Daniel Dennett and the long audit of our explanations of mind
Daniel Dennett died on 19 April 2024. The obituaries called him the man who explained consciousness. His actual project was narrower and more useful: sorting the explanations of mind that do scientific work from the ones that merely feel like explanations. The quarrels that followed are still running.
The obituaries in April 2024 settled quickly on a description Daniel Dennett would have resisted: the philosopher who explained consciousness. He had spent thirty-three years insisting that the title of his 1991 book was a promise about method, not a boast about results. The label stuck because it was short.
A better description is closer to hand. From a 1971 journal article to an essay in The Atlantic eleven months before his death, Dennett's working life went on a single unglamorous task: auditing explanations. He took the accounts of mind people already accepted and asked of each whether it did any predictive work, or merely produced the feeling that something had been explained. Most of them, he thought, did the second thing, and the feeling was the problem.
He died at Maine Medical Center in Portland on 19 April 2024, aged 82, of complications of interstitial lung disease. Two weeks earlier he had held an online class with students on a paper he was still working on, the detail from Liad Mudrik's obituary in Nature most worth keeping.
A Beirut childhood, and an education in two kinds of deflation#
Dennett was born in Boston on 28 March 1942. His father, Daniel Clement Dennett Jr, held a Harvard doctorate in Islamic history and was working in Beirut as a covert agent for the Office of Strategic Services, so the boy spent part of the war in Lebanon; in 1947 his father was killed in a plane crash in Ethiopia and the family returned to Massachusetts. He began at Wesleyan, transferred to Harvard, took a BA in philosophy in 1963 under W. V. O. Quine, and finished his DPhil at Hertford College, Oxford in 1965 under Gilbert Ryle. Six years at UC Irvine followed, then Tufts from 1971, where he stayed as Austin B. Fletcher Professor of Philosophy and co-director of the Center for Cognitive Studies until taking emeritus status in 2022.
Quine and Ryle were both deflationists, and they deflated different things. From Quine he took a naturalism that refused philosophy any position above the sciences; from Ryle, the habit of looking at a mentalistic noun and asking what work it was doing. Dennett's originality lay in refusing the usual consequence of that combination, which is eliminativism. He wanted a third answer.
The stance that made him, and the one popular accounts skip#
Alasdair MacIntyre died in May 2025 insisting that modern moral argument is interminable because the words outlived the world that made sense of them. His diagnosis in After Virtue, the disquieting suggestion, emotivism, Nietzsche or Aristotle, the wait for another Benedict, reads differently in a year of shouting. Here is what holds and what does not.
Charles Taylor, Alasdair MacIntyre, Friedrich Nietzsche
That third answer is the intentional stance, the part of him most often left out in favour of the consciousness material. It arrived in "Intentional Systems" in The Journal of Philosophy in 1971 (vol. 68, no. 4, pp. 87-106), and was worked out at book length in The Intentional Stance (MIT Press, 1987).
The idea is a taxonomy of predictive strategies. To anticipate a system's behaviour you can adopt the physical stance and predict from its material constitution and the laws of physics; or the design stance, predicting from what the thing was built or evolved to do, which is how you know an alarm clock will ring at seven. Or you can adopt the intentional stance. Dennett's description of it (1987, p. 17) is worth quoting exactly:
First you decide to treat the object whose behavior is to be predicted as a rational agent; then you figure out what beliefs that agent ought to have, given its place in the world and its purpose.
You then work out what it ought to want, and predict that it will act accordingly. For a chess program this beats the physical stance by an enormous margin: you will never predict its next move from transistor states, and you will predict it rather well by asking what a competent player would do.
The radical move is what Dennett declined to say next. Asked whether the program really has beliefs, he refused to treat the question as well formed in the way his opponents wanted. The stance works or it does not, and where it works the patterns it tracks are objectively there; he argued this in "Real Patterns" (1991) and called himself a moderate realist rather than an instrumentalist. But he accepted the awkward consequence: where two systems of belief-ascription predict a creature's behaviour equally well, there is no further fact about which is correct. Jerry Fodor, who wanted beliefs to be internal states with determinate content, thought this gave away the store, and said so for two decades.
Their quarrel was never settled, and it matters because everything Dennett later said about machines follows from it. If "does it really believe?" has no fact behind it, "does the chatbot really understand?" will not trouble you as it troubles almost everyone else.
Consciousness Explained, and the objection it has never shaken off#
The 1991 book, published by Little, Brown, did three things at once. It attacked what Dennett called the Cartesian theatre: the assumption, surviving among materialists who believe they have abandoned dualism, that there is a place in the brain where things come together and a moment at which processing becomes experience. He named the residue Cartesian materialism and argued that it was nearly universal and almost never defended, because almost nobody noticed they held it.
In its place came the multiple drafts model: many parallel processes of content-fixation, at different sites and different times, with no master copy and no finish line. His best argument for it was the colour phi phenomenon, in which a green spot and a red spot flashed in quick succession are seen as one spot that changes colour in transit, before the red spot has appeared. You can say the brain revised the memory afterwards, which he called Orwellian, or that it delayed delivery and edited first, which he called Stalinesque. No experiment distinguishes them, and the appearance of a real question is, he argued, an artefact of the theatre metaphor. Third came heterophenomenology: treat reports as data about what the subject believes, granting authority about how things seem and none about what is happening inside.
The objection came immediately and has never gone away: that the book explains consciousness away rather than explaining it. Detractors circulated the alternative titles Consciousness Ignored and Consciousness Explained Away, repeated often enough since to become part of the book's reception, though no single author seems to have originated either. The sharpest named version came from John Searle, whose review in The New York Review of Books of 16 November 1995 charged Dennett with denying the existence of the data his book claimed to explain. Dennett replied on 21 December with a line that shows his temper as well as anything he wrote: "he has one argument, the Chinese Room, and he has been trotting it out, basically unchanged, for fifteen years." Searle's rejoinder: for Dennett "there is no difference between us humans and complex zombies who lack any inner feelings, because we are all just complex zombies." The impasse is the one traced in our account of the Chinese Room.
Dennett's answer had two parts. Intuitions about consciousness are not evidence but the output of thinking tools, which is why he spent so much effort designing better ones; and first-person report is an instrument with severe known distortions, so a theory obliged to vindicate every deliverance of introspection has surrendered before it begins. Thomas Nagel, reviewing From Bacteria to Bach and Back in the NYRB on 9 March 2017, put the counter best: Dennett is "maintaining a thesis at all costs", and an illusion of consciousness is itself a conscious experience, so the move cannot be made without self-refutation.
Why he thought the zombie was incoherent rather than merely wrong#
The quarrel with David Chalmers ran on the same fault line. Chalmers's conceivability argument holds that a being physically identical to you but with no experience is conceivable, that what is conceivable in the relevant sense is possible, and that physicalism is therefore false — the basis of the hard problem, whose thirtieth anniversary we covered last year.
Dennett's reply, in "The Unimagined Preposterousness of Zombies" (Journal of Consciousness Studies, 1995), was not that zombies are possible but unlikely. It was that people who think they have conceived one have not finished the job. A being functionally identical to you would notice its own states, report on them, be surprised by them, and write philosophy papers about the hard problem. Having stipulated all that, the zombie theorist subtracts something further and cannot say what it does. Such theorists, he wrote, "invariably underestimate the task of conception, and end up imagining something that violates their own definition". In "Facing Backwards on the Problem of Consciousness" (JCS, 1996) he called the hard problem "a major misdirector of attention, an illusion-generator", and ran an analogy with vitalism: once the functions of life are explained, nothing remains but the conviction that something ineffable must. Whether the analogy holds is what is in dispute. Dennett treated the question as settled.
Darwin's Dangerous Idea (Simon & Schuster, 1995, 586 pages) was a finalist for the National Book Award and the 1996 Pulitzer, and the most enjoyable thing he wrote. Natural selection, it argues, is a substrate-neutral algorithm, and an algorithm is a universal acid, eating through every concept it touches and leaving the landmarks recognisable but transformed. Explanations appealing to an unexplained source of design he called skyhooks; those building design up from lower-level processes, cranes. He also coined greedy reductionism for the opposite error of skipping intermediate levels, a concession his critics rarely credit him with.
The book's attack on Stephen Jay Gould produced the loudest public quarrel of his career. Gould and Richard Lewontin's 1979 paper on the spandrels of San Marco had argued that adaptationists habitually invent selective stories for features that are structural by-products. Dennett called that refutation an illusion and treated Gould's proposed revolutions, punctuated equilibrium among them, as false alarms.
Gould replied in the NYRB on 12 June 1997 with "Darwinian Fundamentalism", accusing Dennett of defending "a miserly and blinkered picture of evolution", saying the book read "like a caricature of a caricature", and delivering the line everyone remembers: "if T. H. Huxley truly acted as 'Darwin's bulldog,' then it is hard to resist thinking of Dennett, in this book, as 'Dawkins's lapdog.'" A companion piece followed on 26 June. Dennett answered on 14 August, arguing that Gould had lifted a quotation from the very pages in which Dennett was undoing years of Gouldian caricature. Gould's rejoinder: "he offers absolutely no intellectual response to any of my critiques."
The honours are divided. Gould was right that adaptive storytelling is a real methodological vice; Dennett was right that Gould's popular writing had left readers thinking orthodox selection overthrown when it had not. The most damaging review came from neither man. In Boston Review in June 1996 the evolutionary geneticist H. Allen Orr noted that substrate neutrality fails on its own terms, since selection needs particulate rather than blending inheritance to work at all, and that a book about the reach of natural selection said almost nothing about Motoo Kimura's neutral theory, under which much of molecular evolution is chance. For a thesis about universal acid, a large gap.
Competence without comprehension, and the objection to it#
The phrase that has outlived the quarrels is competence without comprehension, developed in From Bacteria to Bach and Back (W. W. Norton, 2017). The image that opens it: set a termite mound beside Gaudí's Sagrada Família. They look alike. One was designed top-down by a man who thought it out in advance; the other was made by millions of insects, none of which had any idea what it was building. "Evolution by natural selection is a process that's breathtakingly competent, but has no understanding at all."
Behind this is what Dennett, in a 2009 paper for the Proceedings of the National Academy of Sciences, called Darwin's strange inversion of reasoning. He borrowed it from a hostile critic, Robert Beverley MacKenzie, who complained in 1868 that on Darwin's theory "IN ORDER TO MAKE A PERFECT AND BEAUTIFUL MACHINE, IT IS NOT REQUISITE TO KNOW HOW TO MAKE IT." MacKenzie meant a reductio; Dennett took it for the discovery itself, and paired it with Turing's: to be a perfect and beautiful computing machine, it is not requisite to know what arithmetic is. Applied to language models the point is useful: competence is not evidence of comprehension, any more than the mound is evidence of an architect. That corrective to the reflex which takes fluency for understanding is why his name keeps surfacing in arguments about what we are doing when we talk to these systems.
The serious objection is that the phrase does less work than it appears to. Comprehension, on Dennett's own account, comes in degrees and is built out of competences; that is the architecture of the 2017 book, with its ascending tower of ways of generating and testing behaviour. If comprehension is a matter of degree, the celebrated phrase names one end of a continuum rather than one side of a division, and the question of where a given system sits is left exactly where it was. Nagel's charge was sharper: that Dennett slides from the true claim that much of what we do happens for reasons which are not our reasons to the far stronger claim that nothing is revealed from the first-person point of view at all. The phrase, his opponents say, sometimes settles by stipulation the question it was meant to open. They have a point.
Breaking the Spell: Religion as a Natural Phenomenon (Viking, 2006) made Dennett one of the Four Horsemen of the New Atheism after the filmed conversation with Dawkins, Harris and Hitchens in 2007. Julian Baggini, in The Philosophers' Magazine in 2010, called him "the least apocalyptic" of the four, and that is fair: the book proposes a research programme, not a polemic. His own defence was blunt. "There is no polite way of asking somebody: have you considered the possibility that your entire life has been devoted to a delusion?"
It cost him anyway. Leon Wieseltier's notice in The New York Times of 19 February 2006 dismissed the book as present-day scientism, and the association with the movement's louder members meant a decade of Dennett's work was read through it by philosophers who had never opened it.
He had been writing about machines from the start. Brainstorms (Bradford Books, 1978) already contains the material, including the thought experiment "Where Am I?"; in 1994 he published "The Practical Requirements for Making a Conscious Robot" in the Philosophical Transactions of the Royal Society, on his work with the MIT team building the humanoid robot Cog.
His late position deserves reporting precisely. In The Atlantic of 16 May 2023 he argued that the pressing danger was not machine consciousness but human credulity. Counterfeiting money has been a grave crime for millennia because it destroys the trust a society runs on, and the same logic, he thought, now applies to the manufacture of convincing artificial persons. His sentence: "These counterfeit people are the most dangerous artifacts in human history, capable of destroying not just economies but human freedom itself." He wanted the analogy taken literally: watermarking obligations, liability reaching company executives, and penalties to match. "The penalties for either offense should be extremely severe, given that civilization itself is at risk."
There is no contradiction with the earlier work, though it is easy to think there is. The intentional stance was always a claim about strategies, not about souls. Treating a system as a believer is something we do because it pays, and a technology that makes the strategy pay when it should not is, on his own terms, an attack on our predictive practices. He spent fifty years saying that "is it really conscious?" was the wrong question; his last argument was that asking the wrong question at scale might be dangerous enough to legislate against.
The durable Dennett is the toolmaker. Intuition Pumps and Other Tools for Thinking (2013) is the book to give a student: a catalogue of devices, including the instruction to treat the word "surely" as an alarm, since it almost always marks the place where an author has run out of argument. It also contains the four rules he took from the game theorist Anatol Rapoport and popularised. Re-express your opponent's position so well that they say "Thanks, I wish I'd thought of putting it that way"; list the points on which you agree; mention what you have learned from them; and only then say a word of criticism.
He did not always follow them. The Gould exchange is no model of rule four, and he could be brisk to the point of dismissiveness. Parts of the work have aged poorly: memetics never became the science he hoped for, the New Atheism years look more parochial each decade, and heterophenomenology has not persuaded many working consciousness scientists that the explanandum is safely accounted for. But the intentional stance is now common property in cognitive science, ethology and increasingly the study of artificial agents, mostly among people with no idea whose idea it was. The auditor's question — does this explanation do any work, or does it only feel as though it does — has survived him in better shape than most of his answers.
Yanis Varoufakis says capitalism is dead and that 'cloud capital' now extracts rent from digital fiefs rather than profit from markets. We set out the technofeudalism thesis as an argument to be tested, then weigh the strongest objections from Morozov, Snow and Gane, and Varoufakis's 2026 reply to his Marxist critics.
The Chinese-American philosopher Tu Weiming has spent sixty years pressing a single claim: that to become modern a society need not become Western. This profile sets out his idea of 'cultural China', the Boston Confucians, and his attempt to revive Confucian humanism as a living philosophy, and asks whether it holds.
.jpg?width=500 "Kant")
Comments
0Loading comments…