Consciousness has many aspects, but the fundamental feature of consciousness is sentience or subjective awareness.  As the philosopher Thomas Nagel put it in his famous essay, "What Is It Like To Be a Bat?" (1979),

The fact that an organism has conscious experiences at all means, basically, that there is something it is like to be that organism.  There may be further implications about the form of the experience; there may even (though I doubt it) be implications about the behavior of the organism.  But fundamentally an organism has conscious mental states if and only if there is something that it is like to be that organism -- something it is like for the organism. We may call this the subjective character of experience.  It is not captured by any of the familiar, recently devised reductive analyses of the mental, for all of them are logically compatible with its absence.  It is not analyzable in terms of any explanatory system of functional states, or intentional states, since these could be ascribed to robots or automata that behaved like people though they experienced nothing.

As Nagel put it in his "Bat" essay, 

"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."

Dualism

• Animals were unconscious machines, operating purely by reflex.  When energy from a stimulus fell on a receptor organ, it was reflected back to the environment in the form of behavior.
• Humans, on the other hand, have consciousness.  They are aware of their own minds and of the objects of thought.  Human beings also have reflexes, because they have bodies, but consciousness frees humans from the dominance by reflexes.  

• Body is characterized by extension, and thus takes up space.
• Mind, in contrast, is characterized by thought, which is composed of an immaterial substance.

• Through the sense organs and the brain, the body mediates the mind's perception of the outside world.
• But the mind's thoughts and desires also cause bodily actions to occur.

But how can a material substance affect an immaterial one, and vice-versa?  Descartes reasoned that this interaction took place within the pineal gland (also known as the pineal body), a structure located near the center of the brain.  He chose the pineal gland partly because of its central location, but also because he believed it to be the only brain structure not duplicated in both hemispheres (there were other reasons as well).

It is now known that the pineal gland is sensitive to light, and when stimulated releases melatonin, a hormone that is important for regulating certain diurnal rhythms.  The pineal gland may not be where mind and body meet, but it does seem to be an important component in the biological clock.

And, for that matter, he was wrong on the first point, too: close microscopic examination reveals that the pineal gland is divided into two hemisphere-like structures.  But Descartes had no way of knowing this, so we shouldn't count it against him.

• First, there was the influence of Catholic religious doctrine, which held that humans possessed souls that distinguished them from mere animals.  Today, the Roman Catholic Church has no problem with the theory of evolution, for the simple reason that, in their view, Darwin's theory applies to the evolution of bodies, whereas souls, including mind and consciousness, comes from God. The equation of soul with mind creates the mind body problem.
• There is a paradox here.  Descartes began his work (and also modern philosophy) by adopting a stance of methodical doubt without skepticism.  That is, he claimed to take nothing on authority, doubting everything until it could be proved through reason, but not doubting that true knowledge was possible.  But it turns out that Descartes didn't doubt everything.  As a devout Catholic, he didn't doubt that humans had immaterial souls that survived bodily death and went to Heaven (or someplace).  If he had doubted this proposition, he might not have come up with the doctrine of substance dualism in the first place, and he wouldn't have been stuck at his impasse. 
• But maybe not.  Setting religious doctrine aside, the fact is that dualism is the common-sense position on mind-body relations.  It has been a fixture of "folk psychology" since the time of Descartes, and probably before then.  Even without Descartes' methodical doubt, we know intuitively that we have bodies and we have minds, that these things are somehow different, and that they somehow interact. Paraphrasing the title of Paul Bloom's book, we are all "Descartes' Babies".
• But it's not mere sentimentality that keeps dualism alive. Dualism also lies at the core of modern scientific psychology, which is based on the doctrine of mentalism -- that mental states stand in relation to action as cause to effect.  If we believe that mental states cause behavior, then we had better take mental states seriously.  This is why James defined psychology as the science of mental life to begin with -- because he believed that mental states had causal powers.  A psychological explanation of behavior is in terms of the individual's mental states.  Accordingly, we need a science that will study mental structures and processes, including the neural correlates of consciousness but also including the relations between consciousness at the environment, including the sociocultural context in which individual behavior occurs.

After we came out of church we stood talking for some time together of Bishop Berkeley's ingenious sophistry to prove the non-existence of matter, and that every thing in the universe is merely ideal.  I observed, that though we are satisfied his doctrine is not true, it is impossible to refute it.  I shall never forget the alacrity with which Johnson answered, striking his foot with mighty force against a a large stone, till he rebounded from it, "I refute it thus.". 

The Explanatory Gap

However, even epiphenomenalism acknowledges consciousness -- it just thinks it's epiphenomenal.  So even epiphenomenalism can't escape what the philosopher Joseph Levine calls the explanatory gap.  Levine argued that purely materialist theories of mind, which rely solely on anatomy and physiology, cannot, in principle, explain how the physical properties of the nervous system give rise to phenomenal experience.  To illustrate his reasoning, consider the following two physicalist explanations:

• The motion of molecules causes heat.
• Electromagnetic radiation of 700 nanometers falling on the retina creates the experience of seeing "red".

• Old Mysterians were radical dualists who believed that mind was immaterial, and thus could not be explained by natural processes.
• New Mysterians, like McGinn, are materialists, but believe that consciousness can't be explained because of human cognitive limitations.
• Final Mysterians, like Schermer himself, believe that consciousness -- at least, the "hard problem" of how neural activity generates subjective experience (aka "the explanatory gap") is not a problem that can ever be solved, even in principle, not even by the most intelligent creature (or machine).  For good measure, Schermer also adds free will (how freedom of choice and action can occur in a deterministic universe) and God (supernatural beings, by definition, cannot be described by natural science; however, Schermer seems to concede that a God who acts in history, like the "Abrahamic" God worshiped by Jews, Christians, and Muslims, would be knowable from His intercession in natural events like miracles).
  

• "Psychology as a Behaviorist Views it" (1913);
  
• and Psychology From the Standpoint of a Behaviorist (1919). 

• Humans are behaving organisms.
• Behavior, and the conditions under which it occurs, can be objectively described.
• Mental states and processes do not determine what we do.
• Rather, humans and other behaving organisms simply react to environmental stimuli by means of unconditioned and conditioned reflexes.
• Psychology, as a branch of natural science, honors no dividing line between humans and other animals.
• The goal of psychology is to predict and control behavior.
• Successful prediction advances scientific theory.
• Successful control promotes human welfare

• At Berkeley, Tolman (in Purposive Behavior in Animals and Men, 1932) argued that during learning organisms acquired knowledge (such as the "cognitive map" representing a maze) which they could later use for their own purposes.  Tolman explicitly used the concept of "expectation" to describe what organisms, whether humans or rats, acquired during learning.
• At Yale, C.L. Hull (in Principles of Behavior, 1943), introduced mathematical models and other quantitative methods that gave an appearance of precision to learning theory.  Because Hull believed that responses were produced by habits energized by drives (as in his famous formula, R = H x D) he did not hesitate to make references to internal mental states such as drive, but these were always measured objectively (e.g., in terms of hours of food deprivation).

• Skinner, who had early ambitions to be a creative writer (he wrote Walden Two, a popular utopian novel) tried to be very careful with his language.  In describing the form of learning he studied, he preferred the term operant conditioning to the more familiar instrumental conditioning.  Skinner understood that organisms do not simply respond to environmental stimuli, in the manner of the classical conditioning studied by Pavlov (and, for that matter, Watson), but that their behavior also changes the environment in which it takes place.  Skinner rejected the descriptor "instrumental" because the latter term smacked of mentalism -- that somehow, the organism engaged in behavior because it was instrumental in achieving some desired state of affairs.  Instead, because the conditioned responses operate on the environment,  he chose "operant" as more neutral.  
• In his research and writing, Skinner placed a great deal of emphasis on the shaping of behavior through reinforcement.  In contrast to E.L. Thorndike, who believed that conditioning brings pre-existing, innate reflexes under stimulus control, Skinner touted shaping as a means of creating wholly new behaviors that were not part of the organism's original, innate repertoire. 

• Analytic, involving an analysis of the meaning of the words in the statement and the statement's grammatical structure.
• Synthetic, involving an empirical appeal to sense-experience -- e.g., the observation of behavior and the conditions under which it occurs.

• treating the mind as a separate component of the body, in addition to its physical parts; and
• identifying the mind with some physical part of the body, such as the brain.

• "Thinking" refers to talking to oneself; and
• "Believing" something refers to behaving as if it were true.

The bottom line in philosophical behaviorism is that we cannot identify mental states independent of behavioral states and the circumstances in which behavior occurs.  This is also essentially the bottom line for psychological behaviorism, to the extent that it talks about mental states at all.  Both psychological and philosophical behaviors seek to get around Descartes' impasse by refusing to talk about mind at all.

The Cognitive Revolution

The hegemony of the behaviorist program within psychology began to fade in the 1950s, with the gradual appreciation that behavior could not be understood in terms of the functional relations among stimulus, response, and reinforcement.

Interestingly, the deep sources of the cognitive revolution lie with some neobehaviorists who made their careers in the study of animal learning, but who critiqued salient features of the behaviorist program:

• At Berkeley in the 1930s, E.C. Tolman's experiments on latent learning (in rats, yet) showed that reinforcement was not necessary for learning to occur.
• At Wisconsin in the 1940s and 1950s, before he began his famous studies of contact comfort, love, and motherless monkeys, Harry Harlow showed that animals were motivated by curiosity to manipulate their environment, and formed learning sets -- general strategies for learning -- during learning trials, not just specific stimulus-response connections (see Love at Goon Park, a biography of Harlow by Deborah Blum, Perseus, 2002)    

These early anomalous findings were generally ignored or dismissed by practitioners of behaviorist "normal science", in a manner that would make Thomas Kuhn proud.  But in the 1960s, evidence for cognitive constraints on learning mounted rapidly:

• Robert Rescorla's studies of classical conditioning showed that associations were formed by contingency (i.e., predictability and controllability) rather than spatiotemporal contiguity;
• Leon Kamin's research revealed that role of predictability, surprise, and attention in classical conditioning;
• research on learned helplessness in avoidance learning by Stephen Maier, Bruce Overmeier, and Martin Seligman  underscored the role of controllability in instrumental conditioning, paralleling the role of predictability in classical conditioning.  

These and other studies, following on the pioneering research of Tolman and Harlow (and others) made it clear that we cannot understand even the simplest forms of learning and behavior in so-called "lower" animals -- never mind complex learning in humans -- without recourse to "mentalistic" constructions such as predictability, controllability, surprise, and attention.  Although the cognitive revolution was chiefly fomented by theorists who were interested in problems of human perception, attention, memory, and problem-solving, the results of these animal studies increased the field's dissatisfaction with the dominant behaviorist paradigm, and increased support for the cognitive revolution in psychology.

The cognitive revolution was also fostered by the development of the high-speed computer, which could serve as a model, or at least a metaphor, of the human mind:

• Computers had input systems (like punch tape and keyboards) analogous to sensation and behavior.
• They also had output systems (like printers and displays) analogous to behavior.
• Computers stored information, analogous to the knowledge stored in the mind (and Shannon's creation of information theory offered a way that knowledge could be quantified).
• Computers operated by means of control processes, analogous to mental processes of attention and thought.
• These control processes guided the flow of information from one storage structure to another, analogous to memory.  

Put concisely, the computer model of the mind showed how a physical system could, in some sense, "think" by acquiring, storing, transforming (i.e., computing) and using information.  In some respects, the computer metaphor solved the mind-body problem in favor of materialism, because it showed how a physical system could operate something like the mind.  But the fact that computers could "think" (or at least process information) raised the question of whether computers were, or could be, conscious.  Put another way, the fact that computers could "think" without being conscious suggested that maybe consciousness was epiphenomenal after all.  In any event, the fact that psychological discourse was re-opened to mental terms revived discussion of the mind-body problem, and promptly stuck us right back in the middle of Descartes' impasse -- exactly where William James had left us in 1890!

The Contemporary Materialist Stance

One difference between the beginning of the 21st century and the end of the 19th is that not too many cognitive scientists  seriously entertain dualism as a solution to the mind-body problem.  Almost everybody adopts some form of the materialist stance, for the same reason that James did -- because we understand clearly that mental states are somehow related to states of the brain.  At the same time, many people remain closet dualists, for reasons articulated earlier: religious belief, the persistence of folk psychology, or disciplinary commitment to the psychological level of analysis (that's my excuse!).  Others, however, have gotten beyond what they see as irrationality, sentimentality, or ideology to proliferate a variety of "new" materialisms -- or, at least, more precise statements of materialist doctrine.

Identity Theory

Modern versions of materialism are based on one or another version of identity theory, which holds that mental states are identical to brain states.  This position is distinct from epiphenomenalism, which holds that mental states are the effects of brain states; and it is also distinct from psychophysical parallelism, which holds merely that mental states are correlated with brain states.  

Identity theory comes in two forms: 

• In type identity theory (Smart, 1959; Feigl, 1960), mental states and brain states refer to the same thing, and differ only in description.  
• In token identity theory (Fodor, 1974; Putnam, 1975), under different circumstances the same mental event may be realized in different brain states.

Eliminative Materialism

If two terms refer to the same object, then any property of the first term must also be a property of the second.

It turns out that when you talk about mind and body, Leibniz' Law isn't necessarily upheld:

• Intentionality ("aboutness"): mental states exhibit intentionality, but brain states do not.
• Spatiality: brain states exist somewhere, but mental states do not.

Anyway, the materialist position has many current adherents, especially among those who wish to abandon folk psychology -- for example.

• Steven Stitch, in From Folk Psychology to Cognitive Science (1983). 
• Paul Churchland, in Matter and Consciousness (1984, rev. 1988).
• Patricia Churchland, in Neurophilosophy: Toward a Unified Science of the Mind-Brain (1986).

Of course, eliminative materialism is a program of reductionism -- specifically, eliminative reductionism, entailing the view that the laws of mental life discovered by psychology can (and should) be reduced to the laws of biological life discovered by biology -- and, presumably, thereafter to the laws of physics.  The implication of reductionism is that physics is the fundamental science, and everything else is just nonscientific "folk" talk, to be eliminated as soon as science allows.  As the physicist Ernest Rutherford once put it:

Never mind that Rutherford got the Nobel Prize in chemistry.  He regarded himself as a physicist, and is generally regarded as the father of nuclear physics.

Of course, such a position is unpalatable to anyone but physicists and those possessed of physics-envy (and perhaps philatelists).  Partly for that reason, the Churchlands, Patricia and Paul, have argued (together and separately) for a program of intertheoretic reductionism, which promises to preserve the legitimacy of the psychological level of analysis. 

• When the propositions and principles of a new theory mirror the propositions and principles of a old theory; and
• When the new theory gives better explanations and predictions than the old one; then
• The new theory contains the correct description of reality and is to be preferred on that ground.

The Churchlands claim not to be eliminative reductionists, but they are also very clear that "science" is better than "common sense", and their writing betrays a somewhat sneering attitude toward "folk psychology" and therefore an implication that neuroscience is not merely different than folk psychology, but that neuroscience is better than folk psychology.  They eagerly anticipate the day when mentalistic concepts are abandoned.  Again, they deny that there are eliminative reductionists, and that intertheoretic reduction preserves the legitimacy of psychology.  But they don't seem to mean it.  Their actions betray their true preference for eliminative psychology: whereas folk-psychology wants to talk about mental states, real science prefers to talk about physical states.  Therefore, they argue that psychology must be grounded in what they call

"the real-world findings of neuroscience".

The clear implication is, that mental states aren't real, and that psychology, as the science of mental life, doesn't have anything to do with the real world.  

Note that it never occurs to them that neuroscience should be grounded in the real-world findings of psychology, which would be the case if they were really interested in the symbiotic relations between two theories, or two levels of analysis.  The clear implication of their work is that the only "real world" is the world of neuroscience.  The world of neuroscience can be objectively described, while the world of mind and consciousness is the world of ether, phlogiston, and fairies. 

You get a sense of what eliminative reductionism is all about when, as quoted in a New Yorker article on their views, Pat says to Paul, after a particularly hard day at the office,

"Paul, don't speak to me, my serotonin levels have hit bottom, my brain is awash in glucosteroids, my blood vessels are full of adrenaline, and if it weren't for my endogenous opiates I'd have driven the car into a tree on the way home. My dopamine levels need lifting. Pour me a Chardonnay, and I'll be down in a minute" (as quoted by MacFarquhar, 2007, p. 69),

It's funny, until you stop to reflect on the fact that these people are serious, and that their students have taken to talking like this too. 

But really, when you step back, you realize that this is just an exercise in translation, not much different in principle from rendering English into French -- except it's not as effective. You'd have no idea what Pat was talking about if you didn't already know something about the correlation between serotonin and depression, between adrenaline and arousal, between endogenous opiates and pain relief, and between dopamine and reward. But is it really her serotonin levels that are low, or is it her norepinephrine levels -- and if it's serotonin, how does she know? Does she have a serotonin meter in her head?  

Only by translating her feelings of depression into a language of presumed biochemical causation -- a language that is understood only by those, like Paul, who already have the secret decoder ring. 

And even then, the translation isn't very reliable. We know about adrenalin and arousal, but is Pat preparing for fight-or-flight (Cannon, 1932), or tend-and-befriend (S. E. Taylor, 2006)?  Is she getting pain relief or positive pleasure from those endogenous opiates? And after going through the first five screens of a Google search, I still couldn't figure out whether Pat's glucosteroids were generating muscle activity, reducing bone inflammation, or increasing the pressure in her eyeballs. 

And note that even Pat and Paul can't carry it off. What's all this about "talk" and "Chardonnay"? What's missing here is any sense of meaning -- and, specifically, of the meaning of this social interaction. Why doesn't Pat pour her own drink? Why Chardonnay instead of Sauvignon Blanc -- or, for that matter, Two-Buck Chuck? For all her brain cares, she might just as well mainline ethanol in a bag of saline solution.  and, for that matter, why is she talking to Paul at all?  Why doesn't she just give him a bolus of oxytocin?  But no: What she really wants is for her husband to care enough about her to fix her a drink -- not an East Coast martini but a varietal wine that almost defines California living -- and give her some space -- another stereotypically Californian request -- to wind down. That's what the social interaction is all about; and this is entirely missing from the eliminative materialist reduction. 

The problem is that you can't reduce the mental and the social to the neural without leaving something crucial out -- namely, the mental (and the social). And when you leave out the mental and the social, you've just kissed psychology (and the rest of the social sciences) good-bye. That is because psychology isn't just positioned between the biological sciences and the social sciences. Psychology is both a biological science and a social science. That is part of its beauty and it is part of its tension.

In her recent memoir, entitled Touching a Nerve (2014), Patricia Churchland writes that she turned from "pure" philosophy to neuroscience when she realized that "if mental processes are actually processes of the brain, then you cannot understand the mind without understanding how the brain works".  To which Colin McGinn, reviewing her book in the New York Review of Books, asked why she stopped there.  If mental processes are actually processes of the brain, and brain-processes are electrochemical processes, then she should have skipped over the neuroscience and gone straight to physics.  And so should everyone else, including historians and literary critics.  But if historians and literary critics don't have to be neuroscientists, why is this an obligation for psychologists?

Anomalous Monism

 Interestingly, it's possible to be a materialist monist but not a reductionist.  This is the stance that the late Donald Davidson (another UCB philosopher ) called anomalous monism (e.g., in "Mental Events", 2001).  Davidson agreed that the world consists only of material entities, thus rejecting Descartes' substance dualism; and therefore, all events in the world are physical events.  At the same time, Davidson denied that mental events, such as believing and desiring, could be explained in purely physical terms.  So, his theory is ontologically materialistic (because the world consists only of physical entities) but explanatorily dualistic (because the network of causal relations is different for mental events than for physical events).

Philosophical Functionalism

Other cognitive scientists prefer token identity theories. They don't seek correspondence between mental events and brain events.  Instead, they classify mental events in terms of their functional roles and ignore the physical systems in which these functions are implemented.  Philosophical functionalism is closely related to psychological and philosophical behaviorism, in that all that matters are the functional relations between inputs and outputs -- what goes on in between doesn't much matter.  

This philosophical functionalism lies at the basis of many programs of artificial intelligence (AI), which attempt to devise computer programs that will carry out the same operations as minds.  The term was originally coined by John McCarthy (then at MIT, later at Stanford), who also invented LISP, a list-processing language which still serves as the backbone of AI.

John Searle has famously distinguished between two views of AI:

• Weak AI seeks only to implement formal psychological theories of some aspect of mental function as an operating computer program.  It is exemplified by the work of Herbert Simon and John Anderson, and is closely related to mathematical psychology.  For weak AI, programs are like minds.
• Strong AI goes further to assert that such a program actually possesses mental states.  Such a claim could only be made by a philosophical functionalist, who think that the "hardware" in which functions are implemented isn't important, only that the functions are performed.  For strong AI, programs are minds.

Although I agree with Searle, I believe it is important to distinguish yet another form of artificial intelligence.  What I call Pure AI is not concerned with psychological theories or mental states at all, but only with the task of getting machines to carry out "intelligent" behaviors.  Most robotics take this form.  So does most work on chess-playing by machine.  Deep Blue, the IBM-produced program that beat Garry Kasparov, the reigning world champion, in 1997,  is a really good chess player, but nobody claims that it plays chess the same way that humans do (in fact, we know it doesn't), and nobody claims that it "knows" it is playing chess.  It is just programmed with the rules of the game and the capacity to perform incredibly fast information-processing operations.  For pure AI, programs are programs.

Kasparov, commenting on his loss, and on chess as a goal for artificial intelligence research, writes:

The AI crowd... was pleased with the result and the attention, but dismayed by the fact that  Deep Blue was hardly what their predecessors had imagined decades earlier when they dreamed of creating a machine to defeat the world chess champion.  Instead of a computer that thought and played chess like a human, with human creativity and intuition, they got one that played like a machine, systematically evaluating 200 million possible moves on the chess board per second and winning with brute number-crunching force.  As Igor Alexander, a British AI and neural networks pioneer, explained in his 2000 book, How to Build a Mind:

By the mid-1990s the number of people with some experience of using computers was many orders of magnitude greater than in the 1960s.  In the Kasparov defeat they recognized that here was a great triumph for programmers, but not one that may compete with the human intelligence that helps us to lead our lives.

It was an impressive achievement, of course, and a human achievement by the members of the IBM team, but Deep Blue was only intelligent the way your programmable alarm clock is intelligent.  Not that losing to a $10 million alarm clock made me feel any better ("The Chess Master and the Computer" by Garry Kasparov, reviewing Chess Metaphors: Artificial Intelligence and the Human Mind by Diego Rasskin-Gutman, New York Review of Books, 02/11/2010).

In 2009, computer scientists at IBM announced Watson (named after the firm's founder, Thomas J. Watson, Sr., not John B. Watson, the behaviorist), based on the company's Blue Gene supercomputer, intended to answer questions on "Jeopardy!", the long-running television game show   If successful, Watson would represent a considerable advance over Deep Blue, because, as noted in a New York Times article, "chess is a game of limits, with pieces that have clearly defined powers.  'Jeopardy!' requires a program with the suppleness to weigh an almost infinite range of relationships and to make subtle comparisons and interpretations.  The software must interact with humans on their own terms, and fast....  The system must be able to deal with analogies, puns, double entendres, and relationships like size and location, all at lightning speed" ("IBM Computer Program to Take On 'Jeopardy!'" by John Markoff, 04/27/2009).  

The IBM Jeopardy Challenge was taped in January and aired on February 14-16, 2011 (the first game of the two-day match was spread out over two days, to allow for presentations on how Watson worked).  It pitted Watson against Ken Jennings, who made a record 74 consecutive appearances on the show in in 2004-5, and Brad Rutter, who won a record $3.3 million in regular and tournament play.  In a practice round reported in January, 2011, Watson beat them both, winning $4,400 compared to Jennings' $3,400 and Rutter's $1,200.  In the actual contest, Watson won again, $77,147 against Jennings' $24,000 and Rutter's $21,600.  The machine made only two mistakes: responding "Toronto" to a question about U.S. cities (you could see the IBM executives' jaws drop open), and repeating one of Rutter's incorrect responses.  

In addition to his linguistic skills, fund of knowledge and rapid access to it, Watson may have had a slight advantage in reaction time, though.  Milliseconds count in Jeopardy, and if so it would be interesting to see a re-run of the contest with Watson constrained to something like humans' average response latencies.

In the contest, Watson displayed a remarkable ability to parse language and retrieve information.

• Watson responded to "A recent best seller by Muriel Barbery is called 'This' of the Hedgehog'" with "What is Elegance?", clearly showing that it "knew" what this referred to.
• In the Final Jeopardy round, Watson responded to "William Wilkenson's 'An Account of the Principalities of Wallachia and Modavia' inspired this author's most famous novel" with "Who is Bram Stoker?" (so did Jennings and Rutter). 
  

In a later match, however, Watson was beaten by Rep. Rush Holt (D.-N.J.), who had been a physicist before being elected to Congress ("Congressman Beats Watson" by Steve Lohr, New York Times, 03/14/2011).  One category that Watson failed in was "Presidential rhymes" (as in "Herbert's military operations": What were Hoover's maneuvers?").  So Watson's not invincible (the machine had won only 71% of its preliminary matches, before the televised contest).  The point is not that Watson is always right, or that he wins all the time, or that he doesn't win all the time.  The point is that Watson is a pretty good question-answering machine.

It is not clear what kind of AI Watson represents.  Probably not Strong AI: nobody claims that Watson will "think", and the newspaper article uses the word "understand" in quotes.  Many of the principles undergirding Watson's software are based on studies of human cognition, which suggests that it is not Pure AI either, and might be a version of Weak AI.  IBM engineers, interviewed after the Challenge was completed, were reluctant to claim that Watson mimicked human thought processes (see "on 'Jeopardy!', Computer win Is All but Trivial" by John Markoff, and "First Came the Machine That Defeated a Chess Champion" by Dylan Loeb McClain, New York Times, 02/17/2011).  Watson's performance on pre-Challenge test runs is described by Stephen Baker in "Can A Computer Win on 'Jeopardy'?", Wall Street Journal, 02/05-06/2011).  

David Gelertner, professor of computer science at Yale (and victim of the Unabomber), wrote ("Coming Next: A Supercomputer Saves Your Life", Wall Street Journal, 02/05-06/2011):

Watson is nowhere near passing the Turing test -- the famous benchmark proposed by Alan Turing in 1950, in which a program demonstrates its intelligence by duping a human being, in the course of conversation over the Web or some other network, into believing that it's human and not software.

But when a program does pass the Turing test, it's likely to resemble a gigantic Watson.  It will know lots about the superficial structure of language and conversation.  It won't bother with such hard topics as meaning or consciousness.  Watson is one giant leap for technology, one small step for the science of mind.  But this giant leap is a major milestone in AI history.

Which strengthens the case for Pure AI.  But even if it had been only remotely successful, instead of spectacularly successful, the article notes that it will be a "great leap forward" in "building machines that can understand language and interact with humans".  For example, IBM is using an improved version of Watson's technology to develop a language-based interactive system for medical diagnosis.  

Will Watson play again?  Probably not.  Deep Blue never played chess again after defeating Kasparov (part of him is on display in the Smithsonian).  He's made his point, it's unlikely that any human will beat him when Jennings and Rutter couldn't, and Watson may have a slight mechanical advantage in reaction time at the buzzer.  

So let's give David Ferucci, who led the IBM development team, the last word (quoted in Markoff's 2011 article):

People ask me if this is HAL [the computer that ran amok in 2001: A Space Odyssey].  HAL's not the focus, the focus is on the computer on Star Trek, where you have this intelligence information seek[ing] dialog, where you can ask follow-up questions and the computer can look at all the evidence and tries to ask follow-up questions.  That's very cool.

Well, maybe not the last word.  Technological advances in artificial intelligence have led researchers to seriously consider the possibility of mind uploading -- that is, of preserving an individual's entire consciousness in what William Graziano, a neuroscientist at Princeton, calls the digital afterlife (see his 2019 book, Rethinking Consciousness: a Scientific Theory of Subjective Experience, and "Will Your Uploaded Mind Still Be You?", an essay derived from it published in the Wall Street Journal (09/14/2019).  Graziano argues that all we need to upload an entire person's mind would be to recreate, in a sufficiently powerful computer combining a connectionist architecture with knowledge from the Human Connectome Project, a complete map of the individual's 100 trillion-or-so synaptic connections among the 86 billion-or-so neurons in the human brain.  Graziano points out that the project is daunting: it took neuroscientists an entire decade to map all the connections among the 300 neurons in a crummy roundworm.  But in principle that's how it would go.  And when we flip the switch (quoting now from his WSJ essay):

A conscious mind wakes up.  It has my personality, memories, wisdom and emotions.  It thinks it's me.  It can continue to learn and remember, because adaptability is the essence of an artificial neural network.  Its synaptic connections continue to change with experience.... 

Philosophically, what is the relationship between sim-me and bio-me?  One way to understand it is through geometry.  Imagine that my life is like the rising stalk of the letter Y.  I was born at the base, and as I grew up, my mind was shaped and changed along a trajectory.  One day, I have my mind uploaded.  At that moment, the Y branches.  There are now two trajectories, each one convinced that it's the real me.  Let's say the left branch is the sim-me ad the right branch is the bio-me.  The two branches proceed along different life paths, with different accumulating experiences.  The right-hand branch will inevitably die.  The left-hand branch can live indefinitely, and in it, the stalk of the Y will also live on as memories and experiences.

The Contemporary Scene

At the turn of the 21st century, theories of mind and body are fairly represented by the various books reviewed by John Searle in The Mystery of Consciousness (1997, reprinting several of Searle's review articles in the New York Review of Books), by the replies of the books' authors, and by Searle's rejoinders.  On first glance, most theorists appear to adopt some version of materialism, holding to the view that consciousness is something that the brain does.  

Searle: Biological Naturalism

This is true even for Searle, who while a vigorous critic of the books under review nonetheless asserts that consciousness is a causal property of the brain (see The Rediscovery of the Mind, 1992).   

For Searle, consciousness emerges at certain levels of  anatomical organization.  Certainly, the human brain, with its billions and billions (apologies to Carl Sagan) of neurons, and umpteen bazillions of interconnections among them, has the complexity to generate consciousness.  This is probably true of the brains of nonhuman primates, which also have lots of neurons and neural connections.  It may also be true for dogs and cats (certainly Searle holds that his dog Ludwig is conscious).  It may not be true of snails, because they may not have enough neurons and interconnections to support (much) consciousness.  It's not true of paramecia, because they don't have any neurons at all.   

And it's certainly not true of thermostats.  Although some computers might be conscious, if they had enough information-processing units and interconnections, by and large Searle is deeply skeptical of any claim that consciousness can be a causal property of anything other than a brain.  Rather than calling himself a materialist, Searle prefers to label his view biological naturalism.  Brains produce consciousness, even if silicon chips (and beer cans tied together by string) don't.

But it's not just possessing neurons that's important for consciousness.  Anatomy alone isn't decisive.  General anesthesia, concussion, and coma render people unconscious, but they don't alter the structure of the nervous system.  However, they do alter physiology: how the various structures in the nervous system operate and relate to each other.

Although his biological naturalism is closely related to materialism, Searle strenuously opposes eliminative materialism and other forms of reductionism.  For him, consciousness cannot be eliminated from scientific discourse because objective, third-person descriptions of brain processes necessarily leave out the first-person subjectivity that lies at the core of phenomenal experience.  First and foremost, consciousness entails first-person subjectivity.  This cannot be reduced to brain-processes because any third-person description of brain-processes must necessarily leave out first-person subjectivity.  For that reason, every attempt to reduce consciousness to something else must fail, because every reduction leaves out a defining property of the thing being reduced -- in this case, the first-person subjectivity of consciousness.  (In making this argument, Searle is basically applying Leibniz's Law.)

Searle believes that first-person subjectivity is an irreducible quality of consciousness, and that this irreducible quality is produced as a natural consequence of human (and probably some nonhuman) brain processes, but he has no idea how the brain actually does it.  But, he says, that's not a philosophical problem.  The philosopher's job is to determine what consciousness is; once that's done, the problem can be "kicked upstairs" to the neurobiologists, whose job it is to figure out how the brain does it (Searle makes the same argument about free will). 

Dennett: The Zombie Bugaboo

The only way out of this problem, it would seem, is to deny the reality of consciousness -- to say that phenomenal awareness is something that does not really exist, that it is literally an illusion.  Or, at least, to say that to the extent that consciousness does exist, it is irrelevant to behavior: That is, consciousness plays no role in the operation of a virtual machine -- a software program that processes stimulus inputs and generates appropriate response outputs, just like a real machine would (which is what computational functionalists think the mind is).

This variant on functionalism appears to be Dennett's position in Consciousness Explained (1991).  According to his "Multiple Drafts" theory of consciousness, qualia are things that just happen to happen while stimulus information is being processed from the environment and responses being generated to them.  And intentionality doesn't exist either -- it's just a rhetorical "stance" we take when we talk about mind and behavior (Dennett also wrote a book entitled The Intentional Stance).

If reduction fails, as Searle says it does, then the only way out for a materialist is to deny the reality of consciousness. For Dennett, consciousness is literally an illusion, only one of many "multiple drafts" of phenomenology, and plays no role in the operation of a virtual machine.  When we ask people what they are thinking, one of these "drafts" pops into mind.  But that draft has no privileged status.  It's just one of many things running along in parallel connecting inputs and outputs.  For Dennett, only behavior matters, and only behavior actually exists.  Conscious mental states don't have any special significance, and they don't play any causal role in what is going on.  For Dennett, it's behavior that matters, not phenomenology, because: 

• Only behavior can be verified empirically.
• Only behavior can be studied scientifically, with the necessary third-person objectivity.
• And only behavior actually exists (because only behavior can be verified).

Those who recognize in Dennett the earlier arguments of John B. Watson are onto something.  For Dennett, the evidence for consciousness is to be found in behavior.  First-person subjectivity doesn't count because it's an illusion -- just one of many "multiple drafts" of phenomenology, none of which have any causal impact on behavior or the world.

Dennett's work is especially provocative for his reconstrual of what might be called the zombie bugaboo.  Philosophical discussions of consciousness often invoke the notion of zombies, defined as molecule-for-molecule replicas of humans, who can do all the things that humans do except that they do them unconsciously.  Thus, consciousness ostensibly differentiates humans from zombies.  But for Dennett, we're zombies too -- not because we're unconscious, but because zombies are conscious too.  From his perspective of radical functionalism, because their behavioral functions are indistinguishable from that of humans, zombies must have the same internal states as humans, including internal states of consciousness.  If we're conscious, so are they.  We're all zombies now.

Dennett recently reaffirmed his views in From Bacteria to Bach and Back: The Evolution of Minds (2017; given a surprisingly even-handed review by Thomas Nagel in "Is Consciousness an Illusion?", New York Review of Books, 03/09/2017).  Dennett, of course, says "yes" (Nagel, of course, would insist "No", which is why the tone of the review is so surprising.  Maybe Nagel got tired of fighting).  Adopting terms introduced by Wilfrid Sellars, a British philosopher, Dennett argues that there is a "manifest image" of the world as it appears to us, and there is a "scientific image" of the world consisting solely of particles in fields of force.  The scientific image is real; the manifest image is literally an illusion.  And it's the manifest image that we're consciously aware of.  The manifest image is a "user-illusion designed by evolution to fit the needs of its users".  And so is consciousness.  Dennett claims that our competence, or our ability to operate adaptively in the world, does not depend on our ability to comprehend what we are doing -- just like a computer (here Dennett invokes the Turing machine) can competently perform calculations without having any idea what it is doing.  But wait: if consciousness is a user-illusion designed by evolution, what are the user's "needs" that it serves?  For Dennett, consciousness allows us to monitor ourselves, and deal with other people.  The widely shared illusion of consciousness allows us to predict each other's behavior, as well as our own.  But consciousness doesn't play any causal role in this behavior.  It's an afterthought.  It's an illusion. 

For Dennett, first-person experience has no particular authority, and our attributions concerning our own behavior have no privileged status over other people's attributions.  They're just opinions:

Curiously, then, our first person point of view of our own minds is not so different from our second-person point of view of others' minds: we don't see, or hear, or feel, the complicated neural machinery churning away in our brains but have to settle for an interpreted, digested version, a user-illusion that is so familiar to us that we take it not just for reality but also for the most indubitable and intimately known reality of all.

About which Nagel concludes:

About the true nature of the human mind, Dennett is on one side of an old argument that goes back to Descartes.  He pays tribute to Descartes, citing the power of what he calls "Cartesian gravity", the pull of the first-person point of view; and he calls the allegedly illusory realm of consciousness the "Cartesian Theater".  The argument will no doubt go on for a long time, and the only way to advance understanding is for the participants to develop and defend their rival conceptions as fully as possible -- as Dennett has done.  Even those who find the overall view unbelievable will find much to interest them in this book.

A profile of Dennett by Joshua Rothman focuses on Dennett's personal life, as well as his ongoing debate with David Chalmers New  ("A Science of the Soul", New Yorker, 03/27/2017; UCB's Terrence Deacon also makes a cameo appearance).  . As an undergraduate at Harvard, Dennett studied with the philosopher W.V.O. Quine (recall, from the lectures on "Introspection", Dennett's essay on "Quining Qualia").  His dissertation advisor at Oxford was Gilbert Ryle, whose famous book about the mind was entitled  The Ghost in the Machine   A wonderful detail: Dennett, a talented wood-carver (and many other things, besides) fashioned a Russian "maryoshka" nested doll of Descartes; open up Descartes, and you find a ghost inside; inside the ghost, you find a robot.  That pretty much captures Dennett's view.

Link to a TED talk by Dan Dennett.

Eccles: Dualism Redux

In the contemporary scene, materialism dominates discussions of consciousness, as arguments rage about whether materialism offers the correct approach to the mind-body problem, and if so which version of materialism is the correct one.  But materialism doesn't exhaust the possibilities, and dualism has been revived as well. 

For example, Karl Popper, the distinguished philosopher of science, and Sir John Eccles, the Nobel laureate in medicine (who I think is largely responsible for this work) have proposed a tri-aspect interactionism which distinguishes among three "worlds":

• World 1 (of physical objects);
• World 2 (of mental activity); and 
• World 3 (of the abstract objects of thought, such as mathematical concepts, scientific theories, works of literature and music). 

According to P&E, World 3 is governed by normative principles such as the rules of logic, literary and musical forms (such as the sonnet and the sonata).  These principles are objectively valid, regardless of whether anyone follows them.  But, they argue, no physical system in World 1 can appreciate the rules of World 3.  Therefore, World 2 is needed as an intermediary -- something that understands the content of World 3 and can apply that understanding in World 1.  Because World 2 must be substantially different from World 1, P&E essentially argue for dualism -- the world of the mind is different from the world of the body.

In arguing against materialism, and for a substantial (no pun intended) difference between mind and body, Popper and Eccles have argued for a dualism that matches our introspections: mental states are not the same as physical states, yet mental states can affect the physical world.  Yet it's not clear that Popper & Eccles have cleared Descartes' impasse.  In fact, they may have made the situation worse, by giving us three worlds to worry about instead of just two!

And, in fact, there may well be a third world to worry about.  Many prominent mathematicians consider the laws of mathematics to be genuine discoveries, not mere products, of the mind -- in other words, even though lacking physical reality, they exist independently of mental activity, somewhat along the lines of Plato's forms.  Among these theorists is UCB's own Prof. Edward Frenkel, legendary instructor of a lower-division course in multivariable calculus,as represented in his book, Love and Math: The Heart of Hidden Reality (2013).  As summarized by Jim Holt ("A Mathematical Romance", New York Review of Books, 12/05/2013):

"How can it be," Einstein asked in wonderment, "that mathematics, being after all a product of human thought independent of experience, is so admirably appropriate to the objects of reality?"  Frenkel's take on this is very different from Einstein's.  For Frenkel, mathematical structures are among the "objects of reality"; they are very bit as real as anything in the physical or mental world.  Moreover, they are not the product of human thought; rather, they exist timelessly, in a Platonic realm of their own, waiting to be discovered by mathematicians.  The conviction that mathematics has a reality that transcends the human mind is not uncommon among its practitioners, especially great ones like Frenkel and [Robert] Langlands, Sir Roger Penrose and Kurt Godel.  It derives from the way that strange patterns and correspondences unexpectedly emerge, hinting at something hidden and mysterious.  Who put those patterns there?  They certainly don't seem to be of our making.

• Among these objects are those of pure mathematics, much as with Popper and Eccles.  
• The statement 7+8=15 is true, but the numbers 7, 8, and 15 do not appear to be part of the natural world of particles moving in fields of force.
  
• Nor do reasons -- or at least the reasons which Scanlon calls normative reasons, which are different from explanatory reasons.
• Explanatory reasons simply explain, in psychological terms, why some actor engaged in some action. 
  
• To use an example from Scanlon, The driver turned the wheel because he wanted to avoid hitting a pedestrian.
• This statement is true only if it is an correct explanation of the driver's behavior.  He could have had other reasons, like a desire to hit someone else.
  
• Normative reasons are reasons about what people should do or think.
• Again, an example from Scanlon: For the driver of an automobile, the fact that the car will strike a pedestrian is a reason to turn the wheel.
• This is true regardless of whether the driver actually turns the wheel.  This reason is relevant to the driver's behavior only if the driver recognizes the reason and is motivated to conform to it.  But the reason exists nonetheless.

Scanlon argues that normative reasons are not simply derived from something else, like a person's attitudes or desires that might justify his behavior.  In fact, the driver's behavior is justified by appeal to something that exists independent of his or anyone's attitudes.  This "something" is a normative reason.  (For an exposition of Scanlon's theory, see "Listening to Reason" by Thomas Nagel, New York Review of Books, 10/09/2014.)

Normative reasons play a big role in analyses of morality, which would take us far beyond the scope of this course.  My only point in mentioning Scanlon's work is that Popper and Eccles are not alone in thinking that there are things which have an observer-independent existence but which are not part of the spatially and temporally bounded reality explored by the natural sciences.

Chalmers: Closing the Explanatory Gap?

A rather different take on the problem of consciousness is offered by David Chalmers, in a provocative (and peculiar) blend of materialism and dualism which neither denies that consciousness exists nor reduces consciousness to neural functions.  Chalmers' work has been extremely provocative: Alone among the philosophers writing on consciousness, he has inspired a whole book of critical responses, and some people think he's actually solved the mind-body problem.  Certainly he does.

• Chalmers begins by distinguishing between two categories of problems of consciousness. "Easy" Problems involve aspects of structure and function that can be investigated by the standard procedures of cognitive science, and explained in terms of computational or neural processes.  
• Examples include the discrimination and categorization of environmental events, information integration, attention, and the distinction between sleep and wakefulness.  Essentially, these problems constitute the subject matter of psychology. 
  
• "Hard" Problems resist standard methods, and are exemplified by the question of how computational and neural processes give rise to phenomenal experience, qualia, aboutness, and the like.  

Chalmers argues that the easy problems are easy because they concern abilities and functions, such that explanation need only specify the proper mechanism.  The hard problem is hard because it remains after all the mechanisms are accounted for, and goes beyond function to experience -- precisely what Levine (1983) called the explanatory gap.

Chalmers asserts that some extra ingredient is needed to close the explanatory gap.  Because everything in physical theory is compatible with the absence of consciousness, we need a new fundamental assumption to get conscious experience out of a physical system.  Chalmers closes the explanatory gap by proposing that experience itself is a fundamental feature of the physical world, just like mass, space, and time.  Specifically, Chalmers proposes that experience is an aspect of information.

In information theory, as developed in the 1940s and 1950s by Claude Shannon at Bell Laboratories, information is defined as the amount of change that occurs between two states of a system. 

• If a system can take two states, such as 0 and 1, then it embodies 1 bit of information.
• If a system can take four states, such as 00, 01, 10, and 11, then it embodies 2 bits of information.

In his double-aspect theory of information, Chalmers asserts that information has a physical aspect (which embodies a difference between one or more physical states of the world) and an experiential aspect (which gives rise to consciousness). Thus, information physically represented in the brain (e.g., by a pattern of neural firings) naturally gives rise to conscious experience, because conscious experience is one aspect of the state of being informative.  But human brains aren't the only things that embody information:

• nonhuman brains do, so therefore nonhuman brains have consciousness;
• computers do, so therefore computers are conscious; and
• thermostats do, so therefore thermostats have consciousness (though arguably it is probably not a very interesting consciousness).

Although this move might strike some observers (including myself) as a deus ex machina, Chalmers insists that everything that embodies information is conscious, because information always has a dual aspect. Still, he never quite addresses the "hard question" of just how information translates into conscious experience.

1. In the final analysis, Chalmers's theory is a restatement, in the terms of modern information theory, of Morton Prince's old mind-stuff theory.  Recall that Prince's solution to the mind-body problem was to assume that every physical object contained some amount of "mind-stuff", and that when enough of it accumulated, the physical object became conscious.  It's a position known as panpsychism -- the idea that consciousness pervades the physical Universe, and inheres not just to humans, not just to humans' close evolutionary relatives, not just to mammals, not just to animals, but also to plants and rocks.  It's a neat solution to the mind-body problem.  But it's a solution that William James tore apart in his Principles (despite the fact that Prince was a friend and colleague).  and it's a solution that should strike us as deeply unsatisfying, on two grounds.
2.  It really is a deus ex machina.  It comes out of the blue, clearly concocted to solve the mind-body problem.
3. It doesn't actually solve the mind-body problem.  To say that any physical system that represents information is conscious doesn't tell us how that state generates consciousness -- leaving us with Chalmers's "hard problem" unsolved after all.
   

Panpsychism may seem like a desperate measure, but it has its proponents.  One of these is Philip Goff, a British philosopher, in Galileo's Error: Foundations for a New Science of Consciousness (2019; reviewed by Julian Baggini in "Of Mind and Matter", Wall Street Journal, 01/04/2020).  In my teaching, I have argued that Immanuel Kant (1724-1804) wrongly characterized psychology as an impossible science because the mind, being composed (as he thought) of an immaterial substance, can't be measured.  Goff takes us back another 150 years, before Kant, and even before Descartes (1596-1652), whose Meditations (1641) gave us the mind-body problem to begin with, to Galileo (1564-1642), who made a similar argument in The Assayer (1623), his seminal essay on scientific method.  For Galileo, measurement is the essence of the scientific method (I'm paraphrasing here: "Nature is a book written in the language of mathematics").   If it can't be measured, it can't be subject to scientific investigation.  And Galileo believed that sensory qualities couldn't be measured -- so that settled that, even before psychology got its name (that was with Christian Wolff, in his Psicologia empirica of 1732 and his Psicologia rationalis of 1734)!  Anyway, because subsequent science embraced Galileo's position on measurement, Goff argues, it rendered itself unable, in principle, to account for consciousness.  Goff rejects dualism, because it can't explain how minds can affect bodies we're talking about free will here); he also rejects materialism, because the implication that consciousness is an illusion strikes him as absurd.  But he embraces panpsychism, by asserting that consciousness is a fundamental property of matter.  Galileo's error was to assume that consciousness could not be part of the material world.

But really, panpsychism doesn't solve the hard problem of consciousness, which is what Chalmers and Goff, and other panpsychists, intend.  It just kicks the hard problem down one level of explanation.  So now, instead of asking neuroscientists to figure out how the brain generates consciousness, instead we ask physicists to figure out how thermostats and solar systems do it.  Doesn't seem like much of a solution to me.

Self-Organizing Complex Systems

A related idea comes from UCB's own Terrence Deacon, a cognitive anthropologist who has long been concerned with the problem of how life, not to mention sentient life, could have evolved from inanimate matter.  Deacon offers a solution in Incomplete Nature: How Mind Emerged from Matter (2012).   The essential ingredient for life, in his view, is the emergence of a complex system that is teleodynamic -- autonomous, self-maintaining, temporally stable, and resistance to entropy.  One such system is the biological cell -- the basic building-block of life.  And the brain, of course, is a complex system of cells.  Whereas Chalmers believes that consciousness is a property of any physical system that embody information, Deacon suggests a somewhat more restrictive view: that the physical system must consist of biological cells, which have the five properties just listed.  Deacon does not quite verge on panpsychism, as Chalmers does, ascribing consciousness to things like computers and thermostats.  But he comes close, asserting that there is sentience -- consciousness -- wherever we find self-preserving dynamic systems.  That includes brains, but also cells (like individual neurons) and molecules. 

But then he edges closer to panpsychism when he asserts that it's not just organic material that can be sentient.  Any entity that is self-organizing will have consciousness, precisely because -- you have to forgive the pun here -- it organizes its self.  That includes biological matter, of course, but presumably it can also include inorganic matter, like perceptrons.  Anyway, Deacon believes that while neurons and the things made of them are conscious, you don't need neurons to be conscious.  All you need is a self-organized, autonomous, temporally stable entity that resists entropy. The key to the argument seems to be that, in order to be conscious, you have to have some sort of self, and self-organizing, teleodynamic systems have a self by definition.

Still, when you ask how teleodynamic systems attain consciousness, simply by virtue of being teleodynamic, all that can be said is: "it's not clear".

For a critique of Deacon's ideas, see "Can Anything Emerge from Nothing?" by Colin McGinn, New York Review of Books, 06/07/2012. An exchange of views between Deacon and McGinn was published in the NYRB on 10/11/2012.

Damasio: Self and Mind

Much the same sort of argument seems to be made by Antonio Damasio, a neurologist who has written frequently on consciousness and related topics. Damasio focuses on the essentially subjective nature of consciousness, and notes that, in order to have subjectivity, or first-person experience, you have to have a sense of self -- a first person in the first place!. So, he tries to figure out which parts of the brain are critical for the sense of self. The general argument of Damasio's book, Self comes to Mind (2011) goes as follows:

• The brain is the physical basis of mind, but for the most part the brain operates unconsciously. There are mental states, and the brain produces them, but these are not accessible to consciousness. Something has to be added to to brain activity in order to create conscious experience.
• These unconscious mental states take the form of images, or percepts, of various objects. These images are created by mapping objects onto patterns of neural activity.
• Conscious mental states are experienced as personal -- as, in some sense, owned by the creature who has them, and not shared directly with anyone else. So, consciousness requires some sense of self as the "owner" of experience or "protagonist" of action.
• The self also consists of neural mappings. In fact, there are three different components of the self. 
• The protoself consists of mental images of the body.
• The core self consists of a sequence of mental images that represent the body in action.
• The autobiographical self represents various facts about the self and its social context, and constitutes our sense of personhood and identity.
• Conscious awareness -- subjectivity -- occurs when neural mappings of objects and events become connected with neural mappings of the self.
• Actually, consciousness entails three things: wakefulness, mental operations, and a self.
• Each of these components of consciousness is generated by activity in the brain stem, the thalamus, and the cerebral cortex. However, there is no isomorphism between the three components of consciousness and the three components of the brain. Each of these brain structures makes its own special contribution to each of the components of consciousness.

For a critique of Damasio's ideas, see "The Mystery of Consciousness Continues" by J.R. Searle, New York Review of Books, 06/09/2011. An exchange of views with Barclay Martin was published in the NYRB 09/29/2011.

Link to a TED talk by Anthony Damasio.

Tononi and Koch: Differentiated and Integrated Information

Both Chalmers and Deacon exemplify what might be considered a search for the "secret sauce" -- an attempt to identify the one thing that makes the difference between consciousness and unconsciousness.

• For Chalmers, the secret sauce is information: For any physical system that represents information, to the extent that it represents information, that system is conscious.
• For Deacon, the secret sauce is self-organization. Any system that represents information, and is also self-organizing, has what it takes to be conscious. So while for Chalmers a thermostat is conscious, for Deacon it isn't, because it isn't self-organizing.

Earlier, Koch joined Francis Crick in proposing that the key to consciousness was the synchronized firing of neurons at 40 cycles per second (hertz). Koch subsequently revised his view, stating only that 40-hertz firings were necessary for the formation of integrated percepts.

At roughly the same time, Tononi joined Gerald Edelman in proposing the dynamic core theory of consciousness, which proposed that consciousness is a product of both highly integrated and highly differentiated information (Tononi & Edelman, 1998; Edelman & Tononi, 2000).

More recently, and working both together and separately, Tononi (in "Consciousness as Integrated Information: A Provisional Manifesto, 2008; see also Balduzzi & Tononi, 2008) and Koch (in The Quest for Consciousness, 2004; Consciousness: Confessions of a Romantic Reductionist, 2012; The Feeling of Life Itself: Why Consciousness is Widespread but Can't be Computed, 2019) have refined the dynamic core theory of consciousness in a new information integration theory.

For a review of Koch's 2004 book, see "Consciousness: What We Still Don't Know" by J.R. Searle, New York Review of Books, 01/13/2005; also an exchange between Searle and Stevan Harnad, NYRB 06/23/2005.

For a review of Koch's 2012 book, see "Can Information Theory Explain Consciousness?" by J.R. Searle, New York Review of Books, 01/10/2013; also an exchange between Searle and Koch and Tononi (writing together), NYRB 02/.

Baars: Global Workspace Theory

Baars, a cognitive neuroscientist who is affiliated with Crick's Neurosciences Institute, has proposed a theory of consciousness known as Global Workspace Theory (GWT; 1988).

GWT holds that the function of consciousness is to combine a number of different cognitive processes into "a single coherent experience".  It does this by making the contents of various subordinate, modular information-processing functions available to each other -- so that, for example, we can think about what we're perceiving, remember similar experiences from the past, control our emotional reactions to them, and plan what to do next.  These different flows of information all come together in a global workspace -- hence the name of the theory.  This global workspace is, essentially, working memory. 

In a nod to neuroscience, Baars's global workspace is also known as the global neuronal workspace, and the neural version of GWT is known as Global Neuronal Workspace Theory  (GNWT; Dehaene, Consciousness and the Brain, 2014).  Based on recordings of event-related potentials elicited by subliminal and supraliminal stimuli, Dehaene has suggested that the neural substrate of GW is in the frontal lobes, and specifically in prefrontal cortex.  On the other hand, in a 2018 essay, Christof Koch identified a large swath of posterior cerebral cortex -- the so-called posterior hot zone -- as the neural basis of the GW, and thus (assuming that Baars is right) of consciousness ("What Is Consciousness?", Scientific American, 06/2018).  For more details, see the lectures on The Mysterious Leap from the Body to the Mind.

Morsella: Supramodular Interaction Theory

Similarly, Morsella (2005) has proposed a Supramodular Interaction Theory.  Following Fodor's (1983) doctrine of modularity, Morsella argues that the outputs of different low-level modules feed into "supramodules" that perform higher-level tasks, leading to overt behavior executed by various musculo-skeletal systems.  For example, one module (processing information from a low-level blood-sugar module) might give rise to feelings of hunger, initiating a search for food.  Another supramodule might take information from the skin senses, generate feelings of pain, and initiate escape or avoidance behavior.  The activities of these supramodules may conflict, as when the act of eating causes pain (because the eater has a toothache?).  The function of consciousness, in this view, is facilitate the resolution of conflict between different musculo-skeletal systems. He argues that only the skeletal musculature is controlled by conscious processes -- everything else done by various effector systems goes on unconsciously.  Thus, it is the phenomenal awareness of conflict between action tendencies that permits the conflict to be resolved.

The Turn Toward Panpsychism

One of the most interesting contemporary approaches to the mind-body problem involves a revival of panpsychism -- the idea that consciousness pervades the physical universe.  Panpsychism was implied in Morton Prince's notion of mind-stuff, and more explicitly implicated in David Chalmers's idea that consciousness is a property of any information system.  Panpsychism was a prominent topic at the 2016 conference on "The Science of Consciousness", and also ran through a series of talks on "The Rise of Human Consciousness" presented at the New York Academy of Sciences on May 23, 2016.

Most prominently, Giulio Tononi (Biological Bulletin, 2008; Oizumi, Albantakis, & Tononi, PLoS Computational Biology, 2014), a neuroscientist at the University of Wisconsin (and former student of Gerald Edelman), has taken a page from Chalmers's book and proposed an Integrated Information Theory (IIT) which holds that consciousness is a property of any physical system that has the ability to store and integrate a large amount of information.  Tononi has also proposed a mathematical unit, phi, to measure how much information integration -- that is, how much consciousness -- any entity has.  Information integration gives a physical system intrinsic causal power -- the ability of elements of the system to exert causal effects on each other through feed-back and feed-forward loops.  Any system with such a reentrant architecture will be conscious; and only such a system can be conscious.

The details of how phi is calculated are beyond my competence to explicate, so I'm not going to even try. 

It is important to note one implication of IIT, at least as noted by Christoph Koch in a recent essay ("Proust Among the Machines", Scientific American, 12/2019).  In IIT, consciousness is not just a property of the brain.  It is the property of any physical system that has achieved a sufficient degree of phi -- including computers (and robots).  In this respect, Koch believes that it differs from GNWT, its chief competitor (at least as of 2020).  In Koch's view, GNWT insists that consciousness emerges from neural systems that have certain properties: the brains of mammals have these properties; but other animals, like jellyfish, do not: like plants, and computers they are not and cannot be conscious because they lack the appropriate neural architecture.  But he might not be right.  In a recent exposition of GNWT, Dehaene et al. (Science, 2017) are clear that as they define it, any physical system with certain computational properties also has consciousness. 

Unfortunately, phi has some problems.  Scott Aaronson, a computer scientist, has calculated in a 2014 blogpost that, following Tononi's formula, a DVD player (technically, a component of a DVD player that corrects for errors induced by flawed media) has more consciousness than a human brain.  Aaronson argues that such a result contradicts our intuition that, however much consciousness a DVD player has, it can't have more consciousness than a typical human.

In reply (also posted to Aaronson's blog), Tononi noted that Aaronson's calculations depended on an earlier version of the Phi formula -- but then conceded that the latest version, known as Phi 3.0, confirms Aaronson's result, and actually increases the superiority of the DVD player.  More important, Tononi isn't bothered by the notion that a DVD player might have consciousness.  He points out, correctly, that empirical science has a long and distinguished history of correcting our intuitions about the nature of things.  And, in fact, he goes on to assert that an even simpler structure, a sufficiently large 2-dimensional grid of logic gates, when physically implemented, would have a large value of phi -- and, thus, be conscious.

Aaronson, in his rejoinder, calls this the "Copernicus of Consciousness Argument", after the astronomer who argued that, despite our intuitions, the sun does not revolve around the earth.  He insists that any formula that indicates that a "simple expander graph" has consciousness doesn't comport with common sense.  Or, as he puts it "When the clock strikes 13, it's time to fix the clock...".

Link to Aaronson's blogpost, which contains a very math-heavy explication of phi.

Link to Tononi's reply.

Link to Aaronson's rejoinder.

Still, the important point in this context is that Tononi has seriously proposed that nonliving matter, including any arbitrary physical system, can be conscious, so long as it can store and integrate a large amount of information. 

• Information is defined in terms of an entity's repertoire of accessible states, or its ability to store large amounts of information.
• Integration is achieved to the extent to which the system is decomposable into independent parts.
• If the system is not decomposable, then it is conscious as a unified whole.
  
• If an information system is decomposable into independent parts, then each of these parts will be conscious to some degree.

Despite this criticism, Max Tegmark (2015), a cosmologist at MIT (and UC Berkeley PhD), has embraced and expanded IIT.  Tegmark actually proposes that consciousness is a state of matter, like solid, liquid, gas, or plasma.  He calls it "perceptronium", in homage to Minsky and Papert's perceptrons -- early versions of self-organizing "neural networks" that were capable of learning.  Tegmark traces several steps that lie between the usual states of matter and states of consciousness.

1. The substance ought to be able to store information -- that is, hold in a particular state -- for a relatively long period of time.  For this reasons, solids are more likely than liquids or gases to have consciousness.  If such a substance is also easy to write to and read from, it has the qualities of memory -- the "first warmup step toward consciousness" (2015, p. 3).
2. A substance with consciousness also has to have what he calls computronium -- that is, the computer-like ability to process information. 
   
3. Finally, a conscious substance will have perceptronium, which allows for subjective self-awareness.

Putting all of this together, Tegmark argues that any physical system will have consciousness if it satisfies six -- but really only four -- principles.

1. Information: substantial storage capacity.
2. Dynamics: information-processing capacity.
3. Independence from the rest of the physical world.
4. Integration of its "nearly independent parts".
5. Autonomy is the combination of dynamics and independence (thus it doesn't really count as a fifth principle).
6. Utility is not necessary (and thus doesn't really count as a sixth principle), but such a principle would explain why conscious systems evolved.

In some ways, Tegmark's version of panpsychism is preferable to Chalmers's and Tononi's, because it puts limits on what can be conscious.  As he writes, clocks and diesel generators tend to exhibit high autonomy, but lack substantial information storage capacity" (2015, p. 4).  Therefore, they are unlikely to be conscious.  Still, it seems to me that any materialist theory that has to solve the mind-body problem by invoking a new substance known as "perceptronium" might just as well cite Morton Prince and talk about "mind-stuff".  It gets us to pretty much the same place.  Which is nowhere.

Nevertheless, IIT, like GWT/GNWT, remains very popular.  In 2023, an international consortium of investigators(known as the Cogitate Consortium), including Stanislas Dehaene (a proponent of GNWT), and Christoph Koch and Giulio Tononi (proponents of IIT), reported on an "adversarial collaboration" in which proponents of the two theories agreed on terms for a comparison.  For example:

1. IIT postulates that conscious contents will be instantiated in posterior brain regions (the "hot zone"), while GNWT predicts that activity in the prefrontal cortex is necessary for consciousness (remember, GWT essentially identifies consciousness with working memory, and GNWT identifies working memory with the executive functions localized in prefrontal cortex).
   
2. IIT predicts sustained activity in the "posterior hot zone" so long as the stimulus is consciously represented, while GNWT predicts an "ignition" of activity in prefrontal cortex at the onset and offset of the stimulus.
3. IIT predicts "short-range connectivity" within posterior cortex between low-level sensory areas (e.g., primary visual cortex) and high-level category-specific areas in (e.g., the fusiform face area), while GNWT "long-range connectivity" between these posterior areas and pre-frontal cortex.
   

The study used fMRI, magnetoencephalography, (mEEG), and electrocorticography (ECG), to record brain activity while subjects viewed suprathreshold (thus conscious) visual stimuli presented for various durations.  The paper was posted (06/23/2023) in advance of peer review, but in the authors' (there are lots and lots of them) view, the results came out about evenly split between the theories.  There was brain activity following stimulation (which I suppose counts as a plus!).

1. In accordance with IIT, consciousness was accompanied by posterior activation, independent of activation in prefrontal cortex;
2. but contrary to IIT, there was little evidence for synchronized activity between low- and high-level posterior areas.
3. In accordance with GNWT, there was "ignition" in prefrontal cortex at the time of stimulus onset;
4. but contrary to GWNT, there was no further "ignition" in prefrontal cortex at the time of stimulus offset.

Here's a link to an article in Science describing the adversarlial collaboration: "Consciousness hunt yields results but not clarity" by Elizabeth Finkel (06/30/2023). 

And, for good measure, Here's a link to a report of the study, posted prior to any peer review (usually a nursery-school no-no).

Still, the panpsychist implications of IIT bother some people (not without justification, we should note).   Just about the time that the Cogitate Consortium paper was posted online, another group (Including Bernard Baars, Patricia Churchland, and Daniel Dennett) posted a declaration that IIT was "pseudoscience".  It's not entirely clear why IIT is to be labeled "pseudoscience", though the authors of the declaration are clearly perturbed by its panpsychic implications.  Admittedly, any theory that predicts that a DVD player is conscious is likely to be wrong.  Interestingly, they are also concerned about the medicolegal implications of IIT: it predicts that fetuses are conscious even at very early stages of development (with implications for the legalization of abortion at early stages of pregnancy), and might be used to impute consciousness inappropriately to coma patients (with implications for the rights of family members to discontinue life support).  But that doesn't make it pseudoscience.  There's a difference between being wrong, as I think IIT is, and being pseudoscientific. 

See "Consciousness theory slammed as ‘pseudoscience’ — sparking uproar" by Mariana Lenharo, Nature, 2023).

And here's a link to the letter castigating IIT as "pseudoscience".

Zap and Zip

Issues of panpsychism aside, Tononi's integrated information theory of consciousness has inspired the development of a strictly biological measure of consciousness, which does not rely on either self-report or behavior, and which can be applied to any species that has a brain: the Perturbational Complexity Index (PCI; Massimini et al., 2005).  The general idea behind the PCI is that conscious processing involves the complex integration of various areas and systems of the brain, especially in the thalamocortical system.  To obtain a value for PCI, Casali et al. first deliver a pulse of transcranial magnetic stimulation (TMS) to the brain: this is the "Zap" of what has been called Zap and Zip.  The the electrical potentials evoked by this stimulus are recorded by EEG by means of a dense array of electrodes.  These recordings, taken from various regions of the cerebral cortex, are then aggregated by an algorithm derived from data-compression techniques (think zipfiles or MP3s) derived from computer  science: this is the "Zip" of Zap and Zip.  To take an analogy: imagine that a an organ sounds a chord in a great cathedral, and we measure how long it reverberates, and how the echo reaches, and is passed around, the various alcoves and chapels that surround the main nave.  PCI is kind of like that.  But there isn't just one "zap".  As with event-related potentials, there are dozens, even hundreds, of zaps.  Each individual subject then receives an aggregate score, ranging from 0 to 1; the investigator may also record the maximum individual PCI score, known as PCI_max. 

In an initial study, Massimini, Tononi, and their colleagues Tononi and Massimini showed that PCI could distinguish between waking and sleeping in normal subjects (Massimini et al., Science, 2005).  A later study showed that PCI could distinguish between subjects who were awake, asleep (NREM sleep), or anesthetized (by various agents), as well as brain-injured patients in various stages of coma (Casali et al., Science Translational Medicine, 2013).  A more recent study found that a PCI value of .31 had 100% specificity and sensitivity in discriminating between subjects who were conscious and those who were unconscious (i.e., in NREM sleep or anesthetized).  That is, 100% of conscious subjects were identified as conscious were indeed conscious (including those who were dreaming), and 0% were unconscious; and 100% of those it identified as having lost consciousness were indeed unconscious (including those in dreamless sleep), and 0% were conscious.  Applying this criterion to patients in various stages of coma, PCI successfully identified all of those who were in the "minimally conscious state", and suggested that a minority of those diagnosed as in the "vegetative state" possessed, at least minimal levels of consciousness.  The upshot of all this work is that Tononi, Massimini, and their colleagues believe that they have created a biological index of consciousness that does not rely on either self-reports or behavioral responsiveness; and that this index is based on Tononi's Information Integration Theory of consciousness.

Back At Descartes' Impasse

At the end of this tour, we appear, like James, to be left stranded at Descartes' impasse. As scientists, we must embrace materialism: the mind is what the brain does, and science must be based on public observation. But our experience as conscious beings inclines us toward dualism: the intuition that mental states are different from physical ones, and the belief that mental states have causal efficacy, and the view that there is something about consciousness that is essentially private and subjective.

As a result, philosophers and other cognitive scientists will continue to argue about the mind-body problem.

• Some, like Dennett and the Churchlands, will encourage us to give up our prescientific folk-psychology.
• Others, like Searle, will ask us to develop a materialism without reductionism.
• Some, like Chalmers, will promote a peculiar blend of materialism and dualism.
• Some, like Popper and Eccles, will attempt to revive a thoroughgoing dualism; and
• Some, like Nagel and McGinn, will lapse in to Mysterianism.

This situation, where after more than 350 years of hard thought we are pretty much back where we started, suggests that Searle is probably right that the mind-body question has been poorly constructed, and that to make progress we have to break out of received categories of mind and body, dualism and monism, idealism and materialism. We know that the mind is a function of the body: the mind is what the brain does, and our mental capacities are the product of biological evolution.  At the same time, dualism has its pleasures.  It comports with our experience of ourselves as sentient beings with free will.  Psychologists, who are primarily concerned with the nature of mental life, are by their very nature dualists of a sort.  Moreover, dualism keeps certain interesting problems alive, such as the problem of free will and the possibility of psychosomatic interactions.

Therefore, one project for neuroscience is to determine how the brain does it -- that is: 

• How does the brain accomplish mental functions, such as sensation and perception, attention and memory, reasoning and problem-solving, judgment and decision-making, emotion and motivation, and all the rest? 
• What is the neural basis of conscious experience?  What are the anatomical structures that are responsible for consciousness?  Are there modules or brain systems for consciousness, like the modules or systems responsible for vision or language?  And what are the physiological processes that are responsible for consciousness.  Are there certain patterns of neural activity that produce consciousness, like the patterns that code for stimulus intensity or visual hue?
  

In 2019, the Templeton World Charity Foundation, a unit of the Templeton Foundation, which supports scholarly research at the intersection of science and religion, announced a $20 million grant program called Accelerating Research on Consciousness (ARC) intended to identity the neural substrates of consciousness (Templeton has long been interested in the problem of free will, which boils down to the question of whether deliberate conscious mental activity plays a causal role in behavior).  The plan, as described in Science ("Rival Theories Face Off Over Brain's Source of Consciousness" by Sara Reardon, 10/18/2019) is to take about a half-dozen prominent theories and pit themselves against each other in what is known as adversarial collaboration, a process proposed by Daniel Kahneman (American Psychologist, 2003) in which the proponents of competing theories work together to test their theories against each other.  That is, the two parties agree on experiments that would decisively test between their respective theories.  The first adversarial collaboration, in which $5 million will be spread out across five international laboratories, will pit a version of Baars's Global Workspace Theory (GWT) against Tononi's Integrated Information Theory (IIT).  Stanislaus Dehaene (in Consciousness and the Brain, 2014) has proposed a version of GWT which emphasizes the role of the prefrontal cortex -- which is reasonable enough, given that the "Global Workspace" is tantamount to working memory.  By the same token, Tononi proposes that the interconnectedness of the brain, allowing the information integration to occur, is greatest in the parietal lobes.  So the research will involve using several methodologies -- fMRI, EEG, and electrocorticography (ECoG, to be used with neurosurgical patients) -- to measure the activation of various brain regions while subjects engage in activities involving conscious processing.  Like what, you say?  We'll see.  In any event, the one whose brain regions activate (or, I suppose, activate the most) during task performance will win the contest.

Actually, a specific research proposal was released in 2021 by a research group of scientists including Koch (Melloni et al., "Making the Hard Problem of Consciousness Easier", Science, 05/28/2021).  They interpret GNWT as predicting that consciousness will occur with an increase in activation of frontal-parietal areas within about 300 msec of stimulus onset, and again within about 300 msec of stimulus offset.  IIT, by contrast, predicts that activation of the posterior hot zone (PHZ) will arise within 300 msec of stimulus onset, and persist until stimulus offset.  For example, brain activity might be recorded by EEG, MEG, or fMRI, or perhaps even by an array of electrodes resting directly on (or in) the cortex, while subliminal and supraliminal stimuli are presented to the subjects.  The differential predictions are illustrated in the image at left.  According to GNWT, activation should be confined to the PHZ, while IIT predicts that activation will also be found in frontal and fronto-temporal areas.  GNWT predicts a period of inactivity between stimulus onset and stimulus offset, while IIT predicts constant activation during this period.  You get the idea. 

In addition, these also researchers proposed to test "first-order" and "higher-order" theories of consciousness against each other.  First-order theories, such as the one proposed by Ned Block, predict that consciousness of stimuli (visual, auditory, etc.) will be associated with activation in the corresponding sensory areas.  Second-order theories, such as the one proposed by David Rosenthal, predict that consciousness will also entail the activation of a "higher-order" brain network that, essentially, "points to" (as it were) the neural representation of the stimulus.  We can see how these theories could also be tested with the imaging methods used to test GNWT against IIT.

Stay tuned for further developments.  But while some cognitive scientists engage in endless debates over consciousness, functionalism, and reductionism, and search (I think fruitlessly) for the consciousness module, or system, or something, in the brain, there are still a number of interesting scientific problems to be addressed.  

Four Problems of Body and Mind

We also know, intuitively, that consciousness has causal efficacy.  Our thoughts, feelings, and desires cause us to do what we do, and our mental states can affect bodily states.  So another project for science is to find out how the brain does this, too. How do mental states affect bodily states and functions, as in placebo effects and psychosomatic effects.  Both projects are materialist in nature, but both projects take mind, and consciousness, seriously -- they don't try to write them out of the picture.

The Mysterious Leap from the Body to the Mind: The Neural Correlates of Consciousness

The Puzzling Leap from the Mind to the Body: Psychosomatics

Mind Without Body?  Spiritualism and Parapsychology

Body Without Mind: Reflex, Taxis, Instinct -- and Zombies

This page last revised 10/09/2023.