Saturday, March 18, 2017

Consciousness: a philosophical tour OWEN FLANAGAN GÜVEN GÜZELDERE

Consciousness: a philosophical tour

OWEN FLANAGAN

GÜVEN GÜZELDERE

DOI:10.1093/acprof:oso/9780198524144.003.0001

Abstract and Keywords

The main concern of this chapter is to provide an overview of the main philosophical positions on consciousness. Ontology is the study of the way things are or the study of the nature of things. The focus of this chapter is on the ontological questions of consciousness: what is its nature and how does it fit into the overall fabric of the universe? The author examines how the concepts of consciousness and mind, and consciousness and intelligence, are not equivalent. Different philosophical positions on the problem of consciousness are presented in the chapter. It explores the Cartesian and epiphenomenalist’s view of the conscious mind and demonstrates how the Cartesian tradition inflates its importance while the latter deflates its significance. In addition, parallelism, new mysterianism, and constructive naturalism’s position on consciousness are illustrated as well.

Introduction

The aim of this chapter is to provide an overview of the main philosophical positions on consciousness. The main focus will be on the ontological question. Ontology is the study of the way things are, the study of the nature of things. With respect to consciousness the ontological questions are what is its nature and how does it fit into the overall fabric of the universe. Answers to questions such as these are philosophical, but they do not come only from philosophers. Philosophy is continuous with science in two senses. Philosophical questions about the nature of things have been best addressed by thinkers, both philosophers and scientists, who pay attention to what the most relevant sciences say about the phenomena in question. Furthermore, scientific theories inevitably assume certain broad philosophical positions. Some world-class neuroscientists in the twentieth century have thought that the mind is the brain and thus that neuroscience holds the key to understanding mental life, while others have embraced dualism, the view that the conscious mind is a phenomenon of an altogether different kind than that countenanced by brain science.

Naturalism and non-naturalism

The most fundamental ontological question is this: why is there something rather than nothing? This almost everyone will admit is a perplexing question. Once we admit that there is something rather than nothing, even if we cannot explain why, other, potentially more tractable, questions arise: given that there is something rather than nothing, what is it that there is? What in the most general terms comprises reality? Why is some of the stuff that exists alive, and why is some of the live stuff conscious? These are among the basic ontological questions.
Naturalism is the view that what there is, and all there is, is natural, or physical, or material stuff. This, of course, is more of a motto or slogan than an analysis, since it is silent about what counts as natural, or physical, or material. The standard move here is to say that what is natural is whatever the physical sciences end up telling us exists, whatever the physical sciences end up being ontologically committed to as they proceed in the project of understanding the nature of things. For such reasons, naturalism is also sometimes called physicalism or materialism.
(p.4) Non-naturalism comes in several versions. The version relevant for present purposes says that in addition to the material world there is an immaterial world. Physical science will suffice to teach us about the nature of the material world. But metaphysics (that which goes beyond physics) will be required to explain the nature of the immaterial world.
With respect to the nature of persons, the naturalist says that we are material beings living in a material world. The mind is the brain; mental events are brain events. The non-naturalist says that persons have an immaterial part, typically their souls or minds. The mind is not the brain, mental events are not brain events. Again, these are slogans. But they serve to capture the spirit of the position.

Consciousness and mind

Some philosophers equate the concept of mind with the concept of consciousness. Others, mostly contemporary thinkers, think that consciousness is one among many faculties of mind or, more plausibly, that the set of mental events is larger than and includes the set of conscious mental events.
According to the first view, a mind is necessarily a conscious mind. All mental events necessarily have the property of being conscious. It is fair to credit Descartes, the great seventeenth century philosopher, mathematician, and scientist, with the view that mental events are necessarily conscious and thus that there are no non-conscious mental events. ‘As to the fact that there can be nothing in the mind, in so far as it is a thinking thing, of which it is not aware, this seems to me to be self-evident.’ (Descartes 1993, p. 171 (fourth set of replies to Arnaud).)
John Locke, despite being engaged in a controversy with the Cartesians over innate knowledge, nonetheless agreed with Descartes and his followers that all mental events were conscious. In 1690 Locke wrote: ‘thinking consists in being conscious that one thinks’, and ‘the idea of thinking in the absence of consciousness is as unintelligible as the idea of a body which is extended without having parts’ (Locke 1959, Book 2, Chapter 1, p. 138).
According to the second view, there are non-conscious mental events. Freud canonized this idea. Indeed, until the time of Freud there was no widely shared theoretical framework in which to reject the Cartesian idea of equating the mind with whatever lay within the scope of one’s consciousness. But the basic insight that mental events need not necessarily be conscious is available in the writings of Leibniz. Leibniz, in a series of responses to Locke written between 1703 and 1705, anticipated some important developments in psychology two centuries ahead of their time, especially those with regard to the nature and role of the unconscious:
[T]here are a thousand indications which lead us to think that there are at every moment numberless perceptions in us, but without apperception and without reflection…. In a word, insensible [unconscious] perceptions are of as great use in psychology as insensible corpuscles are in physics, and it is equally as unreasonable to reject the one as the other under the pretext that they are beyond the reach of our senses. (Leibniz 1951, pp. 374–8.)
(p.5) Now we have not only the dynamic unconscious of Freud but also the information-processing unconscious of cognitive psychology and even the neurobiological unconscious. If one thinks that any of these systems exists, are mental, and are non-conscious, then the concept of mind includes as a subset the conscious mind but is not equivalent to it. In this way questions about mind and consciousness come apart. The relationship of the conscious mind and the body may require a different analysis than the relationship of the non-conscious mind and the body. And the relationship between non-conscious mental events and conscious mental events may require yet a third analysis.
Some distinguished contemporary philosophers reject the picture of mental activity as including both conscious and non-conscious events. John Searle (1992) and Galen Strawson (1994) argue that all mental events are conscious (or potentially conscious). According to both Searle and Strawson it is a misleading linguistic courtesy to extend the tribute of ‘mental’ to non-conscious information-processing states or to neural events that lack the property of being conscious.

Consciousness and intelligence

So far we have seen that according to many but not all contemporary philosophers, the concepts of ‘mind’ and ‘consciousness’ are not equivalent. The same is true for the concepts of ‘consciousness’ and ‘intelligence’. Without defining either term—‘consciousness’ or ‘intelligence’—we can see some reasons for keeping the problem of the nature of consciousness separate from the problem of the nature of intelligence. This is true even though the normal use of our conscious minds may be to guide our actions rationally and intelligently. The reason is this: there are computational systems that display signs of intelligence—they play world-class chess and prove theorems in mathematics, for example. No one thinks that such systems are conscious. But they display flexible, novel, unpredictable behaviour just as humans faced by similar problems do.
Alan Turing, in a famous paper written in 1950, suggested interpreting the question: ‘can machines think?’ as equivalent to the question ‘can a machine successfully mimic human behaviour?’ (Turing 1964). If it can, we should say it thinks, that it is intelligent.
Is such a machine a conscious thinking thing? The answer has only become obvious recently, and the answer is ‘no’. The question of consciousness was dropped by the wayside by the Turing test, the test designed to answer the question: can a machine think? Turing’s insight, one might say, was to see that intelligent systems and conscious systems—smart or dumb—need not be equated.
To get a grip on Turing’s insight one might imagine someone observing a world-class chess-playing computer asking the questions: how can it play so well?; how can a computer be so smart?; how can a mere machine display such intelligence? The answer would come from within a naturalistic ontology: the complex interplay between the software and hardware explains the intelligent behaviour displayed by the computer.
(p.6) Complex adaptive systems such as computers, missiles, and robots perform all sorts of functions similar to those humans perform. Intelligence, reasoning according to the canons of deductive logic and probability theory, performing chemical assays, even ‘visual’ discrimination of objects and locomotion are capacities that most people can understand in naturalistic terms. What has historically been found puzzling is how conscious intelligence, conscious reasoning, conscious perception, consciously deciding to walk from home to the office can be explained in natural terms.
In 1898, physicist John Tyndall wrote: ‘The passage from physics of the brain to the corresponding facts of consciousness is unthinkable.’ (Tyndall 1898, p. 420.)
The difficulty of conceptualizing how physical brains can give rise to experience, how certain features of the objective world can give rise to subjectivity is the problem of ‘the explanatory gap’ (Levine 1983).
In his famous paper ‘What is it like to be a bat?’ (1974), Thomas Nagel writes that ‘Consciousness is what makes the mind–body problem really intractable…. Without consciousness the mind-body problem would be much less interesting. With consciousness it seems hopeless’ (Nagel 1979, pp. 165–6).
One interpretation of Nagel’s remark is this: if the mind-body problem is simply the problem of how physical systems such as ants, bats, and humans can intelligently get around, behave socially, and the like, then the resources of physics, neuroscience, and evolutionary biology can make visible the shape of a naturalistic solution. When consciousness, subjectivity, experience, qualia are added to the mix, confidence in the explanatory power of naturalism can easily wane. As Colin McGinn puts it: ‘How can technicolour phenomenology arise from the soggy grey matter of brains?’ (McGinn 1989).
The point is that problems of intelligence or rationality can, in principle, be prised apart from questions of conscious experience. David Chalmers calls the problems associated with intelligent information processing and action guidance ‘the easy problems’ in the sense that we can picture how the natural sciences can solve them. Consciousness is ‘the hard problem’.
[E]ven when we have explained the performance of all the cognitive and behavioural functions in the vicinity of experience—perceptual discrimination, categorization, internal access, verbal report—there may still remain a further unanswered question: Why is the performance of these functions accompanied by experience? […] This further question is the key question in the problem of consciousness. Why doesn’t all this information-processing go on ‘in the dark’, free of any inner feel? (Chalmers 1995, p. 203.)

The official doctrine of the ghost in the machine

In the opening chapter of The concept of mind, entitled ‘Descartes’ Myth’, Gilbert Ryle writes:
There is a doctrine about the nature and place of minds which is so prevalent among theorists and even among laymen that it deserves to be described as the official theory…[T]he official (p.7) doctrine, which hails chiefly from Descartes, is something like this…every human being has both a body and a mind…His body and his mind are ordinarily harnessed together, but after the death of the body his mind may continue to exist and function. (Ryle 1949, p. 11.)
This is the doctrine of ‘the ghost in the machine’. My non-physical mind is harnessed to my body during my lifetime, but being non-physical it is of a different ontological kind than my body and can exist without it.
What could be at the basis of this powerful intuition that mind and body are ontologically different? For Descartes the intuition expressed a self-evident truth. Bodies are extended in space. Atoms, rocks, tables and chairs, and human bodies are extended in space. These are paradigm case physical objects. Thoughts are, or so it seemed to Descartes, unextended—paradigm case non-physical objects. Thoughts still seem this way to most people. Whether they are this way is a different question.
The official doctrine, despite its intuitive appeal, has some problems. How a non-physical mind which occupies no space can be ‘in’ anything is left unsolved, as is the problem of how the world can get information to the mind and how the mind can move the body. Descartes located the point of commerce between mind and body, between the ghost and the machine, in the pineal gland. But this tells us only where immaterial and material forces meet, it tells us nothing about how causation in either direction is possible (see Popper and Eccles (1977) for a recent attempt to spell out the details of a dualistic theory).

Parallelism and worries about interaction

The official doctrine of the ghost in the machine remains the most popular version of non-naturalism. Philosophers, among them Descartes’ contemporaries and their followers, saw that the interaction problem among different ontological kinds was a big problem:
One strategy was to take the theological background theory available to all the modern philosophers working on the mind-body problem and have God more involved in making the mind-body relation intelligible. This Cartesian picture assumes an ontological continuum with God as pure mind, pure res cogitans, rocks as pure matter, pure res extensa, and humans, while embodied, as mixtures of thinking stuff and extended stuff. Once an omniscient God is part of the background, the Cartesian can say that God somehow solves the interaction problem. We may not understand how God solves the problem, but he, being God, does solve the problem.
Some other ontological non-naturalists or dualists seem to have thought that there were better ways for God to solve the mind-body problem than by allowing interaction among different ontological kinds. Leibniz’ and Malebranche’s different kinds of parallelism were similar ways of retaining dualism without the interaction problem, but at no small cost.
According to Leibniz, God (possibly at the creation of the universe) set a pre-established harmony between mind(s) and body(ies). When I decide to move my arm it moves, but not because my decision causes my arm to move but because God has (p.8) willed that minds and bodies stay in synchrony. Malebranche’s view (1638–1715) differed only in having God involved on each and every occasion in which a mental act and a bodily event needed to co-occur. While both views solve the interaction problem by denying that there is any interaction, they cause trouble for the concept of free will. According to Descartes, writing in 1649, one advantage of the official doctrine is the place it leaves for free will: ‘But the will is so free in its nature, that it can never be constrained…And the whole action of the soul consists in this, that solely because it desires something, it causes a little gland to which it is closely united to move in a way requisite to produce the effect which relates to this desire.’ (Descartes 1968, p. 350.)
The Leibnizean has trouble with free will since he has us picture a world in which God sets all the clocks at creation. Malebranche’s view has resources to avoid this problem (although these are not discussed by him). Since God is at all times involved in every life, he can wait until I make a decision and then (being God) simultaneously get my body to fall into line.
Without getting into unnecessary intricacies we can get a feel for parallelism from this quotation:
If we knew thoroughly the nervous system of Shakespeare, and as thoroughly all his environing conditions, we should be able to show why at a certain period of his life his hand came to trace on certain sheets of paper those crabbed little marks which we for shortness’ sake call the manuscript of Hamlet. We should understand…all this without in the slightest degree acknowledging the existence of thoughts in Shakespeare’s mind. [B]ut, on the other hand, nothing in all this could prevent us from giving an equally complete account of…Shakespeare’s spiritual history, an account in which gleam of thought and emotion should find its place. The mind history would run alongside the body-history of each man, and each point in the one would correspond to, but not react upon, a point in the other. (James 1976, p. 136–7 quoting Clifford.)
Three points about parallelism. First, it might seem like a solution to the interaction problem, but it isn’t since it still requires God, either at creation (Leibniz) or on each occasion (Malebranche), to get body and mind to be or appear to be coordinated. This is no less byzantine than Descartes’ solution which just has God figure out how to make different ontological kinds interact.
Second, and on a different trajectory, the position contains an important insight which has contemporary relevance. Even today, philosophers, psychologists, neuro-scientists, and laypeople will remind us that what we always and only seem to have when it comes to the consciousness-brain problem are correlations. A positron emission tomography (PET) scan shows Shakespeare’s brain lighting up and he claims that he is thinking about writing a play called ‘Hamlet’. What are we to say? That the lighted area is the thought of writing the play or that it is correlated with the thought (and the subsequent taking of pen in hand). The move from correlations to identities is a live ontological and methodological issue in contemporary mind science. We will return to this issue shortly.
Third, parallelism is instructive when thinking about reductionism or, more generally, the issue of levels of explanation. One might think that both stories, the (p.9) mental story of Shakespeare’s life and the bodily story of his life—or just take the Hamlet segment of both stories—are equally explanatory. But this is not obvious: the bodily story explains all his bodily movements but it does not explain the production of the meaningful play we call ‘Hamlet’, and this despite the fact that it does explain how certain marks on paper came to be. Nor, on the other side, does the mental story explain the bodily movements even though it explains the ideas behind the play.
Many contemporary naturalists, known as identity theorists, have taken the correlations identified by dualists of the parallelist persuasion and suggested that what we really have are identities. That is, Shakespeare’s thoughts and intentions about the play are in fact identical to a certain set of events taking place in his nervous system. Just as water is H2o and just as salt is NaCl, so too Shakespeare’s plan to write Hamlet is some set {n1, n2…nn} of neural events. From a logical point of view, when we have strict identities we have synonyms, and synonyms can replace synonyms without any loss of meaning. Thus we should be able to tell the mental story of Shakespeare’s writing of Hamlet in terms of the neural states which constitute the activity, which are, as it were, the very activity. This is reduction. But again, many baulk, since something seems to get lost in the reductive translation, namely, the meaningful, intentional properties of the activity.

Epiphenomenalism: Darwin, Freud, and cognitive science

If the Cartesian tradition inflates the importance of the conscious mind, the epiphenomenalist deflates its importance. Epiphenomenalism says that conscious events are ‘mere’ side-effects of the locus of the really interesting action. Epiphenomenalism comes in both a non-naturalist and naturalist version. The non-naturalist thinks that the conscious side-effects of bodily processes are non-physical; but they needn’t worry those who are trying to develop a science of human behaviour since these non-physical side-effects do no significant causal work. The epiphenomenalist who is also a naturalist says that the side-effects are physical but causally inconsequential. William James quotes Thomas Huxley’s startling version of epiphenomenalism.
The consciousness of brutes would appear to be related to the mechanism of their body simply as a collateral product of its working, and to be completely without any power of modifying that working, as the steam-whistle which accompanies the work of a locomotive engine is without influence upon its machinery. Their volition, if they have any, is an emotion indicative of physical changes, not a cause of such changes…. The soul stands to the body as the bell of a clock to the works, and consciousness answers to the sound which the bell gives out when it is struck…to the best of my judgment, the argumentation which applies to brutes holds equally good of men…. We are conscious automata. (James 1976, p. 135.)
Why would anyone think that consciousness was a mere epiphenomenon, a side-effect of what the system is really doing? Part of the reason comes from evolutionary considerations. Nature abounds with creatures that are reproductively successful but are not conscious. The social insects are fantastically fit as measured by the criteria (p.10) of evolutionary biology, but most philosophers and scientists do not attribute consciousness to the social insects. If consciousness is not necessary for reproductive success, then perhaps it is just an accidental and unnecessary appendage in creatures like us that are conscious.
Although James called epiphenomenalism ‘an unwarrantable impertinence’, claiming that ‘[i]t is inconceivable that consciousness should have nothing to do with a business which it so faithfully attends’ (James 1976, pp. 140–1), epiphenomenalism has been taken more seriously in the twentieth century than at any previous time. One reason for the enhanced status of epiphenomenalism comes from the rejection of the Cartesian equation of mind with consciousness. Psychoanalysis, cognitive information-processing psychology, and neuroscience all attribute significant causal power to non-conscious mental events. This convergence of opinion about the causal efficacy of the non-conscious mind has reduced confidence in the causal powers of consciousness. Consciousness seems as if it runs the show, but then again conscious mental states are the only mental states that seem any way at all. This could easily have led us to overestimate dramatically the causal role of consciousness.
This issue of the causal role of conscious mental events in the overall economy of mental life remains very much a live issue.

Contemporary non-naturalists and agnostics

All the traditional views discussed so far continue to have advocates within the philosophical and scientific communities. Despite what many see as the ascendency of naturalistic or scientific views over more traditional theological views, non-naturalism continues to have articulate advocates. Some contemporary non-naturalists think, just as Descartes did, that consciousness can be made intelligible only if it is understood as a power of a non-physical substance or as composed of non-physical properties (Popper and Eccles 1977). Others think that we need to invoke a supernatural cause to explain why phenomenal qualia, the sensation of red or the scent of a rose, are correlated with specific types of brain states (Adams 1987; Swinburne 1984). Still others think that consciousness is miraculous. Like transubstantiation and the Trinity, it is not for us to fathom.
Thomas Nagel, more than anyone else, has articulated a certain uneasiness with both major ontological options. Call his position principled agnosticism (Nagel 19791986). Naturalism is a position we do not understand, because we do not understand (at least at present) how the relation of consciousness and the brain can be made intelligible in naturalistic terms. We do not understand what it would mean to give an objective account of subjectivity. Since one should not believe a theory one does not even understand, agnosticism is the best policy.
One could add a further consideration in favour of agnosticism to which we alluded at the start. Namely, naturalism follows the lead of the physical sciences in (p.11) determining what counts as natural. But the more science develops, the wilder and woollier the natural world seems. The science of Descartes’ time took extension in space as definitive of matter. But today we countenance electrons as part of the material world and our best theories ask us to think of electrons as point-particles without extension. Developments such as these make the boundary between what is natural and what is non-natural increasingly obscure. Again, on the principle that one should not commit oneself to a position one does not understand, agnosticism is a position one might take for reasons of intellectual integrity.

New mysterianism

A somewhat different position is anticonstructive naturalism, noumenal naturalism, or new mysterianism (McGinn 1991). This is the view mat naturalism is true. There are, in fact, properties of the brain that account naturalistically for consciousness. But we cannot grasp these properties or explain how consciousness depends on them. Consciousness is terminally mysterious to our minds but possibly not to minds of greater intelligence. It is terminally mysterious not because it is a non-natural phenomenon, and not because it is a miracle, but because an understanding of its nature is ‘cognitively closed’ to us. The problem of consciousness is a case where we know how to ask the question but lack the mental powers to find the answer.
To get a feel for this position imagine that the most intelligent creatures in the universe are the social insects. They cannot do science. Nonetheless, the laws of gravity, electromagnetism, relativity theory, quantum physics, and so on, hold in the world they live in. They are simply incapable of asking the questions that would lead them to discover the answers about the world that the world, as it were, exemplifies.
According to McGinn we are in a slightly better position: we can ask certain questions about how consciousness works, what it is, and so on. The social insects cannot even ask the questions. But, for this reason, our situation is considerably more frustrating. Since one doesn’t miss what one doesn’t want, the social insects are not frustrated by not understanding the nature of things. They have no desire to know. We can ask the questions and we want the answers, but at least with respect to the problem of consciousness we are simply not up to the task of answering the questions we ask—or so the new mysterians say.
Non-naturalists have their own reasons for thinking that the problem of consciousness will not yield to science. Anticonstructive naturalism, or new mysterianism, is the surprising view that consciousness, despite being a natural phenomenon, will never be understood. Whether its causal role is significant or not, it will not be understand. The ‘old mysterians’ were dualists who thought that consciousness cannot be understood scientifically because it operates according to non-natural principles and possesses non-natural properties. Consciousness might be understood in other ways, for example, by way of an elaborate metaphysical view about the nature of non-physical things and the ways in which they can interact with physical (p.12) things, or by invoking supernatural phenomena (for some sophisticated contemporary defences of supeinaturalism, see Swinburne 1984 and Adams 1987). Because it is somewhat counterintuitive it needs to be repeated that unlike the old mysterianism or contemporary supernaturalism, new mysterianism is a naturalistic position. Mind and consciousness exist, and they operate in accordance with natural principles and possess natural properties. But new mysterianism is a postmodern position designed to drive a railroad spike through the heart of scientism, the view that science will eventually explain whatever is natural.
Colin McGinn thinks that naturalism must be true. There is no other credible way to think about the relation of consciousness and the brain than as a natural relation. Nonetheless, he thinks, we will never be able to set out a credible constructive theory of that relation.
McGinn (1989, p. 349) writes, ‘We have been trying for a long time to solve the mind-body problem. It has stubbornly resisted our best efforts. The mystery persists. I think the time has come to admit candidly that we cannot resolve the mystery’. McGinn (1989, p. 350) thinks that ‘we know that brains are the de facto causal basis of consciousness’, but ‘we are cut off by our very cognitive constitution from achieving a conception of that natural property of the brain (or of consciousness) that accounts for the psychophysical link’.
Although the doctrine is mischievous, coming from a naturalist, it is a coherent position. There are limitative results in physics and mathematics, for example Heisenberg’s Uncertainty Principle and Gödel’s Incompleteness Theorem, that tell us of in-principle impossibilities faced by the physicist and mathematician. It is conceivable that just as we cannot know the position and momentum of an electron at one and the same time, or just as we can know that a certain sentence in arithmetic is true though it is in principle impossible for us to prove it within arithmetic, so we can know that consciousness is a natural phenomenon though it is in principle closed to us to know what sort of natural phenomenon it is.
It is important to see that new mysterianism is different from principled agnosticism. The agnostic thinks that we do not understand what form a naturalistic solution to the consciousness-brain problem would take, so we ought not to confidently claim that naturalism is true. What makes the principled agnostic position agnostic is that naturalism, materialism, and physicalism are not embraced because they are too poorly understood as ontological positions to commit oneself to; but neither is non-naturalism embraced nor is physicalism declared to be false. Nagel (1979, p. 176) writes: ‘It would be a mistake to conclude that physicalism is false…. It would be truer to say physicalism is a position we cannot understand because we do not at present have any conception of how it might be true’.
In his book The view from nowhere, Nagel (1986, p. 47) puts it this way: ‘We have at present no conception of how a single event or thing could have both physical and phenomenological aspects, or how if it did they might be related’. Because we do not understand what form a constructive naturalistic solution to the problem of consciousness would take, we cannot assign credibility to the claim that physicalism is true or to the claim that it is false. Intellectual honesty requires that we be agnostics.

(p.13) Constructive naturalism

Finally, there is constructive naturalism. Against the anticonstructivist and principled agnostic, the constructive naturalist thinks that there is reason for optimism about our ability to understand that relation between consciousness and the brain—reason for hopefulness that we can make intelligible the existence of consciousness in the natural world. Constructive naturalists resist principled agnosticism because they think that the concept of ‘the natural’ can be filled out in a coherent way, and they resist anticonstructivist naturalism because they do not see the cognitive closure or epistemic impossibility that the new mysterian sees. After all, the main argument for the impossibility of solving the consciousness-brain problem comes from the failure to do so thus far. There is nothing like a formal Gödel-like result which proves that certain obstacles to knowledge in the domain of consciousness exist.
Recent work by David Chalmers (1996), Patricia S. Churchland (1986), Paul M. Churchland (19891995), Daniel Dennett (1991), Fred Dretske (1995), Owen Flanagan (19911992), Valerie Hardcastle (1995), William Lycan (1996), John Searle (1992), Galen Strawson (1994), and Michael Tye (1995) is in the constructive naturalist mode. All these philosophers take conscious experience seriously as a phenomenon or set of phenomena to be explained. And they all are optimistic that philosophy and science can build a credible theory of the nature and function of consciousness. There are many disagreements among these philosophers about a wide variety of issues. Where the views converge is on the ontological commitment to naturalism and optimism that consciousness can at least in principle be understood within such a framework.
The following three principles are not shared by all constructive naturalists but they provide a sense of the sort of commitments that might engender confidence that the problem of consciousness can be made to yield.

1. Principle of supervenience

  1. (a) There exist emergent properties such as liquidity or solidity. Consciousness is in all likelihood an emergent property of complex states of the nervous system.
  2. (b) A microlevel change need not produce a macrolevel change; for example, two H2o molecules do not produce liquidity.
  3. (c) But if there is a macrolevel change—if it gets wet in this vicinity—then there is (must have been) a microlevel change, that is, a sufficient number of H2o molecules must have accumulated.
  4. (d) Emergent, macrolevel properties, can causally interact with other emergent, macrolevel events or processes, as well as with (often because of) interactions with microlevel events and processes. So too can emergent conscious events and processes causally interact with conscious and non-conscious mental events (understood now as emergent neural events).

(p.14) 2. Principle of organismic integrity

That consciousness exists is amazing. But ‘given that consciousness exists at all, there is no mystery in its being connected with what it is connected with’ (Dewey 1922, p. 62). The basic idea behind this principle is to soothe, and then remove, certain troublesome intuitions about subjectivity. Given that emergent properties are possible, and that consciousness is probably such a property, then there should be no surprise in the fact that each person has their own and only their own experiences. It is because of the design of the nervous system. We are uniquely hooked up to ourselves. Given that mere are experiences at all, it makes perfect evolutionary and physiological sense that I have my experiences and that you have yours.

3. Principle of biological naturalism

Consciousness…is a biological feature of human and certain animal brains. It is caused by neurobiological processes and is as much a part of the natural biological order as any other biological features such as photosynthesis, digestion, or mitosis. (Searle 1992, p. 90.)
Stated this way, the principle does not deny that consciousness could conceivably occur in systems with alternative biologies (non-carbon based ones, for example) or even in robots made of inorganic parts. It simply proposes that if you want to understand consciousness, study the systems that exist in our vicinity that are known to be conscious.

Conclusion

The aim of this chapter has been to provide an overview of the main ontological positions concerning the nature of consciousness, to provide a quick tour of the main live philosophical positions on the problem of consciousness. Although almost all philosophers and mind scientists are committed to one of the broad ontological positions discussed here, it would be premature to say that anyone knows which picture is ‘the right one’. It is possible that the best ontological position has not yet been thought of. Happily, the area of consciousness research is at present engaging more good minds than at any point in human history. There is reason to be hopeful that philosophy and science will illuminate each other and take us in the direction of getting a better hold on the nature and function of consciousness. Time will tell.

Thirty years of object recognition Glyn W. Humphreys

Thirty years of object recognition Glyn W. Humphreys

DOI:10.1093/acprof:oso/9780199228768.003.0012

Abstract and Keywords

This chapter looks at changes and developments in the study of object recognition during the past thirty years. It discusses the Marrian revolution attributed to David Marr, who took ideas and concepts from psychophysics with the aim of translating them into working computer models. Another major development during this period was the recognition and serious research on the neural basis of visual perception and object recognition.

My perception, in the 1970s

I completed my undergraduate degree in 1976.I had started university when student protest was still common, buildings would be taken over, flags bravely unfurled and then abandoned, flapping lifelessly from windows. By 1976 this had begun to seem faintly silly. Rather than pulling things down for some vague goal, one wanted to see how things could be made to work.
I suppose I studied psychology because I wanted to understand how humans worked. I was fortunate to sit in social psychology lectures given by Henri Tajfel on group dynamics, fuelled by his own experiences as a wartime exile. I spent hours learning about various reinforcement regimes, which at least seemed factual and enabled a sort of understanding of which type of learning might apply in which situation. I even undertook a rare gem of a course on mathematical psychology covering topics such as Luce's choice theorem and Bayesian analyses of decision-making. But, though of course I received the statutory courses, I just didn't ‘get’ Perception. Although the classes were filled with enough demonstrations to satisfy even a Royal Society Christmas Lecture audience, I couldn't figure out what it all amounted to, what the mechanisms were. Things began to click together only when I attended a lecture on cognitive psychology, where I remember the idea of using converging operations was discussed. Suddenly some larger picture began to fall into place. The lecture introduced the Atkinson-Shiffrin model of short-term memory—my first encounter with a theory formulated in a box-and-arrows framework, where the representations inside the boxes were specified along the connections between the boxes. Here was something that could direct experiments and was open to empirical evaluation. Most excitingly, this approach could be tested (indeed a converging operations advocate would argue that it should be tested) using different lines of evidence—not just from studies of free and serial recall by normal participants, but also, for example, from patients whose brain lesion might mean that one part of the model may not function properly. If correct, the model should predict the pattern of impairment (p.152) found in neuropsychological populations. The idea of using converging evidence to test models was a revelation, suggesting that one should be able to link together work from different fields to construct an overarching account of human cognition. Moreover it encouraged the idea that different lines of converging evidence could then be designed to assess different component processes in the cognitive system. All this was somehow lacking in my understanding of perception. Our lectures explained that adaptation was the psychophysicist's microscope, but, as it were, all I could see were single cells. I really had no idea of what a perceptual system might comprise.
It would be unfair if these comments were read as a specific criticism of perception as I was taught it, because the fragmented picture reflected something of the state of affairs at that time. There were many clever experiments and interesting, non-intuitive ideas on aspects of perception (the notion that visual coding might operate through spatial frequency analysis had begun to infiltrate the undergraduate curriculum), but it was rare to find the different strands of work being linked together. Failing to see what could be done in perception, I went on to conduct a PhD in an area that would subsequently become known as visual cognition—inspired by Michael Turvey's (1973) (to this day beautiful) work on how different forms of masking could be used to probe sequential stages of visual processing. I hoped to advance our understanding of letter recognition by analysing the time course over which different types of information were made available. I had friends who were perception guys. They studied after-effects and visual gratings. I spent hours in dark labs listening to BBC Radio 4 and trying to detect low contrast patterns. But the world my friends inhabited was a different one to mine. Their vocabulary was foreign.

The Marrian revolution

Then, during my PhD, my psychophysicist friends started to talk about someone called David Marr, who was taking ideas from their field with the aim of translating them into working computer models. What was interesting was that, to do this, you had not only to specify how inputs were coded but also how different codes might be integrated, to think about the order of events, and to specify how the evolving representation could access stored knowledge that might allow the model to do something useful—like recognizing an object. In other words, to have a working computer model, you had to think of perception as a system. Suddenly I could see an analogy with models with which I had grown familiar, particularly accounts such as the dual-route model of word recognition being proposed by Max Coltheart and colleagues (e.g. Coltheart 1978). Marr's ideas offered a new kind of scaffolding to link (p.153) different aspects of perception together using converging operations from computer science as well as from visual psychophysics. It gave scientists coming with very different approaches a common language.
The initial paper that kindled my interest was Marr and Nishihara (1978), built on Marr's earlier proposals (Marr 1976; Marr and Hildreth 1980). Note the incremental approach: Marr had earlier dealt with feature coding, and now Marr and Nishihara went on to consider higher-level representations where features were integrated and then associated with past knowledge. This in itself felt a novel way of thinking—a reflection of a computational approach in which a complex system could be built by linking together modules that each performed their own particular job. Moreover, the proposals put forward by Marr and Nishihara, for how you might go from feature representation to code a surface-based description of an object, and then from that to a modal three-dimensional (3D) representation, specified mechanisms for how object recognition might actually take place. These mechanisms could be tested.
As I came to read more, it became clear that Marr was not the first person to think of constructing explicit theories of pattern and object recognition using ideas from psychology and physiology—proposals such as the Pandemonium model of Selfridge (1959) long predated the work—but Marr's arguments still felt revolutionary. In part I think this was because they came with a well worked-through philosophy for how different approaches to perception could be linked. Marr argued for the utility of having different levels of description. He proposed a computational level of theory which set out the constraints that would impact on any system that used vision for object recognition—as relevant to computers as to humans. For example, his work on developing a model for stereopsis (Marr and Poggio 1976) used constraints such as: no point in the world should be represented at more than one point in an internal representation of depth, to limit possible mappings between points in the two eyes. Similar constraints still influence computational models today, for example the Heinke and Humphreys (2003) account of visual selection.1 Beneath the computational level of theory, Marr suggested that one could have an algorithmic theory, based on abstracted processing mechanisms, which could be implemented in different kinds of hardware. Further, underneath this, he suggested that there could be a theory of the hardware, which dealt with how particular algorithms were realized in different physical systems. As Coltheart notes in Chapter 5, much of subsequent cognitive science has been built on (p.154) the idea that theories can be abstracted from the hardware on which processes operate. Marr's framework for different levels of theorizing makes this explicit, and has had a profound influence on the field—though, as I shall describe, the boundaries between, for example, the algorithmic and hardware levels have become increasingly blurred over time.2
After my PhD, I was lucky to gain a lectureship at Birkbeck College where Max Coltheart had recently taken up a chair in psychology, and the department was a hotbed of research into aspects of reading. This work had a particular flavour. It employed functional accounts of performance, such as the dual-route model, to guide experiments, and it used data from neuropsychological patients with disorders of reading alongside data from normal participants. The neuropsychological work seemed especially exciting. Here theorists went outside the laboratory and addressed real-life problems that people experienced after brain lesions. The models could also be used to guide therapy (e.g. Coltheart et al. 1992), and so could be useful practically. As a young lecturer it was impossible not to become infected.3 The functional account of cognition offered by the dual-route model could be thought of as an algorithmic level theory, in much the same way as Marr and Nishihara's proposed mechanisms underlying visual object recognition. It was thus not difficult to think of testing the Marr and Nishihara account using similar procedures to those used to test dual-route theory—with cognitive neuropsychological studies providing an important part of the empirical armoury. This was how my own work in this area started. It was not a profoundly original approach, and Graeme Ratcliff and Freda Newcombe were already embarked on a similar analysis (Ratcliff and Newcombe 1982). However, up to that date I think it is true to say that neuropsychological data had had little impact on theories of normal object recognition, and indeed there were still controversies within the neurological literature over whether ‘true’ disorders of visual object recognition could occur without contamination from peripheral visual disturbances or more profound cognitive impairments (Bender and Feldman 1972).
My own work was helped enormously on two counts. One was meeting Jane Riddoch, who was beginning a PhD under Max Coltheart's supervision as I came Birkbeck and who wanted to carry out neuropsychological studies of cognition. It was Jane's insights into patient disorders that helped frame the (p.155) questions posed in our joint work, and her access to patients made the research possible. The second was meeting HJA, a profoundly agnosic patient with a wonderfully persevering nature. HJA had many low-level visual processes as well as high-level cognitive capacities preserved, refuting the argument that frank disorders of ‘intermediate’ visual processes could not exist (contra Bender and Feldman 1972). HJA subsequently loyally helped our research for over 25 years (e.g. Riddoch and Humphreys 1987a; Riddoch et al. 1999). Patients such as HJA, with selective disturbances of particular aspects of cognition, have made great contributions to the field, and single-case studies should not be overlooked despite current-day emphases on group-based lesion analyses (see The Biological revolution, below).
The first neuropsychological papers on disorders of object recognition that I read were those of Warrington and Taylor (19731978). These distinguished between groups of patients who had deficits either in matching objects depicted in different views or in matching between physically different exemplars of objects used to perform the same basic function (e.g. a wheelchair and a deckchair, both of which serve the function of being used to sit on). Such data provided early suggestions that aspects of object recognition could be fractionated; for example, the ability to achieve viewpoint-independent matching was distinct from access to semantic/functional knowledge about objects. Moreover, the data indicated that some of the processes proposed by Marr and Nishihara had psychological reality (e.g. that there might be some process that derived common object structures across viewpoints). The basic fractionation made by Warrington and Taylor has also continued to influence much of the work in the field; indeed the question of how objects can be recognized across different points of view has generated enormous heat and perhaps rather less light than one would hope (see Biederman and Gehardstein 1993; Tarr and Bülthoff 1998). Interestingly, findings that patients with problems in matching objects across different viewpoints can retain an ability to recognize objects in prototypical views (e.g. Davidoff and Warrington 1999) remain perhaps one of the strongest pieces of evidence suggesting that Marr and Nishihara's account was not correct in its details. For example, according to Marr and Nishihara, some form of view-independent object representation needs to be constructed to enable recognition to occur. If patients cannot construct a view-independent representation, then their recognition of objects in all views should be impaired. However, perhaps the more important point is that, through formulating their account of the perceptual system underlying object recognition, Marr and Nishihara paved the way for questions about view-independent representation to be addressed in a theoretically coherent way.

(p.156) After the first revolution

Following from Marr's work, subsequent theories of object recognition have differed in many critical ways. One distinction concerns whether surface-based and 3D representations of objects need to be coded for recognition to take place. For example, Biederman's (1987) influential ‘Recognition by Components’ theory supposed that object representations could be assembled directly from the edges of visual objects, without the need to generate any intermediate surface-based representations. Other theorists have proposed a more direct image-based approach to recognition, where multiple, view-specific, memory representations may be held and used to match objects appearing in different viewpoints (Edelman and Bülthoff 1992). Hybrid accounts, in which view-independent and view-specific procedures operate in parallel, have also been proposed (Hummel and Stankiewicz 1998). These hybrid models hold that view-independent coding requires attentional processes that ensure that the parts of objects are coded in appropriate relative spatial locations, bringing into the play the issue of how attention may play a modulatory role in object recognition. Studies in which attention is manipulated in normal participants, or which use patients who are limited in attending across all the parts of objects, have provided some support for hybrid accounts (e.g. Stankiewicz et al. 1998; Vernier and Humphreys 2006).
A further question highlighted by post-Marrian theories concerns the role of colour and surface texture on object recognition, as edge-based approaches to object recognition maintain that colour and surface texture should play little causal role. Here there is again converging evidence from studies with normal participants and with patients pointing to there being an influence of colour and surface texture at least for some object classes and for objects for which surface information is a reliable cue (e.g. Humphrey et al. 1994; Price and Humphreys 1989; Riddoch and Humphreys 2004; Tanaka and Presnell 1999; Wurm et al. 1993).
We can think of this empirical work as refining our ideas about what we might term the intermediate representations involved in object recognition, such as the surface- and 3D-model representations suggested by Marr and Nishihara (1978). In addition to this, converging experimental work with normal participants and patients has helped to ‘flesh out’ our understanding of how the input into these intermediate representations is coded (how perceptual features are integrating and organized) and also what later processes are required for object recognition (the involvement of different forms of stored knowledge). For example, we have argued that work with patient HJA distinguishes between processes that group oriented elements into edges, (p.157) and subsequent processes that code the relations between edges within and across objects (Humphreys 2001; Humphreys and Riddoch 2006). HJA can perform normally on tasks requiring that local oriented elements are grouped (Figure 11.1), but he is profoundly impaired at encoding the correct relations between edges within and across shapes—indeed his recognition errors often involve inappropriate segmentation of shapes based on misinterpreting an internal edge as a segmentation cue (Giersch et al. 2000; Riddoch and Humphreys 1987a). It is thus possible to elaborate on different stages of visual grouping and perceptual organization. Work by Mary Peterson and colleagues also provides evidence that perceptual organization operates in a top-down as well as a purely bottom-up manner, so that processes such as edge assignment (in figures with ambiguous figure-ground relations) are influenced by whether the edge forms part of a known object representation (see Peterson and Skow-Grant 2003). This notion—that earlier visual processes can be ‘penetrated’ by top-down knowledge—is a critical point that contrasts with the ideas put forward by Marr and colleagues. In keeping with the idea of stand-alone computational modules, Marr proposed a bottom-up approach to object recognition whereby early processes were not affected by feedback from processes at higher levels of representation. The questions of whether, when, and how top-down processes might influence earlier stages of object recognition are ones that will drive research in this field for some time to come.
                   Thirty years of object recognition
Figure 11.1 Example stimuli used by Giersch et al. (2000) to examine the grouping of oriented elements in the agnostic patient, HJA. HJA had a normal threshold for detecting a round group of elements aligned by collinearity, when shown against a ‘noisy’ background (see (a)). Nevertheless, HJA had major problems distinguishing between figures when they overlapped and had difficulty organizing lines within the context of complex shapes. Figure (b) illustrates how HJA assigned edges in shapes, when asked to colour different shapes in contrasting hues. Each shading pattern represents a different hue.
(p.158) Distinctions between different forms of higher-level representations in object recognition have also been suggested. Evidence for structural representations of objects separate from semantic/functional representations in normal participants comes from reports by Schacter and Cooper (1993) that normal participants showed long-term priming for novel but plausible 3D shapes (with minimal semantic representations), but no priming for implausible shapes. They argue that plausible but not implausible 3D shapes must have persistent structural representations. In neuropsychological studies, several investigators (Fery and Morais 2003; Hillis and Caramazza 1995; Riddoch and Humphreys 1987b; Sheridan and Humphreys 1993; Stewart et al. 1992) have documented patients who can distinguish reliably between real objects and structurally similar non-objects, but who remain impaired at accessing semantic knowledge about the objects, for example in matching together semantically related objects. Such dissociations indicate a separation between stored structural representations of objects and stored semantic knowledge. The framework put forward by Marr and Nishihara (1978) needs to be expanded to take account of these additional distinctions.
One other major change in experimental and neuropsychological work on perception after the Marrian revolution has been to emphasize the importance of visual information for action. If you followed courses on Perception from the 1970s through to the mid-1990s you would have hardly thought that vision was used for anything other than describing the visual world and recognizing objects. Of course, in everyday life vision is used for much more than this—particularly for guiding our actions on the world. In the 1990s, David Milner, Mel Goodale, and colleagues (Milner and Goodale 1995; Milner et al. 1991) described the agnosic patient DF who showed an impairment apparently even earlier in the visual stream than that suffered by HJA, as she showed profound limitations when making perceptual judgements about groups of visual elements or the orientations of single lines. Strikingly, though, DF was able to reach and post a letter through a letterbox positioned at different orientations! Milner and Goodale argued that there is a distinction between the visual information that is used for conscious perceptual judgements and for object recognition (processes that are damaged in DF), and the visual information used for action (spared in DF). Subsequently, Goodale and associates have attempted to derive converging evidence from studies of visual illusions in normal participants. Here, it has been argued that our actions are much less susceptible to some illusions than our conscious perceptual judgements (e.g. Agolioti et al. 1995; Bridgeman 2002; Haffenden and Goodale 1998; for alternative views see Franz et al. 2000; Pavani et al. 1999). Other work has suggested that the actions we intend to make can modulate how we attend to (p.159) objects, and, through this, alter how objects are coded (Linnell et al.2005). The step towards thinking of what behavioural outcomes result from visual processing has to have been a healthy one in terms of thinking about real-world applications, and also one that now enables converging work to be developed between vision scientists and computer scientists and engineers working on robotic systems.

The biological revolution

There is one other change I believe worth highlighting, that has taken place after the Marrian revolution. This is that the neural basis of visual perception and object recognition (indeed, of all of cognition) is now taken much more seriously. One of the main drivers for this has been the development of functional brain imaging, which now allows us to assess which brain regions are active when we, for example, recognize particular types of object. My view is that brain imaging can contribute to our understanding of the functional basis of object recognition, not least because it brings another type of converging evidence to bear. The new evidence is concerned with where in the brain a given process operates. Now, because we have prior knowledge of what a given brain region is typically involved in, new information indicating that this area is recruited when a given stimulus is processed, can constrain our account of what kind of processing is involved. As a concrete example, Moore and Price (1999) contrasted the neural regions activated when participants named black and white line drawings relative to when they named colour images. They found differential activation in a number of posterior areas in the right hemisphere. One functional account of why coloured images can be easier to identify than black and white images of objects is that coloured images specifically facilitate name retrieval (Ostergaard and Davidoff 1985). A contrasting account is that colour images facilitate the object recognition process itself. Given that changes are observed in the right hemisphere, and that the right hemisphere is not usually thought to modulate name retrieval in normal right-handed participants, these imaging data suggest that the effects of colour are on object recognition itself. Arguments such as this, of course, start to blur Marr's distinction between the algorithmic level of description and descriptions of the hardware. Accounts of what particular regions of the ‘hardware’ are doing can be used to inform accounts of what algorithms might be involved. I find nothing ideologically objectionable in this. It seems simply to be a case of using extra (dare I say converging) evidence to help refine out arguments about complex processes such as object recognition.
(p.160) This ‘biological revolution’ is still evolving, but some new emphases are apparent. Imaging data suggest that distinct brain regions may be recruited when different objects are recognized. This is perhaps most obvious when contrasting faces and other objects, given the highly reliable finding that small regions of the occipital cortex and fusiform gyrus show enhanced activity to faces compared with other stimuli (Grill-Spector et al. 2004; Kanwisher and Yovel 2006). However, neural specialization can be observed for other classes of object too. Haxby et al. (2001) raised the possibility that there are not generic ‘object recognition procedures’, but rather that contrasting processes may be called into play, depending on the object involved. This idea of recruitment may be important here. For example, there is evidence that there is activation of medial temporal cortex, left parietal and ventral frontal cortex when tools are recognized (Grabowski et al. 1998; Grafton et al.1997). The interesting point that medial temporal cortex is associated with motion processing (Beauchamp et al. 2002), and parietal and ventral frontal regions are associated with tool use (Decety et al. 19941997), suggests that associations with object motion and functional actions may come into play as we process tools, and these associations may even help us recognize the object involved. These suggestions from imaging sit alongside neuropsychological studies showing that patients can have selective deficits (or sparing) in processing faces versus other objects (Buxbaum et al. 1996; Riddoch et al. 2008; Rossion et al. 2003; Rumiati et al. 1994) or relatively impaired (or preserved) recognition of tools compared with living things (Riddoch and Humphreys 2004; Sirigu et al. 1991). Whereas the earlier emphasis from neuropsychological studies was primarily on the functional deficit involved, arguments about the lesion site now also become relevant. Of course, it can be difficult to argue about lesion site from single cases, given the (relative) idiosyncrasy of different brains, and so this also leads to a change in the way that research is done, moving work towards case series of patients rather than single cases (e.g. Humphreys and Riddoch 2003). Though, as I have argued, the continuing importance of single cases, and of functional dissociations, should not be lost when we add in further information about common lesion site over groups of patients.
One can caricature the box-and-arrow models that emerged during the Marrian revolution as being static, based on established representations in set boxes, and set connections between the boxes. However, an emergent emphasis from studying the biological basis of visual processing is that perceptual systems are not static but change dynamically over time. In studies of functional imaging, the importance of dynamic change has been highlighted by techniques such as adaptation (a return of the psychophysicist's electrode?), which have been developed to provide a finer-grained analysis of the neural (p.161) substrates of processing (e.g. Kourtzi and Kanwisher 2001). Imaging studies show that neural areas responding to a stimulus have reduced activity if the same stimulus is adapted repeatedly. This would be consistent with the cells responding to that stimulus in that region entering a refractory state. The extent to which there is recovery of activity when the same stimulus is shown under different conditions (e.g. when the viewpoint changes) or when a new stimulus is presented, indicates both whether the same neurones in that region code the different stimuli, and whether the region contains different populations of neurones that can now be prized apart by the selective adaptation of the neurones to responding to one stimulus. Given the limited resolution of much of present-day functional imaging (e.g. using voxel sizes of 2×2×2 mm, say), adaptation has proved to be an important way of probing the selectivity of neural responding. But, perhaps even more than this, it indicates that dynamic changes operate continuously in perception, with both short- and longer-term changes being evident (see Kourtzi and DiCarlo 2006; Kourtzi and Kanwisher 2001). Understanding these dynamic changes is a critical issue for future research. The emphasis on dynamic change and learning also enables links to be formed with neural network models that incorporate dynamic fluctuations in activity as part of their normal operation, and with studies of how perceptual systems evolve as they develop. The importance of converging operations will not go away.

Acknowledgments


I thank Jane Riddoch for comments. The work was supported by the Medical Research Council and the Stroke Association (UK).