Online research in philosophy

If connectionism is to be an adequate theory of mind, we must have a theory of representation for neural networks that allows for individual differences in weighting and architecture while preserving sameness, or at least similarity, of content. In this paper we propose a procedure for measuring sameness of content of neural representations. We argue that the correct way to compare neural representations is through analysis of the distances between neural activations, and we present a method for doing so. We then use the technique to demonstrate empirically that different artificial neural networks trained by backpropagation on the same categorization task, even with different representational encodings of the input patterns and different numbers of hidden units, reach states in which representations at the hidden units are similar. We discuss how this work provides a rebuttal to Fodor and Lepore's critique of Paul Churchland's state space semantics.

Grush and Churchland (1995) attempt to address aspects of the proposal that we have been making concerning a possible physical mechanism underlying the phenomenon of consciousness. Unfortunately, they employ arguments that are highly misleading and, in some important respects, factually incorrect. Their article ‘Gaps in Penrose’s Toilings’ is addressed specifically at the writings of one of us (Penrose), but since the particular model they attack is one put forward by both of us (Hameroff and Penrose, 1995; 1996), it is appropriate that we both reply; but since our individual remarks refer to different aspects of their criticism we are commenting on their article separately. The logical arguments discussed by Grush and Churchland, and the related physics are answered in Part l by Penrose, largely by pointing out precisely where these arguments have already been treated in detail in Shadows of the Mind (Penrose, 1994). In Part 2, Hameroff replies to various points on the biological side, showing for example how they have seriously misunderstood what they refer to as ‘physiological evidence’ regarding to effects of the drug colchicine. The reply serves also to discuss aspects of our model ‘orchestrated objective reduction in brain microtubules – Orch OR’ which attempts to deal with the serious problems of consciousness more directly and completely than any previous theory.

Age-old battle lines over the puzzling nature of mental experience are shaping a modern resurgence in the study of consciousness. On one side are the long-dominant "physicalists" who view consciousness as an emergent property of the brain's neural networks. On the alternative, rebellious side are those who see a necessary added ingredient: proto-conscious experience intrinsic to reality, perhaps understandable through modern physics (panpsychists, pan-experientialists, "funda-mentalists"). It is argued here that the physicalist premise alone is unable to solve completely the difficult issues of consciousness and that to do so will require supplemental panpsychist/pan-experiential philosophy expressed in modern physics. In one scheme proto-conscious experience is a basic property of physical reality accessible to a quantum process associated with brain activity. The proposed process is Roger Penrose's "objective reduction" (OR), a self-organizing "collapse" of the quantum wave function related to instability at the most basic level of space-time geometry. In the Penrose- Hameroff model of "orchestrated objective reduction" (Orch OR), OR quantum computation occurs in cytoskeletal microtubules within the brain's neurons. The basic thesis is that consciousness involves brain activities coupled to a self-organizing ripples in fundamental reality.

The search for neural correlates of consciousness (or NCCs) is arguably the cornerstone in the recent resurgence of the science of consciousness. The search poses many difficult empirical problems, but it seems to be tractable in principle, and some ingenious studies in recent years have led to considerable progress. A number of proposals have been put forward concerning the nature and location of neural correlates of consciousness. A few of these include.

In this commentary, arguments are made for a dendritic code being preferable to a temporal synaptic code as a model of conscious experience. A temporal firing pattern is a product of an ongoing neural computation; hence, it is based on a neural algorithm and an algorithm may not provide the most suitable model for conscious experience. Reiteration of a temporal firing code as suggested in a preceding article (Helekar, 1999) does not necessarily improve the situation. The alternative model presented here is that certain synaptic activity patterns, possibly those possessing universal features as suggested by Helekar, can become encoded in the dendritic structure. Following dendritic encoding, quantum phenomena in those specific dendrite sets could illuminate the static image of that encoded synaptic activity. It is the activation of the static image that would be equivalent to conscious experience; thus, conscious awareness would not be directly affiliated with synaptic activity. This dendrite encoding model may go farther than other models to explain the gestalt nature of consciousness, insofar as quantum entanglement could produce an interconnectedness between specific sets of dendrites-an interconnectedness that need not be based on neural computation or neural connections.

O'Regan & Noë (O&N) are pessimistic about the prospects for discovering the neural correlates of consciousness. They argue that there can be no one-to-one correspondence between awareness and patterns of neural activity in the brain, so a project attempting to identify the neural correlates of consciousness is doomed to failure. We believe that this degree of pessimism may be overstated; recent empirical data show some convergence in describing consistent patterns of neural activity associated with visual consciousness.