Tuesday, April 18, 2006

Notes from Tucson

I was fortunate to be able to attend a good portion of the Toward a Science of Consciousness 2006 conference in Tucson. This seventh biennial event, sponsored by the University of Arizona’s Center for Consciousness Studies, brings together all strands of inquiry into consciousness, including neuroscience, cognitive science, psychology, philosophy, physics and more. The conference had 500+ attendees and while I understand it used to be even larger, I can attest it was very robust and diverse. In addition to the main stage talks there was an unbelievably rich set of concurrent talks (most of which I missed) and poster presentations. It seems to me that the topic demands an interdisciplinary forum like this, and Tucson ably fills the role. I met and spoke to a number of interesting people there; everyone was extremely nice and there was a great vibe to the proceedings.

Below are notes on three of the talks I heard, those of Giulio Tononi, John Searle, and Paul Davies. I may look to add more later on.

Uriah Kriegel has 2 posts at Desert Landscapes about the conference (here and here). Unfortunately, I did not get a chance to meet Uriah, but happily did catch his main stage talk, where he effectively presented his self-representational theory of consciousness.

I. Giulio Tononi presented his information integration theory of consciousness. He was a very effective speaker and made a strong case for his work.

Here are a few of his preliminary remarks. Many have searched for “the” neural correlate of consciousness. Attempts to localize the phenomenon have not worked well. Today many think the correlate is in widely distributed thalamocortical activity. (But as an aside, attempts to identify a synchronicity mechanism for binding brain states is not promising and Tononi doesn’t seem to see a need for something like synchronization via 40 hz oscillation).

A couple of other thoughts: consciousness is more than introspective/reflective/higher-order self-consciousness. It is more and essentially simpler. It also goes beyond consciousness of the environment (dreams show us there is something of a world inside the skull).

Tononi offers these 2 paradoxes help us to think about consciousness (to which he returns at the end):
1. Why is the correlate in the cerebrum, not the cerebellum? The cerebellum has even more neurons, connections.
2. Why is it present in wakefulness rather than (non-REM) sleep?

Here’s a brief summary of Tononi’s approach (warning: these notes are taken on the fly and may do a poor job outlining the theory).

We have a large differentiated repertoire of possible states. Each state conveys a lot of information about itself as well as ruling out all other states. Also, each state is integrated (you cannot really subdivide it).

Tononi will use mathematics to characterize this phenomenology. Define “effective information” to be the entropy obtained by part of a system when you perturb another part. The minimum information partition you can put into the system (the ‘weakest’ link) will be used to define this. This can be measured, and is a measure of information integration.

It turns out that to maximize information integration, all elements in the system should participate by connecting to each other, but with non-uniform number and/or pattern of connections. Fewer connections is bad, but so is uniformity/simplicity of connections (specialization is good).

Analysis of the thalamocortical system shows that it has the right combination of specialization and integration to maximize effective information in the system. The cerebellum, in contrast, consists of separated modular systems which are not well integrated. Analysis of waking vs. sleeping states also shows how consciousness is correlated with this combination.

My summary thoughts: First, one implication I think is important is that consciousness is a graded phenomenon, not an all-or-nothing one. Second, it must be noted that a theory of this sort does not explain the interior/subjective and qualitative nature of consciousness (the hard problem). However, it impressed me and other listeners as presenting a good path for exploring how the particular richness as well as cognitive effectiveness of human consciousness is grounded in brain organization.

II. John Searle is an excellent speaker and I’m glad I got a chance to hear him talk. For those who take the “hard problem” seriously, however, Searle remains annoying. His talk was entitled "Dualism Revisited"

Searle defines consciousness as qualitative, subjective, unified and intrinsically intentional. He asserts consciousness is real and irreducible. Functional or behavioral approaches cannot explain consciousness (he has always been effective with his criticism of these).

He then says consciousness is caused in the brain by low-level neuronal processes, and is realized in the brain as a higher-level or system feature. Consciousness is a biological (specifically neurobiological) phenomenon.

He insists despite its irreducible first person ontology, consciousness just is a biological fact in the world. Once we have a good neurobiological theory, it will show how neuronal processes cause consciousness, and then we will be done.

He wants to say that the hard problem is just a conceptual confusion. We have different levels of description of the phenomena, but he denies this shows a need for any special further explanation. He uses (in my opinion) faulty analogies regarding different levels of description of say, an automobile engine (faulty because both levels are third-person, not first and third!). “Just stop worrying about it!” is the message.

Later in Searle’s talk, he speculated about why progress in explaining consciousness in neuroscientific terms so far seems to be slow. His answer was that the work focuses on localized phenomena and not enough on the large scale conscious “field”. These comments were in tune with some of Tononi’s mentioned above about the problem researchers have had when trying to localize the NCC.

III. Paul Davies on quantum biology and on a possible link between cosmology and strong emergence.

Paul Davies discussed his exploration of a number of speculative ideas which could advance our understanding of life and mind. First he discussed the view that quantum mechanics has something to contribute to this understanding.

He said that are 2 ways QM could play a role in life. First, a negative effect, with quantum indeterminacy limiting efficiency at the micro-level. Second, a positive effect, where life actually harnesses or exploits QM to improve efficiency or accomplish difficult tasks (with a comparison to the potential gains present in quantum computing).

Davies said there is plenty of circumstantial evidence for the role of QM in life. Certainly, we know many small biological structures do operate on the “quantum edge” in terms of their size. He gave a few examples of this.

He also discussed his suspicion that QM played a role on the origin of life (OOL). He very briefly outlined an idea of “quantum replicators” which could have been simpler precursors of living things.

He discussed the biggest challenge for quantum biology, which is decoherence. Simple models and experimental results indicate that quantum systems decohere very rapidly. But he says we may not know the whole story yet, and there may be environments where coherence can be sustained.

The next subject was on the subject of QM interpretations and whether there was an objective environmental trigger of wave function collapse (such as Roger Penrose’ idea that gravity plays a role). Davies discussed the idea that a specified level of complexity could engineer the collapse of the wave function and define the microscopic/macroscopic frontier. Quoting Seth Lloyd’s idea of the universe as a quantum computer, he said one could estimate the large but finite amount of classical information the universe can contain. He quoted an estimate of 10^120 bits. Quantum systems leveraging coherence can handle a huge amount of information, but perhaps the finite cosmological limit cuts off the size of these coherent quantum systems. A system of, say, 400 or so entangled particles could represent the approximate limit given this complexity ceiling imposed by the cosmos.

Finally, if there is indeterminacy due to a finite information capacity of the universe, then Davies said this could remove the obstacle to micro-level causal closure of the universe, and open the door to downward causation. “Physical systems are causally open when they reach a certain level of complexity – a level determined by the information processing capacity of the universe.” The origin of consciousness may lie within the quantum to classical transition at this critical threshold.

I like Davies a lot. At this point in his career he has been freed by his publishing success, Templeton prize, etc. to engage in diverse speculations rather than work within normal academic boundaries. But he is a very serious fellow. The main virtue being that even if these ideas have a very small chance of being correct, the fact that they would be very big deals if they did work out makes it worthwhile.

I got to talk with him in the lobby of the amphitheatre briefly and asked what he thought it would take to get more serious research time and money allocated to quantum biology. He said we needed a result that was both secure and non-trivial. He said we do have evidence of quantum level mechanisms in biology (citing enzymes and proton tunneling in DNA mutation), but these are relatively trivial, not interesting, effects. There are many leads to follow, though, and he is optimistic about some research which is underway.


Peter said...

Wow! I'm envious.

Hope you do manage to post more notes in due course.

state said...

I went to the first one of these, back in....'96, I think; it was okay. Obviously, the guest roster has gotten better. If I'd known, I might've gone.