The beauty of this sort of system is that while it is a multiplicity, its behavior is largely coherent and adaptive. This later characteristic derives from the fact that the workspace serves as a "global integration and dissemination system". It is out of the operation of this system that conscious awareness arises (and, over time, our sense of being a coherent "I"). And it is this unitary awareness -- not any agent or homunculus -- that is globally superordinate, not in any hierarchical sense, but in the sense of the GW being accessible for representing and potentiating the activities of any of the competing arrays of "specialized bioprocessors".
Of course, such a system is prone to inefficiencies and pathological perturbations, but these seem consistent with the scientific literature concerning human consciousness (see Baars, 1988, 1995).... (from Newman, Baars & Cho, in press).
References
Baars, B.J. (1988). A Cognitive Theory of Consciousness. Cambridge: Cambridge University Press.
Gray, C.M. (1994). Synchronous oscillations in neuronal systems: mechanisms and functions. Journal of Computational Neuroscience 1, 11- 38.
Knight, R.T (1985) Electrophysiology in Behavioral Neurology. Chap 9 in M-M Mesulam (Ed.) Principles of Behavioral Neurology. Philadelphia, PA: F.A. Davis Co.
Newman, J. (1995b) Commentary: Reticular-thalamic activation of the cortex generates conscious contents. Behavioral and Brain Sciences, 18:4, 691-692.
Newman, J. (1995a) Review: Thalamic contributions to attention and consciousness. Consciousness and Cognition, 4:2, 172-193.
Newman, J. & Baars, B.J. (1993) A neural attentional model for access to consciousness: A Global Workspace perspective. Concepts in Neuroscience, 4:2, 255-290. Newman, J., Baars, B.J. & Cho, S-B (1995) A neurocognitive model for attention and consciousness. Tenth Annual Conference on AI and Cognitive Science (AISB-95), Sheffield, England.
Newman, J., Baars, B.J. & Cho, S-B. (in press) A neurocognitive model for attention and consciousness. In S. O'Nuallain et al.(Eds.) Two Sciences of Mind (Advances in Consciousness Research Series). New York: John Benjamins.
Scheibel, A.B. (1980) Anatomical and physiological substrates of arousal: A view from the bridge. In J.A. Hobson & M.A.B.Brazier (Eds.), The Reticular Formation Revisited, pp. 55-66. New York: Raven Press.
Steriade, M., Jones, E.G. & Llinas, R. (1990) Thalamic Oscillations and Signalling. New York: John Wiley & Sons.
Taylor, J.G. (1992). Towards a neural network model of the mind. Neural Network World, 2, 797-812.
jcs-online menu