A dialogue with Ordinary Person Keith

Eric Schwitzgebel and I have had a long-running debate about whether there is a minimal, theory-neutral conception of phenomenal consciousness, which everyone can agree picks out something real. Eric thinks there is, but I am sceptical. You can find some of the exchanges in print — see this comment by Eric, my reply in this paper, and my second thoughts in a long footnote in this forthcoming paper — but Eric and I have also continued the debate by email.

In a recent exchange, Eric suggested that we have a grip on phenomenality simply in virtue of having a grip on our conscious experiences. Phenomenality, Eric remarked, is ‘the obvious property that those conscious experiences share in common, in virtue of which you (not as philosopher Frankish but as ordinary person Keith) recognize them all as worth labeling with the term “conscious experiences”‘.

So I, Philosopher Frankish (PF) asked Ordinary Person Keith (OPK) about this. To avoid biasing him towards illusionism, I played devil’s advocate. Here’s how it went.

PF: Hi Ordinary Person Keith!

OPK: Hey, Philosopher Frankish! Nice to see you. It’s been ages. What have you been up to?

PF: Oh, you know, the usual. Making distinctions, questioning intuitions, denying consciousness. Nothing spectacular. You?

OPK: The usual too. Looking at stuff. Assuming stuff. Riding the Clapham omnibus.

PF: May I ask you something? It’s for my friend Eric.

OPK: Sure — fire away. I have views about all sorts of things.

PF. It’s about consciousness.

OPK: Uh-oh.

PF: Think about your conscious experiences.

OPK: How’d you mean?

PF: When you’re seeing something, smelling something, tasting something, feeling pain — states like that.

OPK: States?

PF: When you do those things — seeing, smelling, and so on — you enter a certain kind of state, the state we call a ‘having a conscious experience’.

OPK: I don’t call it that, but fair enough.

PF: Now, you can recognize when you’re in a state of that kind, can’t you?

OPK: You mean I can tell when I seeing something, smelling something, feeling pain, and so on?

PF: Yes. Now, how do you tell it?

OPK: Come again?

PF: What do all those experiences have in common that enables you to recognize them as conscious experiences?

OPK: I’ve no idea. Do they have anything in common? Seeing a blue sky doesn’t have much in common with smelling a rose or feeling a pain in my toe.

PF: But don’t they all have a certain feel to them?

OPK: A feel? The pain has a feel. It feels bad. But seeing the blue sky and smelling a rose don’t. There’s just the colour and the smell. I suppose they might make me feel nostalgic or something like that . . .

PF: But think about what it’s like to see blue. Forget about the emotions it causes. There’s a quality to the blue experience that you can immediately recognize.

OPK: You mean the blue that I see?

PF: The blue quality of the experience.

OPK: You mean the blue quality of the sky. You said that my experience is the state of me seeing the sky. I don’t turn blue when I see the sky!

PF: No, but focus on what seeing blue is like for you, subjectively, on the inside.

OPK: On the inside of what?

PF: In your mind.

OPK: It’s like seeing blue. What else can I tell you?

PF: Right, the mental state of seeing blue has a certain quality to it that you can attend to. So does smelling a rose, feeling a pain. Every conscious experience has its own distinctive quality.

OPK: Are you saying that the act of seeing something blue is itself blue?

PF: In a way. Not in the same way that the sky is blue, obviously.

OPK: Sheesh! So when I see the sky I’m aware of two kinds of blue — the blue of the sky and the blue of my experience of the sky?

PF: Yes — that’s it!

OPK: Seriously?

PF: Seriously. How else do you recognize that you’re having the experience?

OPK: Let me get this straight. When I see blue . . .

PF: When you consciously see blue . . .

OPK: If you insist. When I consciously see blue, I detect a sort of blue in my mind and it’s this mental blue that alerts me to the fact that I’m seeing blue in the world?

PF: Sort of, yes. It’s that mental blue that makes the experience conscious.

OPK: Is it? And the same’s supposed to go for sounds, smells, tastes, and all other experiences?

PF: Yes, every conscious experience has a distinctive mental quality to it that makes it conscious.

OPK: What if it didn’t? Couldn’t I just see the blue sky without being aware of a mental blue as well?

PF: Maybe, but then you’d be a zombie!

OPK: Blimey. Well, maybe that’s what it’s like for you. But I just see the blue of sky. Maybe I am a zombie.

It seemed that Ordinary Person Keith didn’t have a good grip on phenomenality after all. Of course, I could have gone on to talk about the physics of colour, after-images, inverted spectra, and so on, but Eric and I had agreed that such heavy theorizing wasn’t appropriate if we were trying trying to establish a minimal notion of phenomenality.

I reported back to Eric. He replied that I had confused Ordinary Person Keith by trying to shove too much theory down his throat. Instead, he said, I should ask him ‘whether seeing the sky, feeling pain, pondering how to get to grandma’s house during rush hour, and having a sudden rush of fear have anything in common that immune system response and heart rate regulation lack. When he says yes, they have something in common, ask him whether hearing a tune in his head belongs on the former list or the latter.

So I tried again.

PF: Hi again, Ordinary Person Keith. Look, I’ve had a word with Eric, and he says I’m confusing you. He suggests  that I ask you whether seeing the sky, feeling pain, pondering how to get to Grandma’s house during rush hour, and having a sudden rush of fear have anything in common that immune system response and heart rate regulation lack.

OPK: So that’s seeing, feeling pain, pondering something, and having a rush of fear?

PF: Yes, do they have something in common that immune system response and heart rate regulation lack?

OPK: I’m not sure. Seeing, feeling, thinking, a rush of fear — they all seem very different.

PF: Sure. But aren’t they all even more different from immune response and heart regulation?

OPK: Well, I’m not aware of the immune response and the heart regulation, but I am aware of what I’m seeing, feeling, and thinking. I can tell you about it.

PF: Yes — seeing, feeling pain, and so on are conscious states. Immune response heart regulation isn’t. Right?

OPK: OK.

PF: And what about hearing a tune in your head? Does that belong on the former list or the latter?

OPK: I’m aware of it, so on the former. But what follows?

PF: That all those conscious states have some property you’re aware of — a quality, which makes them conscious.

OPK: Do they?

PF: Yes, when you’re seeing, feeling, thinking and so on, you’re aware of what’s it’s like to be in those states!

OPK: I don’t get it. When I see the sky I’m aware of what the sky is like — blue or cloudy, or whatever. When I feel pain. I’m aware of what the pain is like — where it is, whether it’s dull or sharp, and so on. When I think about getting to Grandma’s house I’m focusing on the house — which direction it’s in, where it’s located on the street plan. Same with the rush of fear and the song in my head. The sky, the pain, the house, the fear, the song — they don’t have anything in common, except that I’m aware of them.

PF: Oh dear. I’m confusing you again, aren’t I? I’d better ask Eric for advice…

What do we have access to?

Indirect realism and idealism are not popular with contemporary academic philosophers (in the 2022 PhilPapers survey, only 6.6% endorsed idealism and only 5.0% sense data theory), and to many — including me — they are profoundly unattractive.

Yet they retain some popularity among a wider public. When discussing philosophy of mind on social media, I frequently find people asserting as an unquestionable truth that the only things we are directly aware of are our own experiences and that all our beliefs about the external world are inferences from beliefs about our experiences. (My response is usually to ask (a) Who or what does ‘we’ refer to in this context? and (b) How do we gain direct access to our experiences? The only answers that work in the context are ‘a soul’ and ‘magic’, or some more nuanced variants of them.)

Why is this? I don’t know (the philosophical arguments are certainly not compelling), and I’d be interested to know what others think. But I suspect that one factor may be the way neuroscientists write about consciousness. They tend to stress that the brain has no direct access to the world it represents. Here’s an example from David Eagelman:

Here’s the key: the brain has no access to the world outside. Sealed within the dark, silent chamber of your skull, your brain has never directly experienced the external world, and it never will … Everything you experience — every sight, sound, smell — rather than being a direct experience, is an electrochemical rendition in a dark theater.

David Eagleman, The Brain: Story of You (Pantheon, 2017), p. 41

Now, I’m not suggesting that Eagleman or others who write like this endorse indirect realism. They are, I take it, making two perfectly good points: first, which aspects of the world we experience and how we experience them are determined by processes in our brains, which (let us suppose) create internal models of the environment, and, second, the brain has to construct its models from completely uninterpreted data — essentially, spiking patterns in neurons.

But such talk can easily be misinterpreted. A reader might reason as follows: ‘Scientists tell me that my brain doesn’t have access to the world outside. But I have access to something — this rich buzzing, booming world of sensory qualities. So this something must be an internal world, the model constructed by my brain. That’s what I’m directly aware of, and all my beliefs about the outer world are inferences from beliefs about this inner world.’

This is, of course, fallacious. My brain may not have access to the world beyond the skull, but it doesn’t follow that I, the whole organism, do not. The brain’s job (or one of its jobs) is to put the organism to which it belongs into a relation of tight sensitivity to the world around it — the relation we call ‘awareness’ or ‘experience’ — and the models the brain constructs are part of the subpersonal machinery that creates this personal relation.

Of course, our personal awareness of the world is not immediate and perfect; far from it. It is dependent on the hugely complex processes neuroscientists describe, and it is simplified, distorted, and caricatured in ways that reflect our needs as evolved biological organisms. But it is an awareness of the world, not of some inner simulacrum of it.

Indeed, if we were personally aware of an internal model of the world, then, from an explanatory point of view, we’d be back to square one. Neuroscientists would point out that the brain has no direct access to the models created in other brain regions and that it must construct our awareness of them from uninterpreted neural signals — making models of models. Drawing the same fallacious inference, indirect realism would become doubly indirect, then triply, and so on.

Animal sentience and animal welfare

A postscript to yesterday’s post on animal sentience. Some readers took me to be proposing that we drop the concept of sentience and stop asking which animals are sentient and which aren’t. Since it’s generally agreed that sentient creatures have ethical claims on us that non-sentient ones don’t, such a policy might have worrying ethical implications.

That wasn’t my intention. I no more want to eliminate the notion of sentience than you, in my imagined conference scenario, would want to eliminate the notion of life. You would want the conference participants to revise their conception of life — to start thinking of it as a cluster of biological processes rather than as a hidden essence that is only contingently connected to those processes. Substituting ‘psychological’ for ‘biological’, that’s what I want to do with consciousness.

Revising our conceptions of life and sentience in these ways would not prevent us from continuing to ask about the distribution of those properties in the natural world, and it would, in fact, make the task much more tractable. Nor would it prevent us from continuing to regard life and sentience as ethically significant. (Indeed, the revised conceptions would provide a much better foundation for ethical concern than the old ones, which treated those features as mysterious essences, which might have no casual role in the physical world.)

This isn’t to say that the revisions would have no consequences. For one thing, they would change the way we frame questions about the distribution of life and sentience. Instead of asking ‘Is this creature alive/conscious?’, we would ask ‘Which aspects of the cluster of biological/psychological functions constitutive of life/consciousness does this creature possess, and to what degree?’.

Focusing on sentience, we would cease to think of consciousness as a binary feature and and cease to ask whether or not a creature possesses it tout court. Instead, we would think of sentience as a multi-dimensional space of possibilities, whose axes correspond to different psychological sensitivities and abilities, and ask whereabouts in this space a creature is located. In short, we would replace a neat but intractable metaphysical question with a messy but tractable empirical one.

We would also change how we approach the ethical issues. If sentience were binary, then our task would be to divide animals into the sentient sheep, who have an ethical claim on us, and the non-sentient goats, who don’t. But if it’s a multi-dimensionally graded feature, then we would need to adopt a much more nuanced approach. We would need to to determine where each creature was located in the region of sentience space and ask what kind of ethical claims creatures in that region have on us, given their characteristic sensitivities and abilities. Instead of asking, ‘Should we care about this creature?’ we would ask, ‘How should we care about this creature?’

I think that would be progress.

How I feel about work on animal sentience

Yesterday on Twitter Robert Long asked me how I, as an illusionist, feel about the recent surge in work on animal sentience, such as that being done under the umbrella of Jonathan Birch’s Foundations of Animal Sentience project.

The short answer is: enthusiastic and optimistic. It’s great to see this work. It should shake up our anthropocentric assumptions, both theoretical and ethical, and give us a much better understanding the diversity and complexity of the minds of the creatures with whom we share this planet. The bulk of the experimental work being done will be useful regardless of whether one takes a realist or illusionist view of the metaphysics of consciousness.

Having said that, I do often feel uneasy at the way the debate about animal sentience is framed. I’ll illustrate my unease with a little story.

Imagine you go to a conference on animal life. Everyone there is debating furiously about which animals are really alive. Most are convinced that mammals are alive, but there is deep disagreement about whether birds, reptiles, cephalopods, and insects are, and only a few brave souls are prepared to argue that flatworms are alive.

You are puzzled at first, but then you realize that what they mean by ‘life’ is different from what you mean. They do not think of life as a loosely defined cluster of biological functions, such as growth, perception, metabolism, and reproduction. They think of life as an extra feature — an essence or spirit — which can’t be defined in functional terms and can’t be directly detected.

Most of them agree that this feature is closely associated with the biological functions you think define life, and some even think it is identical with some cluster of them, but they they can’t decide which biological processes are the best indicators of its presence. They cite vast quantities of experimental work on life in animals, but it all concerns the presence of some biological function or other, and since there is no agreement about which function is the best indicator of life, none of it is decisive. What one theorist regards as definitive evidence of the presence of life another dismisses as a confounder.

You tentatively suggest that life is nothing more than a cluster of biological processes and that the extra feature the participants are looking for is illusory, but everyone stares at you with incomprehension. Some declare that that you’re a monster for denying that animals are alive.

Now you know how I feel about work on animal sentience.

A dilemma for illusionists — and another for realists!

David Chalmers often — rightly — presses me to clarify exactly what it is that I and other illusionists deny. We say that we deny the existence of phenomenal properties, or qualia, but what exactly do we mean by this? Here, it seems, we illusionists face a dilemma. Is the notion of a phenomenal property a theoretical one or an observational one?

Suppose we say that the notion is a theoretical one — say, that of a property of experience that is nonphysical, not publicly observable, and immediately known. Then phenomenal realists such as Chalmers will say that that’s not what they mean by a phenomenal property. They mean simply the feel of experience — what it’s like. They do not define these feels as having features such as non-physicality, though they have arguments for thinking that they do in fact have them. Thus, if illusionists take this option, they are not denying the existence of properties realists believe in, but merely denying that those properties have the features many realists believe they have. Chalmers will be more than happy to engage with them on the latter point.

Suppose, then, that illusionists say that the concept of phenomenality is an observational one — the concept of the kind of feature people detect when they introspectively detect that they are having an experience of some kind. Then illusionists are denying something realists believe in, but also, it seems, making the absurd claim that experiences have no introspectively detectable aspect — that you detect nothing when you detect that you’re in pain!

This is the dilemma facing the illusionist. Francois Kammerer has likened it to the task Odysseus faced in navigating his ship between the two sea monsters Scylla and Charybdis. (See his talk ‘Defining consciousness and denying its existence’ at a workshop in Bochum.) What should the illusionist do?

The short answer is that they should reject the dichotomy and point out that all observation, including introspection, is theory-laden and that the concepts we employ in introspection are themselves infused with theoretical commitments. This is the line Kammerer takes, and I think it’s right. I won’t develop the response here, however. (I discuss the issue at much more length in a new paper forthcoming in a special issue of the French journal Klēsis, edited by François Loth.) Instead, I want to point out that illusionists can pose a parallel dilemma for the realist.

Here’s how they can do it. First, they need to sketch some positive account of what consciousness is, couched entirely in terms of informational and reactive processes centred on the brain. For present purposes, the details do not matter; the crucial thing is that the account is framed entirely in functional terms. The realist will say that the account omits, or at least fails to explain, something — the central phenomenal aspect of experience. (Note that it does not matter whether the proposed account is actually true; the point is that the realist will claim to know a priori that it is not true, or at least not fully explanatory, since it misses out something crucial.) Now comes the dilemma. How do realists conceive of the missing properties?

Suppose they conceive of them in the observational way, as the properties people detect when they introspectively detect that they are having experiences, with no commitment whatsoever as to the nature of these properties. This option was unattractive to illusionists, but it is awkward for realists too. How can they know that the detected properties are omitted from, or not explained by, the illusionist’s account? Why should they even suspect that they are not included? After all, they claim to have no idea what the properties are.

Suppose, then, that realists allow that the notion of phenomenality does have a theoretical component after all, and that they conceive of phenomenal properties as having certain specific features. Then they can maintain that the illusionist account is inadequate, on the grounds that it omits or fails to explain properties with those features. But if they make this move, then illusionists can deny the existence of phenomenal properties without denying that experiences have an introspectively detectable aspect. For they can say that experiences have an introspectively detectable aspect but this aspect is not phenomenal. What you detect when you detect that you are in pain is not really a phenomenal property.

Sauce for the goose is sauce for the gander!

Where does this leave us? It leaves us having identified the core questions as, first, what theoretical commitments are packed into the notion of phenomenality, and second, whether experiences have phenomenal properties as characterized by those commitments.

In response to the first question, I’m inclined to say that the key commitments are to phenomenal properties being (a) not characterizable in functional terms and (b) clearly revealed to introspection. Other claims, such as that phenomenal properties are non-physical, resist scientific explanation, and are radically private, derive from those core ones. I discuss this more in the Klēsis article.

As for the second question, I answer in the negative, of course.

Updated 3/12/22

Some fallacies (or heuristics?)

Some common fallacies in debate. (Or are they? Might some of these moves have heuristic value?)

The Bad Company Fallacy: People who believe p typically also believe q, which is false. Therefore not-p.

The Bad Reasons Fallacy: People often believe p for bad reasons. Therefore not-p.

The Desire Fallacy: The people arguing for p want it to be true that p. Therefore not-p.

The Fairness Fallacy: It would not be fair if p were true. Therefore not-p.

The Fallacy Fallacy: It is a fallacy to think that p follows from q. Therefore not-p.

The Incredulity Fallacy: I cannot believe that you are suggesting that p. Therefore not-p.

The Tribal Fallacy: My political opponents believe that p. Therefore not-p.

Technology and the human minds

As some of you may know, I have my own particular take on the dual-process theory of reasoning, and I recently I wrote a longish paper applying this take to issues surrounding AI and cognitive enhancement. The abstract of the article runs as follows:

According to dual-process theory, human cognition is supported by two distinct types of processing, one fast, automatic, and unconscious, the other slower, controlled, and conscious. These processes are sometimes said to constitute two minds—an intuitive old mind, which is evolutionarily ancient and composed of specialized subsystems, and a reflective new mind, which is distinctively human and the source of general intelligence. This theory has far-reaching consequences, and it means that research on enhancing and replicating human intelligence will need to take different paths, depending on whether it is the old mind or the new mind that is the target. This chapter examines these issues in depth. It argues first for a reinterpretation of dual-process theory, which pictures the new mind as a virtual system, formed by culturally transmitted habits of autostimulation. It then explores the implications of this reinterpreted dual-process theory for the projects of cognitive enhancement and artificial intelligence, including the creation of artificial general intelligence. The chapter concludes with a brief assessment of the risks of those projects as they appear in this new light.

The paper appeared in a 2021 collection titled The Mind-Technology Problem, edited Robert W. Clowes, Klaus Gärtner, and Inês Hipólito. If you don’t have access the volume, here is an eprint of the article.

In addition, Anna Strasser has prepared a fantastic PowerPoint presentation summarizing the main ideas of the article, which she has kindly given me permission to share.

The brain repair experiment

Looking through old teaching materials, I found a little philosophical story I’d written as an exercise for students of my first-year introduction to philosophy of mind course at Sheffield in the mid-1990s. I thought it was moderately amusing and that others might like to see it, so here it is. I have tidied it up and removed the exercises that accompanied it, but I haven’t attempted to update it. All the characters are fictional. The Brain Repair Experiment

A dialogue between Philobios and Philopraxis

Scene: The agora of ancient Athens.

Enter Philobios and Philopraxis, two philosophers.

Philobios: Greetings, Philopraxis! Have you a moment? I have been thinking about the nature of life, and I should like to share my ideas with you.

Philopraxis: Ah well, actually I’m busy. I’m searching for my dog, you see. He’s run off again.

Philobios: It won’t take long. Then I can help you search for your dog.

Philopraxis: Very well. If it won’t take long.

Philobios: Excellent. To begin, then: Some things are alive and some things are not, and we are good at telling which are which. You and I are alive, obviously, and so are many other things, from bears to beetles to beetroots. It seems there is some property they all share, a vital property, a that-it’s-aliveness, which we can immediately recognize when we encounter it. But what is this property exactly, and how does inanimate matter come to possess it?

Philopraxis: Isn’t that a question for our scientists? Like young Aristoteles, for example.

Philobios: Ah, no. Scientists describe processes that occur in living things — responsiveness to stimuli, metabolism, growth, reproduction, and so on — but this doesn’t get to the heart of the matter. We can easily imagine all those processes occurring without life. A machine might detect things, convert resources into energy, use raw materials to augment itself, and even make copies of itself, without really being alive. Processes such as responsiveness and metabolization may be necessary for life, but they aren’t sufficient for it, and explaining how those processes occur in living things won’t be sufficient to explain life itself. We’d still have to explain how things that support those life-related processes acquire a vital property, an intrinsic that-it’s-aliveness. This is a really hard problem! We need a new science of life, which takes life seriously as a fundamental aspect of reality. Perhaps life emerges spontaneously, by natural necessity, when sufficiently complex life-related processes occur. Or perhaps all matter possesses a tiny spark of life, and the vital properties of complex organisms like us are produced by combining the micro-level vital properties of their atomic constituents. We should explore these exciting ideas! Don’t you agree?

Philopraxis: Erm, no, actually, I don’t. I think processes of the sort you mentioned are all there is to life. Being a living thing is just a matter of being a persisting system that is responsive, metabolizes resources, grows, and so on. There’s no need for an extra animating ingredient. Maybe it seems otherwise to you because we can recognize life so easily. When we encounter a living thing, we don’t have to check that all those processes are occurring within it. We can just see that it’s alive. And that tempts us to think that we are detecting a special property that is distinct from all those processes. I can understand. I feel the same, in fact. But I don’t trust my feelings here. I think it’s a sort of illusion. For her own good reasons, Nature has tuned up our perceptual systems to be highly sensitive to indications of life, so that animate things leap out at us perceptually. And this leads us to think that they possess a special property — a vital property, a that-it’s-aliveness — which can’t be explained in more basic terms. That’s how it seems; animation is a fundamental feature of the world as we take it to be — of what we might call our manifest image. But it’s not a fundamental feature of reality independent of us, and we would be making a big error if we were to set about asking how it arises, whether all matter possesses it, and so forth. We’d be mistaking a feature of our reaction to the world for a feature of the world itself.

Philobios: But that’s not taking life seriously. In fact, it’s denying life! That’s the silliest view I’ve ever heard. What’s more, it’s an immoral and dangerous view. If you deny that life exists, then presumably you don’t think that murder is possible? I assume you think that the tyrant Hippias was a fine man. For on your view, he never deprived anyone of life!

Philopraxis: You’re missing my point. I’m not denying that life exists; I’m denying that life is what you think it is…

Philobios: You’re equivocating! Does life exist? Yes or no!

Philopraxis: Yes, but…

Philobios: So you concede that vital properties are real…

We pan away to a dog scratching itself. The dog settles down to sleep in the sun as Philobios and Philopraxis continue to argue in the background.

KF: Philobios and Philopraxis continue to argue, but I think you will have got the gist of their views about life, and I hope you will agree that Philopraxis’s view is better and that it is unfair of Philobios to accuse him of silliness and immorality.

Now substitute ‘consciousness’ and ‘phenomenal properties’ for ‘life’ and ‘vital property’ and replace the references to biological processes with ones to perceptual and cognitive processes. Philobios’s position is now a robust form of phenomenal realism and Philopraxis’s is a form of illusionism.

Do you still agree with Philopraxis? If not, why not?

George Harrison - What Is Life

Some replication crisis resources

Note: A few years ago, I made a start on a blog post listing some resources on the replication crisis in psychology. I have just found the draft and thought I would post it, in case it is of use to anyone. It’s very incomplete and outdated. If you would like to improve the post by adding more references in the comments, please do!

Blogposts and journalism

Articles

Podcasts

Twitter

Some thoughts on LLMs

Here are some thoughts on large language models (LLMs), derived from a recent email exchange concerning the digital simulation of Daniel Dennett recently created by Eric Schwitzgebel, Anna Strasser, Matthew Crosby.

Do LLMs such as GPT-3 perform speech acts? Is there a continuum between what they do and what we do when we speak? Or is there a sharp difference — we performing genuine speech acts and GPT-3 merely giving the appearance of doing so?

There’s a case for thinking that GPT-3 can perform speech acts. It can make appropriate contributions to conversations within domains it has been trained on. If game-playing AIs such Alpha-Zero can make genuine moves within a board game, why can’t GPT-3 make genuine moves with in a language game?

I think it can. It can make conversational moves with words. The problem is that that’s all it can do. It can’t do any of the other things we do with words. It can’t inform, instruct, persuade, encourage, suggest, imply, deceive, and so on. It can’t perform any illocutionary acts. It can’t even assess its own conversational contributions and select the best.

It can’t do any of this, of course, because it lacks the raft of psychological and social attitudes and aptitudes that illocutionary acts require. (As Dennett has noted, Grice’s theory of communication is an attempt to elucidate this implicit background; see chapter 13 of his From Bacteria to Bach and Back.) Even a non-human animal with a simple call system can perform more types of speech act than GPT-3 can.

And here there is a big contrast with the game-playing AIs. Alpha Zero can’t do much with go counters; it’s limited to making clever moves in games of go. But we don’t do much more with go counters ourselves. Go is a game, and making clever moves is the object. Of course, we can achieve other things through playing go, such as winning a bet, but we also play the game simply to pass the time.

Language, by contrast, is far more than a game. It’s a tool — a hugely complex Swiss army knife of a tool — and one that GPT-3 doesn’t have a clue how to use. If there’s a sharp divide here, it’s between systems that play language like a game and ones that use it as a tool.

It is true that humans sometimes treat language as a game. Think of a simultaneous translator whose only object is to reproduce the moves made in one language with parallel moves in another. But these are rare cases. (This example was suggested by a comment by Douglas Hofstadter.)

A final, broader point. One of the worries about AIs such as GPT-3 is that we may end up creating more complex versions of them, which display a wide range of human-like behaviours but lack a rich human psychology — counterfeit humans, which would trick us into treating them as equals. I suspect this is a red herring. We have little reason to create machines that behave like us, with all our limitations, weaknesses, and occasional concomitant sublimities, especially as we can already produce new humans by biological means.

What we will want are machines that aren’t like us — ones that can do things we can’t do or that make a better job of things we can. We will want artificial astronauts, explorers, builders, doctors, companions, inspirers, inventors — and some will want artificial cheaters, exploiters, and fighters. The danger is that we will give these beings a veneer of humanity — an ability to play social interaction like a game — which will lead us to both overestimate and underestimate their real abilities.

Revised 7/12/22

Caricaturing attempts to crack consciousness

On 15 May this year, The Spectator magazine published an by Kit Wilson titled ‘Cracking consciousness: how do our minds really work?’. I disagreed with much of the article, which I felt gave a seriously unbalanced view of consciousness science, but I was particularly annoyed by the author’s presentation of the illusionist view to which I myself subscribe. Partisanship is one thing, but downright misrepresentation is another, and I sent a brief letter to the magazine protesting about it. Since the editor decided not to publish it, I am posting it here.

Sir: As the person who coined the term ‘illusionism’, I want to protest against the mischaracterization of the illusionist view in Kit Wilson’s article ‘Cracking consciousness: how do our minds really work?’ (15 May 2022). Wilson says that illusionists ‘deny the mind exists’, and he uncritically quotes dismissive assessments (‘essentially gibberish’, ‘the silliest claim ever made’) from two of the view’s most hostile critics. Your readers deserve better.

Illusionists do not deny that mind and consciousness exist. They merely reject a certain view of what consciousness is. Specifically, they deny that consciousness involves awareness of what philosophers call ‘qualia’ — private mental versions of colours, sounds, tastes, and so on. Illusionists argue that qualia are illusory, and they offer alternative accounts of consciousness that do not mention them. Wilson suggests that this view is ‘self-evidently self-defeating’ since the illusion of consciousness would itself be a conscious one, but this facile objection ignores the fact that illusionists conceive of consciousness (and thus of illusion) as not involving qualia.

Illusionism may not be the right view, but it is an alternative to the outlandish and untestable speculation Wilson discusses elsewhere in his piece, and it deserves to be assessed on its merits rather than dismissed on the basis of a crude caricature.

Keith Frankish
(Honorary Reader, University of Sheffield)
Heraklion, Greece

Like a rainbow

Do illusionists deny the reality of consciousness? I’ve been discussing this on Twitter recently (see this thread, among others), and it has promoted me to try to think of analogies that might illuminate the illusionist perspective.

Here’s one: rainbows. Rainbows are real, aren’t they? You can see them with your own eyes — though you have to be in the right position, with the sun behind you. You can point them out to other people — provided they take up a similar position to you. Heck, you can even photograph them.

But what exactly is it that’s real? It seems as if there’s an actual gauzy, multi-coloured arc stretching across the sky and curving down to meet the ground at a point to which you could walk. Our ancestors may have thought rainbows were like that. We know better, of course. There’s no real coloured arc up there. Nor are there any specific physical features arranged arcwise — the rainbow’s “atmospheric correlates”, as it were. There are just water droplets evenly distributed throughout the air and reflecting sunlight in such a way that from your vantage point there appears to be a multi-coloured arc.

To sum up:

  • Rainbows, whatever they are: real
  • Coloured, spatially located aerial arcs: illusory
  • Experiences as of coloured, spatially located aerial arcs: real
  • Atmospheric conditions that cause experiences as of multi-coloured, spatially located aerial arcs: real

That’s very much how illusionists think of consciousness. It’s real enough. It’s the condition you’re in when you attend to things perceptually. I have it; you have it, scientists can study it. But, like a rainbow, it’s not what we naively take it to be.

When I reflect on my own experience, it seems to me that my consciousness is an inner world, where the world around me is rendered in private mental qualities — “qualia” — for my benefit alone. But there isn’t such a world. Neuroscience finds nothing like it in the brain, nor even anything isomorphic to it. Rather, it finds complex trains of neural activity proceeding in parallel and triggering a host of reactions — physiological, psychological, and behavioural. My sense of having a rich qualia-filled inner world is an impression created by all these processes, but the processes themselves are as different from the supposed inner world as a moisture-infused mass of air is from a colourful aerial arc.

To sum up:

  • Consciousness, whatever it is: real
  • A private qualia-filled mental world: illusory
  • The impression of a private qualia-filled mental world: real
  • Brain processes that produce the impression of a private qualia-filled mental world: real

I know what you are going to say! You’re going to ask about that impression of a private qualia world. What’s that exactly? Isn’t it a conscious experience — like the experience of seeing a rainbow — which itself belongs to a private qualia-filled mental world? If so, the whole idea is circular. I seem to be saying that the qualia world exists only in another qualia world. Does that second qualia world exist only in a third one, then, and so on? Ridiculous!

If illusionists thought like that, then their view would indeed be ridiculous. But they don’t. They don’t think that experiences exist in qualia worlds at all. They offer alternative accounts of what experiences are that don’t mention qualia. On the view I favour, experiences are complex sets of perceptually triggered psychological reactions and reactive dispositions. To have an experience as of a colourful arc in the sky is to form beliefs, memories, emotions, and a host of other reactive dispositions appropriate to the presence of such an arc. It is to be, as it were, in “sensing sky arc mode”. Similarly, to be under the impression that one has an inner qualia world is to react psychologically as if one had an inner qualia world — to think, talk, and react in countless other way as if such a world existed. That suggestion needs a lot of fleshing out, of course, and you might think it won’t work, but at least it shows that illusionists aren’t making a ridiculously circular claim.

That’s the analogy then. Consciousness is as real as a rainbow. It exists, but it is not a private qualia world, any more than a rainbow is a physical arc in the sky. So trying to find the neural correlates of the qualia world is as sensible as trying to find an arc-shaped structure in the atmosphere after a rain shower. And searching for a solution to the Hard Problem is like looking for the pot of gold at the end of a rainbow!

PS. After posting this piece, I was delighted to find that Daniel Dennett gave it his stamp of approval:

The Rolling Stones - She's A Rainbow (Official Lyric Video)

Do you know what sensations you’re having?

Last week, Matt Lieberman posted the video below on Twitter. (I believe it was originally posted on TikTok by the dancer featured in it.) The video illustrates change blindness. Large visual changes occur during the course of the video, which the viewer typically doesn’t notice. You may like to watch the video a few times and see if you notice the changes.

If you still haven’t noticed the changes, I’ll tell you. [Spoiler follows]

. . .

The colours of the dancers’ tops change markedly during the video. You can see the difference easily by jumping to different points in the video and noting the colours of the tops at each point.

You may be surprised that you missed such a large change, but it is not really surprising. Our visual systems are designed to keep track of things through changes in their appearance. Imagine you’re in a forest with a tiger stalking you. As the tiger moves in and out of the shade, the light reflected from its coat changes continually, but you need to ignore these changes and focus on what’s constant: the individual animal moving through the trees. And that is exactly what your brain does. In the case of the video, your brain accurately tracked the individual dancers, not confusing one with another, and it ignored changes in colour that were irrelevant to this task. Looked at in this way, change blindness is not a bug in our visual system but a feature, and we should expect it to occur all the time.

But this feature of vision presents a problem for a certain view of consciousness, which I’ll call the three-stage view. The view goes like this. Conscious perception proceeds in three stages. First, physical stimuli impact on our sense organs and produce neural activity in sensory processing regions of our brains. Second, this neural activity produces a visual sensation — a private mental quality which makes it like something to perceive the stimuli. Third, the sensation in turn produces reactions appropriate to its nature (unpleasant sensations cause us to shun the things generating them, pleasant sensations cause us to seek them out, etc). We react as we do because of what our experiences are like.

Now, when the colours changed without your noticing it, what happened at each of the three stages? There were certainly changes at the first stage. The light rays hitting your retina changed, and, presumably, at least some of the activity in your visual cortex did, too. There were no changes at the third stage, however — or at least no major ones. Maybe there were subtle changes in your behavioural dispositions that could be detected under experimental conditions. (I don’t know if this has been investigated; it would be an interesting project.) But your reactions did not change in a way that was noticeable either to you or to a casual observer.

What about the second stage? Did your visual sensations change? Was there a change in what the experience was like for you?

If there wasn’t, and your sensations didn’t change until you reacted, then the three-stage view is undermined. For the view says that our reactions are produced by our sensations; we react as if there is something green in front of us because we are having a sensation of green. But if we don’t have a green sensation until we react, then the sensation can’t be playing this role. Indeed, it is not clear why we need sensations at all, since the reactions occur independently of them.

So a defender of the three-stage view must say that our sensations change before we react. They can then continue to claim that when we do react we are reacting to a change in our sensations, albeit after a delay.

But this has a strange consequence. It means that we can be mistaken about our own current sensations. At the mid-point of the video, you thought you were having the same colour sensations as at the start, but you were in fact having different ones. And does that even make sense? Your sensations are supposed to define how things seem to you, and while you may be wrong about what colour something is, you can’t be wrong what colour it seems to be.

That’s not all. If you can fail to notice your sensations changing, then maybe you could fail to notice them fading out altogether. Maybe your sensations faded out an hour ago, and you haven’t noticed yet. You’ve continued to react as if they hadn’t faded out, of course, but during the video you continued to react as if your colour sensations hadn’t changed. Maybe you’ve never had colour sensations at all. Maybe you’re just reacting as if you have them. How do you know? All you can be sure of is what you think about your sensations. Maybe sensations are a sort of illusion.

A confession

I have a confession to make. I often fail to reply to people who email or message me with invitations, requests, and questions. I have an excuse, of sorts. I have many demands on my time and haven’t been in the best of health. But I feel guilty about it.

I am particularly bad at replying to messages that pose philosophical questions. Replying to such a message takes time and thought — especially if the question is a good one — and I often put it off, though fully intending to reply at some point. And then, of course, more messages come in, and my to-reply list grows longer and longer. The senders must think I am just ignoring them.

Why don’t I at least send a short acknowledgement, saying that I have received the message and will reply if and when I have time? I have thought of doing this, but the danger is that it would make me even less likely to write a full reply. Having sent the acknowledgement, I might feel that I’d dealt with the matter and could forget it. By contrast, If I don’t reply at all I feel guilty, and the guilt does sometimes push me to reply properly.

So here’s a message to anyone who has written to me and not received a reply:

First, I am sorry; please accept my apologies.

Second, please don’t think that I felt your message wasn’t worth a reply. The chances are it was just the opposite: I felt that it deserved a long and careful reply, which I just didn’t have time to write.

Third, please feel free to re-send your message. I won’t be annoyed and will reply eventually!

Zac’s disappearing nonconsciousness

Clive Wearing is a former musicologist and musician who has chronic amnesia, both anterograde and retrograde. He can’t form new memories or retrieve old ones. Although he is fully conscious, Clive lives trapped in a perpetual present, with no awareness of his own past (according to Wikipedia his episodic memory has a capacity of only 30 seconds). He continually feels that he has just emerged from a long period of unconsciousness, and in his diary he repeatedly affirms that he is now at last awake, scoring out earlier entries which affirmed the same and which now mean nothing to him:

8:31 AM: Now I am really, completely awake.
9:06 AM: Now I am perfectly, overwhelmingly awake.
9:34 AM: Now I am superlatively, actually awake.

Clive’s condition is a personal tragedy for him and his loved ones, but it is a also a fascinating psychological case study, and I want to take it as the basis for a philosophical thought experiment. I hope this does not suggest any disrespect to Clive or his family. I certainly intend none.

Clive suffers from what we might call disappearing consciousness. At every waking moment he is fully conscious, but those conscious moments rapidly pass into oblivion, and Clive’s conscious life has a brief temporal extension. We can imagine his memory span reducing further — to twenty seconds, ten, one, less — until his conscious life is the merest flicker in an otherwise dark existence.

I want to imagine someone who is the inverse of Clive — someone who suffers from disappearing nonconsciousness. Consider Zac. Zac has suffered an illness which has left him a zombie in the philosopher’s sense. His brain functions exactly as it did before, taking in the same range of sensory information and using it to produce responses just like those of a fully conscious person. He can give detailed reports on the world around him and his own bodily condition, noticing everything a fully conscious person would. However, his brain no longer produces any subjective experience — his mental states no longer have any phenomenal feel to them, no ‘what-it-is-likeness’. The inner light of phenomenal consciousness has gone out, and Zac experiences the world unconsciously. Given the standard way of thinking of consciousness, this should be at least conceivable.

There is a twist, however. Zac is not a complete zombie. Unlike other philosophical zombies, he is aware that he is not phenomenally conscious. So when he describes the visual scene before him, he adds that he isn’t really seeing it. He knows what the world is like, but his experience is not like anything. If zombies are conceivable, then a partial zombie like Zac should be, too. (I have borrowed the notion of a partial zombie from Allin Cottrell, who uses it to question the conceivability of zombies. See his excellent paper ‘Sniffing the Camembert’.)

Zac’s lack of phenomenal consciousness troubles him, and he repeatedly notes it in his diary. (He can write perfectly well, of course, and is aware of what he is writing, though the experience of writing is not like anything for him.)

8:31 AM: Now I am really, completely unconscious.

So Zac is unlike Clive in not being conscious. He is unlike Clive in another way, too. Whereas Clive’s brain fails to record memories of being conscious, Zac’s brain keeps fabricating them. When it forms episodic memories of scenes and events, it tags them as having been experienced in the normal way, just as they were before Zac’s illness. So although Zac experiences the world unconsciously, he remembers it has having been experienced consciously.

The upshot is that Zac suffers from the opposite delusion to Clive. Whereas Clive continually feels that he is waking up into consciousness and denies his past consciousness, so Zac continually feels that he losing consciousness and denies his past nonconsciousness:

8:31 AM: Now I am really, completely unconscious.
9:06 AM: Now I am totally, overwhelmingly unconscious.

9:34 AM: Now I am fully, actually unconscious.

And as the latency period between experience and memory shrinks, Zac’s sense of nonconsciousness shrinks too, until it is a mere flicker of darkness in an otherwise illuminated life.

What is the moral of this? The obvious one is that even if our experiences do have a subjective phenomenal feel to them, this feel may play little or no role in generating our sense of being conscious — of its being like something to be us. So maybe consciousness doesn’t have much to do with phenomenal feel after all.

Designed for destruction?

Suppose that our universe is intrinsically special in a way that it would not have been if the physical constants had taken other possible values. (As I argued in this previous post, it’s important that it be intrinsically special, not just special in the eyes of the creatures who inhabit it.) Then this would seem to raise the credence we should give to the hypothesis that the constants were not randomly set but shaped by a purposive agency of some kind, which sought to produce a universe with this special feature. For the sake of argument, let’s grant that it does.

So is our universe intrinsically special? What might be the special-making feature be? An obvious candidate is life. Life looks pretty special, perhaps even intrinsically valuable. Now, I’m not convinced that life is intrinsically valuable in the relevant sense. (We often treat it as intrinsically valuable, of course, and it’s right that we do, but that’s another matter.) But let’s concede it. And life certainly could not have existed if the physical constants had not been within a vary small range. So we have an argument that a life-seeking agency shaped our universe to enable to it to support life. Cool!

But wait a minute! Our universe’s physical laws and constants may allow life to emerge, but they also condemn life to complete annihilation. The second law of thermodynamics secures that. The relentless growth of entropy, which actually drives the local development of complexity, will eventually dismantle all complexity, leaving the universe a cold, dead realm of undifferentiated low-energy soup. This doesn’t look like something a life-seeking agency would build in. If we were designing a giant spaceship to carry humanity to a new star system, we wouldn’t build in a self-destruct mechanism set to trigger automatically halfway through the voyage.

In fact, a more plausible hypothesis is that the special making feature is the destruction of life, and that the agency that shaped the constants was seeking this goal — an inference not dissimilar to that drawn by Gloucester in King Lear:

As flies to wanton boys are we to th’ gods,
They kill us for their sport.

A defender of the life-agency hypothesis might reply that the agency could only do so much. It could tweak the physical constants but couldn’t change the second law. But what reason do we have to think that? We know nothing of this agency and the constraints under which it operated. (Unless of course, we think that revealed religion provides such information.) The agency has been introduced simply in order to explain why the universe supports life on the assumption that life is what makes this universe intrinsically special. If the universe is not really so special in the life department, then we can’t save the life-agency hypothesis by saying that the life agency did its best. It’s begging the question. We need to identity the special-making feature independently of the process posited to explain it.

Notes on emergence

Last April, Bryce Huebner and I had a Twitter exchange about emergence in biology. I felt that we were talking past each other and tried to work out how it had happened. I suspected that it was due to the way the concept of emergence had been used in different traditions, and I sketched the following tentative history. (I should stress that this is speculative; I’m not a historian.)

1. Late 19th – early 20th century: Some British philosophers (Broad, Lloyd Morgan, Alexander) started using ’emergentism’ to describe a metaphysical view on which the world is radically layered, with new fundamental properties and powers spontaneously appearing at each level, building up to the mental.

2. Mid 20th century: That kind of emergentism was empirically falsified, and philosophers rejected it. However, dualist philosophers continued to claim that some mental properties and powers are fundamental, emerge spontaneously, and exert a ‘top-down’ influence on the physical world, causing particles to behave in ways that physics alone could not predict.

3.Late 20th century: Scientists adopted the term ’emergence’, ignoring its metaphysical baggage and using it in a different sense to characterize global features of complex systems, especially self-sustaining ones. Confusingly, these scientists often spoke of these global features as exhibiting a ‘top-down’ causal influence — meaning that the components of complex systems are causally sensitive to global conditions, not that global features appear spontaneously and exert a fundamental causal influence on the physical world, additional to that of the components.

5. Late 20th – early 21st century: To capture the difference between the philosophical and scientific uses, philosophers drew a distinction between strong and weak emergence. However, they tended to define weak emergence in terms of deducibility and denial of ‘top-down’ influences, which made scientists think that that ‘weak’ emergentists denied the kind of global influences they needed to talk about.

6. Same time: Some dualist philosophers seized on this, claiming that scientists themselves were now talking of strong (spontaneous, fundamental) emergence. If strongly emergent features are common in the natural world, then why shouldn’t aspects of the mind be strongly emergent? Science vindicates dualism!

7. April 2021: Bryce and Keith talk past each other, Bryce adopting the scientific use, Keith the philosophical one.

At the time, Bryce and I talked of writing this up as a short paper, which we may still do. Meanwhile, if anyone has any comments, I’d be grateful for them. Comments from historians of science and philosophy would be particularly welcome!

Illusionism cover image

Front cover of Illusionism book

The cover image for my 2017 book Illusionism as a Theory of Consciousness is a painting of a rather flamboyant stage magician performing a variety of tricks simultaneously. I found the image on Wikimedia and thought it was an appropriate visual metaphor for the theory discussed in the book. But who was this magician? Wikimedia says only this about the image:

Zan Zig performing with rabbit and roses, including hat trick and levitation. Advertising poster for the magician (who seems to have left no other trace behind).

Could it be Julius Zancig? It seemed unlikely to me, given the description of Julius’s act, which he performed with his wife Anges. I put a message on my website asking if anyone could find out more.

In 2020, Jeff Miner (a former student of Kent Bach at SFSU, now working in tech) contacted me to say that he’d done some research on the image, which he has kindly given me permission to share. Jeff wrote:

I asked around and crowd-sourced a bit.

It looks as though he was a magician (possibly from the Cincinnati area) who named himself in such a way as to be confused with Zancig.

There’s a second lithograph of him at the LoC.

This response I got seems the most complete:

It was fairly common for second and third rank magicians to use names that resembled those of first-rank magicians (e.g. Hoodini, Howdini, Houdyni, etc.). The Zancigs, (Agnes and Julius) toured with a two-person telepathy (second sight) act. They didn’t do bunnies, doves and goldfish. It’s possible there was a fellow who called himself “Zan Zig” in hopes that people would think they were seeing the Zancigs.

Since the US Copyright Office is part of the Library of Congress, and since this lithograph was copyrighted, I expect the LOC got the name from the copyright registration. Librarians are pretty good about researching items in their collections, and the LOC librarians are some of the best in the world.

As to who “Zan Zig” might have been, at least on the Google-Indexed Web all hits on “Zan Zig” are to this picture (or the second lithograph also at the LOC showing a fellow who looks like the same guy doing four different illusions.) The fellow top left in that second lithograph is the standard stage version of Mephistopheles from Faust.

Doing a search in The New York Clipper and the New York Dramatic Mirror for the year 1899 might reveal more, or maybe not if this fellow was a local Cincinnati magician.

In a second email, Jeff forwarded further information from one of his correspondents:

There was a magician of this period named “Zanzic,” He was referred to in Leaves From Conjuror’s Scrapbooks, with this paragraph:

Another magician by the name of Robinson has been traveling in the Western States for the past few years, but is going under the professional name of Zanzic. Being a clever performer, it would seem he ought to have originated a more healthful-sounding name, which is “sick’led o’er with a pale cast of thought.”

There is quite a bit more about him in Jim Steinmeyer’s book The Glorious Deception, pages 135-138. The first paragraph there reads:

Zanzic was a tall, slender, dark-haired magician. His real name was Robertson or Brenner, and he also worked under the name Henry Andre. Zanzic had been born in New Orleans, the son of a Creole fortune-teller. He was six years younger than Will [Robinson], an old friend from the days when they were both starting out in magic and used to meet at Martinka’s shop. His associates thought of Zanzic as unstable: accident-prone and filled with half-baked schemes. He certainly had the skill to be a good magician. But for Zanzic, the adrenaline of a performance was like a strange, addictive sort of poison that made him giddy and stupid.

In short, the illustration is of a hyperactive magician who was hoping to deceive people about his very identity. I think that makes it even more appropriate for a book about the addictive self-illusions our brains create.

The pinprick solution to the Fermi paradox

Why are there no signs of aliens, given the vastness of the galaxy and the apparent abundance of opportunities for life to develop? This is the Fermi paradox. Here’s a possible solution to it. I’ve not seen it proposed before, though I don’t know the literature well.

Suppose the following claims are true:

1) There is a multiverse. (It may not be strictly necessary to assume this, but it makes exposition easier.)

2) Any universe that is capable of supporting life is also fragile, in the sense that a relatively small, highly localized ‘pinprick’ event can cause the whole thing to collapse, as a pinprick can destroy an inflated balloon.

3) ‘Pinprick’ events do not occur naturally. (Again, it may not be strictly necessary to assume this.)

4) An advanced technological civilization will almost certainly create a pinprick event inadvertently as soon as it is capable of doing so. (Suppose that there is a technological process that promises huge benefits in say, in cheap energy or interstellar travel, but which will also, and unforeseeably, create a pinprick event.)

Then we have a solution to the Fermi paradox. Every universe in which a technological civilization develops is rapidly destroyed by that civilization — or by another that develops around the same time. No universe contains more than a very small number of technological civilizations, and we shouldn’t expect to see signs of another one in our galaxy.

And we’re going to destroy the universe soon.