Pilgrims on an unknown road

"Curiously enough . . . it is sometimes those basic problems that look impossibly difficult to solve which yield the most easily. This is because there may be so few even remotely possible solutions that eventually one is led inexorably to the correct answer." These words of Francis Crick (in What Mad Pursuit) could turn out to apply to the problem of consciousness.

This book, the selected proceedings of a conference held at the University of Arizona in 1994, is a landmark in the study of consciousness. It contains 56 articles from disciplines as diverse as philosophy, cognitive science, neuroscience, neural networks, quantum theory, hierarchical organisation and phenomenology. Many are specialised, but there are also some for the general reader.

Three criteria by which to gauge the articles might be as follows. First, do they take the first-person perspective of subjective experience seriously? Second, is a satisfactory explanatory framework supplied? Third, is an approach evident that brings together both philosophy and empirical science?

It is often thought that the first-person perspective can only be investigated in human subjects not animals. Thus blindsight is a phenomenon which has previously been studied in people with a damaged striate-cortex. They claim to be blind in the visual field dealt with by this damaged area but when shown something they cannot "see" and asked to guess what it is, they choose correctly with a high statistical significance. Here, in a fascinating paper by Petra Stoerig and Alan Cowey, monkeys with apparent blindsight are investigated. Through a series of experiments, Stoerig and Cowey coax the monkeys into effectively telling them whether they can guess the position of something in their blindsight or damaged receptive field, and reporting whether they can see it.

Britt Anderson and Thomas Head produce evidence of a kind of blindsight in language. They show that patients who are severely "aphasic" and overtly lack language comprehension, respond covertly to funny limericks, as demonstrated by changes in the electrical potential of their skin. It is amazing that the full comprehension of a complex spoken utterance can be demonstrated in the absence of overt comprehension, and possibly of conscious access.

The timing of conscious experience is looked at in an article which outlines the pioneering work of Benjamin Libet. He shows that people appear to backdate experiences, so as to discount the delay from initial stimulus to conscious experience. Here experiments pit neuronal activations against the accompanying subjective reports made by the individual. The paper by Mikael Bergenheim et al, which investigates simultaneity and synchronisation of sensory information in human subjects, makes a good companion piece to Libet's.

The way that different modalities and domains may interact in a person is examined in Thaddeus Cowan's paper. Here a curious effect is examined in which an after-image of the subject's hand is created by a bright flash of light in an otherwise pitch-black room. When the subject moves his arm, he sees the after-image move with it in the darkness. The visual map of where the hand is and a motor map of where the hand should be, interact.

Alfred Kaszniak and Gina DiTraglia Cristenson review research on patients who have no idea of their acquired disabilities, such as paralysis and memory loss. This lack of self-awareness seems to be correlated with specific types of brain damage peculiar to Alzheimer's patients (localised in the frontal cortex).

This brings us to the question of perceptions about perceptions. Guven Guzeldere, a philosopher of mind, argues that the so-called higher-order perceptions are a conceptual fallacy; they muddle up what is being represented with who is doing the representing. But Guzeldere fails to note that the inner structure of perceptions may well involve perceptions about perceptions anyway. This may be part of the "bootstrap" process that gets consciousness off the ground. We do not understand enough about perception to make the claim Guzeldere does.

All the above papers take subjective experience seriously. Other papers in the book demonstrate how techniques such as EEGs, Pet scans and the study of split-brain patients are revealing how conscious experience correlates with brain function.

A number of writers claim that quantum physics can explain consciousness. This may be a reductionist fallacy. How could the collapse of the wave function capture what it feels like to smell a rose? Nevertheless, several articles do draw fascinating parallels between quantum physics and the mind. These include the smeared-out nonlocality of both mental and physical events, their holism (where parts interact to form a global unity) and the actuality of the present. But these parallels may just be parallels. Quantum physics may not have anything directly to do with consciousness.

Stuart Hameroff and Roger Penrose tell a particularly elegant story in which a macroscopic quantum field is supported by the crystalline structure of microtubules. Consciousness occurs when this field collapses. Much is made of the "noncomputability" of this collapse, said to be induced by quantum gravity. But if one is not hung up on noncomputability, there seems little reason to buy into this story. Is Penrose tilting at shadows?

An important neuroscience article by Christof Koch outlines his work with Crick on the neural correlates of consciousness. Koch suggests that area V1 of the visual cortex is not conscious, because it does not project directly to the frontal cortex, which is involved in planning and voluntary motor outputs. But it might be a mistake to discount the mail boy just because he has never met the chief executive. Maybe the mail boy has a soul as well: area V1 does not just pass on information, it also gets feedback from above.

What is known as the "binding problem" refers to the question of how the brain joins outputs from separate, functionally distinct regions together into the perception of a unified object, without having an assembly area. Synchronous 40 Hz firing between spatially separate groups of neurons might tie their informational content together by their coherent rhythmical activity. Valerie Hardcastle is sceptical on the basis of the evidence. However, this is a hot topic and new evidence is accumulating rapidly.

Neuronal rhythms might also be involved in attention. Tokiko Yamanoue presents a simple neural network model with oscillatory activity. Without any additional mechanism, it seems to mimic a number of well-known properties associated with human attention amazingly.

Alwyn Scott attempts to draw a global picture of the hierarchical structure of the universe, of which consciousness is one level among many. This is a delightful read. However, we need a tighter conceptual framework that will capture the logic of how consciousness is generated, why phenomenal experience is self-evident, what role self-reference plays, and why the mind both floats on the workings of the nervous system, yet is sealed off from it. I think mathematical logic may be a clue here.

Although we have to take the first-person perspective of subjective experience seriously, explanatory conceptual frameworks are seldom developed from this point of view. David Chalmers argues that it is the organisation and circuitry underpinning consciousness which matters, not whether it is realised in neurons or silicon. If the substrate were suddenly to change from organic neurons to silicon chips, how could this be accompanied by a change in the subject's consciousness, from, say, seeing red to seeing blue? Since there is nothing in the subject's behaviour to indicate a difference, and, more profoundly, there is nothing internally to allow the subject to know there is a difference, Chalmers concludes that the substrate cannot matter.

Another way of arguing this, I suggest, is to start by realising that explanations are purely relational structures or maps. If someone argues that the material substrate of the circuitry matters, in explaining why he would have to unpack the material and the workings of the circuits, ie regard them as a relational structure. But we could then substitute an alternative substrate to instantiate this structure, and so leave its conscious correlate unchanged.

Explanations as relational structures may apply, I suggest, at the phenomenal as well as the physical level. Thus we could ask what is required in order to distinguish between two percepts, such as seeing red and seeing green. In order to answer this question, we might consider a scenario where someone loses this discrimination, and can see only "gred" when either red or green is presented to the senses. How might he now reinstate his faculty of discrimination? He would need two independent markers that allowed the subject to tag "gred" in two different ways, one pair of which (say tag 1 plus gred) would then be the percept "red" and the other (say tag 2 plus gred) would be the percept "green". These tags must themselves be independent conscious percepts (if they are to allow the subject to know there is a difference). Seen in this way, the essential components of sensations are other related sensations, which are assembled to form more complex recursive hierarchies.

As Crick surmises, there may be so few even remotely possible solutions to the problem of consciousness that we are led inexorably to the correct answer. But this will require scientists to understand the philosophical constraints on a theory of consciousness, and philosophers to have the courage to create theories with testable predictions.

Paul Caro is honorary research associate, department of mathematics, University College, London.