Monday, April 29, 2024

An expanded view of human minds and their reality.

I want to pass on this recent essay by Venkatesh Rao in its entirety, because it has changed my mind on agreeing with Daniel Dennett that the “Hard Problem” of consciousness is a fabrication that doesn’t actually exist. There are so many interesting ideas in this essay that I will be returning to it frequently in the future.  

We Are All Dennettians Now

An homage riff on AI+mind+evolution in honor of Daniel Dennett

The philosopher Daniel Dennett (1942-2024) died last week. Dennett’s contributions to the 1981 book he co-edited with Douglas Hofstadter, The Mind’s I,¹ which I read in 1996 (rather appropriately while doing an undergrad internship at the Center for AI and Robotics in Bangalore), helped shape a lot of my early philosophical development. A few years later (around 1999 I think), I closely read his trollishly titled 1991 magnum opus, Consciousness Explained (alongside Steven Pinker’s similar volume How the Mind Works), and that ended up shaping a lot of my development as an engineer. Consciousness Explained is effectively a detailed neuro-realistic speculative engineering model of the architecture of the brain in a pseudo-code like idiom. I stopped following his work closely at that point, since my tastes took me in other directions, but I did take care to keep him on my radar loosely.

So in his honor, I’d like to (rather chaotically) riff on the interplay of the three big topics that form the through-lines of his life and work: AI, the philosophy of mind, and Darwinism. Long before we all turned into philosophers of AI overnight with the launch of ChatGPT, he defined what that even means.

When I say Dennett’s views shaped mine, I don’t mean I necessarily agreed with them. Arguably, your early philosophical development is not shaped by discovering thinkers you agree with. That’s for later-life refinements (Hannah Arendt, whom I first read only a few years ago, is probably the most influential agree-with philosopher for me). Your early development is shaped by discovering philosophers you disagree with.

But any old disagreement will not shape your thinking. I read Ayn Rand too (if you want to generously call her a philosopher) around the same time I discovered Dennett, and while I disagreed with her too, she basically had no effect on my thinking. I found her work to be too puerile to argue with. But Dennett — disagreeing with him forced me to grow, because it took serious work over years to decades — some of it still ongoing — to figure out how and why I disagreed. It was philosophical weight training. The work of disagreeing with Dennett led me to other contemporary philosophers of mind like David Chalmers and Ned Block, and various other more esoteric bunnytrails. This was all a quarter century ago, but by the time I exited what I think of as the path-dependent phase of my philosophical development circa 2003, my thinking bore indelible imprints of Dennett’s influence.

I think Dennett was right about nearly all the details of everything he touched, and also right (and more crucially, tasteful) in his choices of details to focus on as being illuminating and significant. This is why he was able to provide elegant philosophical accounts of various kinds of phenomenology that elevated the corresponding discourses in AI, psychology, neuroscience, and biology. His work made him a sort of patron philosopher of a variety of youngish scientific disciplines that lacked robust philosophical traditions of their own. It also made him a vastly more relevant philosopher than most of his peers in the philosophy world, who tend, through some mix of insecurity, lack of courage, and illiteracy, to stay away from the dirty details of technological modernity in their philosophizing (and therefore cut rather sorry figures when they attempt to weigh in on philosophy-of-technology issues with cartoon thought experiments about trolleys or drowning children). Even the few who came close, like John Searle, could rarely match Dennett’s mastery of vast oceans of modern techno-phenomenological detail, even if they tended to do better with clever thought experiments. As far as I am aware, Dennett has no clever but misleading Chinese Rooms or Trolley Problems to his credit, which to my mind makes him a superior rather than inferior philosopher.

I suspect he paid a cost for his wide-ranging, ecumenical curiosities in his home discipline. Academic philosophers like to speak in a precise code about the simplest possible things, to say what they believe to be the most robust things they can. Dennett on the other hand talked in common language about the most complex things the human mind has ever attempted to grasp. The fact that he got his hands (and mind!) dirty with vast amounts of technical detail, and dealt in facts with short half-lives from fast-evolving fields, and wrote in a style accessible to any intelligent reader willing to pay attention, made him barely recognizable as a philosopher at all. But despite the cosmetic similarities, it would be a serious mistake to class him with science popularizers or TED/television scientists with a flair for spectacle at the expense of substance.

Though he had a habit of being uncannily right about a lot of the details, I believe Dennett was almost certainly wrong about a few critical fundamental things. We’ll get to what and why later, but the big point to acknowledge is that if he was indeed wrong (and to his credit, I am not yet 100% sure he was), he was wrong in ways that forced even his opponents to elevate their games. He was as much a patron philosopher (or troll or bugbear) to his philosophical rivals as to the scientists of the fields he adopted. You could not even be an opponent of Dennett except in Dennettian ways. To disagree with the premises of Strong AI or Dennett’s theory of mind is to disagree in Dennettian ways.

If I were to caricature how I fit in the Dennettian universe, I suspect I’d be closest to what he called a “mysterian” (though I don’t think the term originated with him). Despite mysterian being something of a dismissive slur, it does point squarely at the core of why his opponents disagree with him, and the parts of their philosophies they must work to harden and make rigorous, to withstand the acid forces of the peculiarly Dennetian mode of scrutiny I want to talk about here.

So to adapt the line used by Milton Friedman to describe Keynes: We are all Dennettians now.

Let’s try and unpack what that means.


As I said, in Dennettian terms, I am a “mysterian.” At a big commitments level, mysterianism is the polar opposite of the position Dennett consistently argued across his work, a version of what we generally call a “Strong AI” position. But at the detailed level, there are no serious disagreements. Mysterians and Strong AI people agree about most of the details of how the mind works. They just put the overall picture together differently because mysterians want to accommodate certain currently mysterious things that Strong AI people typically reject as either meaningless noise or shallow confusions/illusions.

Dennett’s version of Strong AI was both more robustly constructed than the sophomoric versions one typically encounters, and more broadly applied: beyond AI to human brains and seemingly intelligent processes like evolution. Most importantly, it was actually interesting. Reading his accounts of minds and computers, you do not come away with the vague suspicion of a non-neurotypical succumbing to the typical-mind fallacy and describing the inner life of a robot or philosophical zombie as “truth.” From his writing, it sounds like he had a fairly typical inner-life experience, so why did he seem to deny the apparent ineffable essence of it? Why didn’t he try to eff that essence the way Chalmers, for instance, does? Why did he seemingly dismiss it as irrelevant, unreal, or both?

To be a mysterian in Dennettian terms is to take ineffable, vitalist essences seriously. With AIs and minds, it means taking the hard problem of consciousness seriously. With evolution, it means believing that Darwinism is not the whole story. Dennett tended to use the term as a dismissive slur, but many, (re)claim it as a term of approbation, and I count myself among them.

To be a rigorous mysterian, as opposed to one of the sillier sorts Dennett liked to stoop to conquer (naive dualists, intelligent-designers, theological literalists, overconfident mystics…), you have to take vitalist essences “seriously but not literally.” My version of doing that is to treat my vitalist inclinations as placeholder pointers to things that lurk in the dank, ungrokked margins of the thinkable, just beyond the reach of my conceptualizing mind. Things I suspect exist by the vague shapes of the barely sensed holes they leave in my ideas. In pursuit of such things, I happily traffic in literary probing of Labatutian/Lovecraftian/Ballardian varieties, self-consciously magical thinking, junk from various pre-paradigmatic alchemical thought spaces, constructs that uncannily resemble astrology, and so on. I suppose it’s a sort of intuitive-ironic cognitive kayfabe for the most part, but it’s not entirely so.

So for example, when I talk of elan vital, as I frequently do in this newsletter, I don’t mean to imply I believe in some sort of magical fluid flowing through living things or a Gaian planetary consciousness. Nor do I mean the sort of overwrought continental metaphysics of time and subjectivity associated with Henri Bergson (which made him the darling of modernist literary types and an object of contempt to Einstein). I simply mean I suspect there are invisible things going on in the experience and phenomenology of life that are currently beyond my ability to see, model, and talk about using recognizably rational concepts, and I’d rather talk about them as best I can with irrational concepts than pretend they don’t exist.

Or to take another example, when I say that “Darwin is not the whole story,” I don’t mean I believe in intelligent design or a creator god (I’m at least as strong an atheist as Dennett was). I mean that Darwinian principles of evolution constrain but do not determine the nature of nature, and we don’t yet fully grok what completes the picture except perhaps in hand-wavy magical-thinking ways. To fully determine what happens, you need to add more elements. For example, you can add ideas like those of Stuart Kauffman and other complexity theorists. You could add elements of what Maturana and Varela called autopoiesis. Or it might be none of these candidate hole-filling ideas, but something to be dreamt up years in the future. Or never. But just because there are only unsatisfactory candidate ways for talking about stuff doesn’t mean you should conclude the stuff doesn’t exist.

In all such cases, there are more things present in phenomenology I can access than I can talk about using terms of reference that would be considered legitimate by everybody. This is neither known-unknowns (which are holes with shapes defined by concepts that seem rational), nor unknown-unknown (which have not yet appeared in your senses, and therefore, to apply a Gilbert Ryle principle, cannot be in your mind).

These are things that we might call magically known. Like chemistry was magically known through alchemy. For phenomenology to be worth magically knowing, the way-of-knowing must offer interesting agency, even if it doesn’t hang together conceptually.

Dennett seemed to mostly fiercely resist and reject such impulses. He genuinely seemed to think that belief in (say) the hard problem of consciousness was some sort of semantic confusion. Unlike say B. F. Skinner, whom critics accused of only pretending to not believe in inner processes, Dennett seemed to actually disbelieve in them.

Dennett seemed to disregard a cousin to the principle that absence of evidence is not evidence of absence: Presence of magical conceptualizations does not mean absence of phenomenology. A bad pointer does not disprove the existence of what it points to. This sort of error is easy to avoid in most cases. Lightning is obviously real even if some people seem to account for it in terms of Indra wielding his vajra. But when we try to talk of things that are on the phenomenological margins, barely within the grasp of sensory awareness, or worse, potentially exist as incommunicable but universal subjective phenomenology (such as the experience of the color “blue”), things get tricky.

Dennett was a successor of sorts to philosophers like Gilbert Ryle, and psychologists like B. F. Skinner. In evolutionary philosophy, his thinking aligned with people like Richard Dawkins and Steven Pinker, and against Noam Chomsky (often classified as a mysterian, though I think the unreasonable effectiveness of LLMs kinda vindicates to a degree Chomsky’s notions of an ineffable more-than-Darwin essence around universal grammars that we don’t yet understand).

I personally find it interesting to poke at why Dennett took the positions he took, given that he was contemplating the same phenomenological data and low-to-mid-level conceptual categories as the rest of us (indeed, he supplied much of it for the rest of us). One way to get at it is to ask: Was Dennett a phenomenologist? Are the limits of his ideas are the limits of phenomenology?

I think the answers are yes and yes, but he wasn’t a traditional sort of phenomenologist, and he didn’t encounter the more familiar sorts of limits.

The Limits of Phenomenology

Let’s talk regular phenomenology first, before tackling what I think was Dennett’s version.

I think of phenomenology, as a working philosophical method, to be something like a conceited form of empiricism that aims to get away from any kind of conceptually mediated seeing.

When you begin to inquire into a complex question with any sort of fundamentally empirical approach, your philosophy can only be as good as a) the things you know now through your (potentially technologically augmented) senses and b) the ways in which you know those things.

The conceit of phenomenology begins with trying to “unknow” what is known to be known, and contemplate the resulting presumed “pure” experiences “directly.” There are various flavors of this: Husserlian bracketing in the Western tradition, Zen-like “beginner mind” practices, Vipassana style recursive examination of mental experiences, and so on. Some flavors apply only to sense-observations of external phenomena, others apply only to subjective introspection, and some apply to both. Given the current somewhat faddish uptick in Eastern-flavored disciplines of interiority, it is important to take note of the fact that the phenomenological attitude is not necessarily inward-oriented. For example, the 19th century quest to measure a tenth of a second, and factor out the “human equation” in astronomical observations, was a massive project in Western phenomenology. The abstract thought experiments with notional clocks in the theory of relativity began with the phenomenology of real clocks.

In “doing” phenomenology, you are assuming that you know what you know relatively completely (or can come to know it), and have a reliable procedure for either unknowing it, or systematically alloying it with skeptical doubt, to destabilize unreliable perceptions it might be contributing to. Such destabilizability of your default, familiar way of knowing, in pursuit of a more-perfect unknowing, is in many ways the essence of rationality and objectivity. It is the (usually undeclared) starting posture for doing “science,” among other things.

Crucially, for our purposes in this essay, you do not make a careful distinction between things you know in a rational way and things you know in a magical or mysterian way, but effectively assume that only the former matter; that the latter can be trivially brushed aside as noise signifying nothing that needs unknowing. I think the reverse is true. It is harder, to the point of near impossible, to root out magical ideas from your perceptions, and they signify the most important things you know. More to the point, it is not clear that trying to unknow things, especially magical things, is in fact a good idea, or that unknowing is clarifying rather than blinding. But phenomenology is committed to trying. This has consequences for “phenomenological projects” of any sort, be they Husserlian or Theravadan in spirit.

A relatively crude example: “life” becomes much less ineffable (and depending on your standards, possibly entirely drained of mystery) once you view it through the lens of DNA. Not only do you see new things through new tools, you see phenomenology you could already see, such as Mendelian inheritance, in a fundamentally different way that feels phenomenologically “deeper” when in fact it relies on more conceptual scaffolding, more things that are invisible to most of us, and requires instruments with elaborate theories attached to even render intelligible. You do not see “ATCG” sequences when contemplating a pea flower. You could retreat up the toolchain and turn your attention to how instruments construct the “idea of DNA” but to me that feels like a usually futile yak shave. The better thing to do is ask why a more indirect way of knowing somehow seems to perceive more clearly than more direct ways.

It is obviously hard to “unsee” knowledge of DNA today when contemplating the nature of life. But it would have been even harder to recognize that something “DNA shaped” was missing in say 1850, regardless of your phenomenological skills, by unknowing things you knew then. In fact, clearing away magical ways of knowing might have swept away critical clues.

To become aware, as Mendel did, that there was a hidden order to inheritance in pea flowers, takes a leap of imagination that cannot be purely phenomenological. To suspect in 1943, as Schrodinger did, the existence of “aperiodic covalent bonded crystals” at the root of life, and point the way to DNA, takes a blend of seeing and knowing that is greater than either. Magical knowing is pathfinder-knowing that connects what we know and can see to what we could know and see. It is the bootstrapping process of the mind.

Mendel and Schrodinger “saw” DNA before it was discovered, in terms of reference that would have been considered “rational” in their own time, but this has not always been the case. Newton, famously, had a lot of magical thinking going on in his successful quest for a theory of gravity. Kepler was a numerologist. Leibniz was ball of mad ideas. One of Newton’s successful bits of thinking, the idea of “particles” of light, which faced off against Huygens’ “waves,” has still not exited the magical realm. The jury is still out in our time about whether quantized fields are phenomenologically “real” or merely a convenient mnemonic-metaphoric motif for some unexpected structure in some unreasonably effective math.

Arguably, none of these thinkers was a phenomenologist, though all had a disciplined empirical streak in their thinking. The history of their ideas suggests that phenomenology is no panacea for philosophical troubles with unruly conceptual universes that refuse to be meekly and rationally “bracketed” away. There is no systematic and magic-free way to march from current truths to better ones via phenomenological disciplines.

The fatal conceit of naive phenomenology (which Paul Feyerabend spotted) is the idea that there is privileged reliable (or meta-reliable) “technique” of relating to your sense experiences, independent of the concepts you hold, whether that “technique” is Husserlian bracketing or vipassana. Understood this way, theories of reality are not that different from physical instruments that extend our senses. Experiment and theory don’t always expose each other’s weaknesses. Sometimes they mutually reinforce them.

In fact, I would go so far as to suggest—and I suspect Dennett would have agreed—that there is no such thing as phenomenology per se. All we ever see is the most invisible of our theories (rational and magical), projected via our senses and instruments (which shape, and are shaped by, those theories), onto the seemingly underdetermined aspects of the universe. There are only incomplete ways of knowing and seeing within which ideas and experiences are inextricably intertwined. No phenomenological method can consistently outperform methodological anarchy.

To deny this is to be a traditional phenomenologist, striving to procedurally separate the realm of ideas and concepts from the realm of putatively unfactored and “directly perceived” (a favorite phrase of meditators) “real” experiences.

Husserlian bracketing — “suspending trust in the objectivity of the world” — is fine in theory, but not so easy in practice. How do you know that you’re setting aside preconceived notions, judgments, and biases and attending to a phenomenon as it truly is? How do you set aside the unconscious “theory” that the Sun revolves around the Earth, and open your mind to the possibility that it’s the other way around? How do you “see” DNA-shaped holes in current ways of seeing, especially if they currently manifest as strange demons that you might sweep away in a spasm of over-eager epistemic hygiene? How do you relate, as a phenomenologist, to intrinsically conceptual things like electrons and positrons that only exist behind layers of mathematics describing experimental data processed through layers of instrumentation conceived by existing theories? If you can’t check the math yourself, how can you trust that the light bulb turning on is powered by those “electrons” tracing arcs through cloud chambers?

In practice, we know how such shifts actually came about. Not because philosophers meditated dispassionately on the “phenomenology” with free minds seeing reality as it “truly is,” but because astronomers and biologists with heads full of weird magical notions looked through telescopes and microscopes, maintained careful notes of detailed measurements, informed by those weird magical theories, and tried to account for discrepancies. Tycho Brahe, for instance, who provided the data that dethroned Ptolemy, believed in some sort of Frankenstein helio-geo-centric Ptolemy++ theory. Instead of explaining the discrepancies, as Kepler did later, Brahe attempted to explain them away using terms of reference he was attached to. He failed to resolve the tension. But he paved the way to Kepler resolving that particular tension (who of course introduced new ones, while lost in his own magical thinking about platonic solids). Formally phenomenological postures were not just absent from the story, but would have arguably retarded it by being too methodologically conservative.

Phenomenology, in other words, is something of a procedural conceit. An uncritical trust in self-certifying ways of seeing based entirely on how compelling they seem to the seer. The self-certification follows some sort of seemingly rational procedure (which might be mystical but still rational in the sense of being coherent and disciplined and internally consistent) but ultimately derives its authority from the intuitive certainties and suspicions of the perceiving subject. Phenomenological procedures are a kind of rule-by-law for governing sense experience in a laissez-faire way, rather than the “objective” rule-of-law they are often presented as. Phenomenology is to empiricism as “socialism with Chinese characteristics” is to liberal democracy.

This is not to say phenomenology is hopelessly unreliable or useless. All methodologies have their conceits, which manifest as blindspots. With phenomenology, the blindspot manifests as an insistence on non-magicality. The phenomenologist fiercely rejects the Cartesian theater and the varied ghosts-in-machines that dance there. The meditator insists he is “directly perceiving” reality in a reproducible way, no magic necessary. I do not doubt that these convictions are utterly compelling to those who hold them; as compelling as the incommunicable reality of perceiving “blue” is to everybody. I have no particular argument with such insistence. What I actually have a problem with is the delegitimization of magical thinking in the process, which I suspect to be essential for progress.

My own solution is to simply add magical thinking back into the picture for my own use, without attempting to defend that choice, and accepting the consequences.

For example, I take Myers-Briggs and the Enneagram seriously (but not literally!). I believe in the hard problem of consciousness, and therefore think “upload” and “simulationism” ideas are not-even-wrong. I don’t believe in Gods or AGIs, and therefore don’t see the point of Pascal’s wager type contortions to avoid heaven/hell or future-simulated-torture scenarios. In each case my commitments rely on chains of thought that are at least partly magical thinking, and decidedly non-phenomenological, which has various social consequences in various places. I don’t attempt to justify any of it because I think all schemes of justification, whether they are labeled “science” or something else, rest on traditional phenomenology and its limits.

Does this mean solipsism is the best we can hope for? This is where we get back to Dennett.

Dennett, to his credit, I don’t think he was a traditional phenomenologist, and he mostly avoided all the traps I’ve pointed out, including the trap of solipsism. Nor was he what one might call a “phenomenologist of language” like most modern analytical philosophers in the West. He was much too interested in technological modernity (and the limits of thought it has been exposing for a century) to be content with such a shrinking, traditionalist philosophical range.

But he was a phenomenologist in the broader sense of rejecting the possible reality of things that currently lack coherent non-magical modes of apprehension.

So how did he operate if not in traditional phenomenological ways?

Demiurge Phenomenology

I believe Dennett was what we might call a demiurge phenomenologist, which is a sort of late modernist version of traditional phenomenology. It will take a bit of work to explain what I mean by that.

I can’t recall if he ever said something like this (I’m definitely not a completist with his work and have only read a fraction of his voluminous output), but I suspect Dennett believed that the human experience of “mind” is itself subject to evolutionary processes (think Jaynes and bicameral mind theories for example — I seem to recall him saying something approving about that in an interview somewhere). He sought to construct philosophy in ways that did not derive authority from an absolute notion of the experience of mind. He tried to do relativity theory for minds, but without descending into solipsism.

It is easiest to appreciate this point by starting with body experience. For example, we are evolved from creatures with tails, but we do not currently possess tails. We possess vestigial “tail bones” and presumably bits of DNA relevant to tails, but we cannot know what it is like to have a tail (or in the spirit of mysterian philosopher Thomas Nagel’s What is it Like to Be a Bat provocation, which I first read in The Mind’s I, what it is like for a tailed creature to have a tail).

We do catch tantalizing Lovecraftian-Ballardian glimpses of our genetic heritage though. For example, the gasping reflex and shot of alertness that accompanies being dunked in water (the mammalian dive reflex) is a remnant of a more aquatic evolutionary past that far predates our primate mode of existence. Now apply that to the experience of “mind.”

Why does Jaynes’ bicameral mind theory sound so fundamentally crackpot to modern minds? It could be that the notion is actually crackpot, but you cannot easily dismiss the idea that it’s actually a well-posed notion that only appears crackpot because we are not currently possessed of bicameral mind-experiences (modulo cognitive marginalia like tulpas and internal family systems — one of my attention/taste biases is to index strongly on typical rather than rare mental experiences; I believe the significance of the latter is highly overstated due to the personal significance they acquire in individual lives).

I hope it is obvious why the possibility that the experience of mind is subject to evolution is fatal to traditional phenomenology. If despite all the sophistication of your cognitive toolchain (bracketing, jhanas, ketamine, whatever), it turns out that you’re only exploring the limits of the evolutionarily transient and arbitrary “variety of mind” that we happen to experience, what does that say about the reliability of the resulting supposedly objective or “direct” perceptions of reality itself that such a mind mediates?

This, by the way, is a problem that evolutionary terms of reference make elegantly obvious, but you can get here in other ways. Darwinian evolution is convenient scaffolding to get there (and the one I think Dennett used), but ultimately dispensable. But however you get there, the possibility that experiences of mind are relative to contingent and arbitrary evolutionary circumstances is fatal to the conceits of traditional phenomenology. It reduces traditional phenomenology in status to any old sort of Cartesian or Platonic philosophizing with made-up bullshit schemas. You might as well make 2x2s all day like I sometimes do.

The Eastern response to this quandary has traditionally been rather defeatist — abandoning the project of trying to know reality entirely. Buddhist and Advaita philosophies in particular, tend to dispense with “objective reality” as an ontologically meaningful characterization of anything. There is only nothing. Or only the perceiving subject. Everything else is maya-moh, a sentimental attachment to the ephemeral unreal. Snap out of it.

I suspect Western philosophy was starting to head that way in the 16th century (through the Spinoza-vs-Leibniz shadowboxing years), but was luckily steered down a less defeatist path to a somewhat uneasy detente between a sort of “probationary reality” accessed through technologically augmented senses, and a subjectivity resolutely bound to that probationary reality via the conceits of traditional phenomenology. This is a long-winded way of saying “science happened” to Western philosophy.

I think that detente is breaking down. One sign is the growing popularity of the relatively pedestrian metaphysics of physicists like Donald Hoffman (leading to a certain amount of unseemly glee among partisans of Eastern philosophies — “omigod you think quantum mechanics shows reality is an illusion? Welcome to advaita lol”).

But despite these marginally interesting conversations, and whether you get there via Husserl, Hoffman, or vipassana, we’re no closer to resolving what we might call the fundamental paradox of phenomenology. If our experience of mind is contingent, how can any notion of justifiable absolute knowledge be sustained? We are essentially stopped clocks trying to tell the time.

Dennett, I think favored one sort of answer: That the experience of mind was too untrustworthy and transient to build on, but that mind’s experience of mathematics was both trustworthy and absolute. Bicameral or monocameral, dolphin-brain or primate-brain, AI-brain or Hoffman-optimal ontological apparatus, one thing that is certain is that a prime number is a prime number in all ways that reality (probationary or not, illusory or not) collides with minds (typical or atypical, bursting with exotic qualia or full of trash qualia). Even the 17 and 13 year cicadas agree. Prime numbers constitute a fixed point in all the ways mind-like things have experience-like things in relation to reality-like things, regardless of whether minds, experiences, and reality are real. Prime numbers are like a motif that shows up in multiple unreliable dreams. If you’re going to build up a philosophy of being, you should only use things like prime numbers.

This is not just the most charitable interpretation of Dennett’s philosophy, but the most interesting and powerful one. It’s not that he thought of the mysterian weakness for ineffable experiences as being particularly “illusory”. As far as he was concerned, you could dismiss the “experience of mind” in its entirety as irrelevant philosophically. Even the idea that it has an epiphenomenal reality need not be seriously entertained because the thing that wants to entertain that idea is not to be trusted.

You see signs of this approach in a lot of his writing. In his collaborative enquires with Hofstadter, in his fundamentally algorithmic-mathematical account of evolution, in his seemingly perverse stances in debates both with reputable philosophers of mind and disreputable intelligent designers. As far as he was concerned, anyone who chose to build any theory of anything on the basis of anything other than mathematical constancy was trusting the experience of mind to an unjustifiable degree.

Again, I don’t know if he ever said as much explicitly (he probably did), but I suspect he had a basic metaphysics similar to that of another simpatico thinker on such matters, Roger Penrose: as a triad of physical/mental/platonic-mathematical worlds projecting on to each other in a strange loop. But unlike Penrose, who took the three realms to be equally real (or unreal) and entangled in an eternal dance of paradox, he chose to build almost entirely on the Platonic-mathematical vertex, with guarded phenomenological forays to the physical world, and strict avoidance of the mental world as a matter of epistemological hygiene.

The guarded phenomenological forays, unlike those of traditional phenomenologists, were governed by an allow list rather than a block list. Instead of trying to “block out” suspect conceptual commitments with bracketing or meditative discipline, he made sure to only work with allowable concepts and percepts that seemed to have some sort of mathematical bones to them. So Turing machines, algorithms, information theory, and the like, all made it into his thinking in load-bearing ways. Everything else was at best narrative flavor or useful communication metaphors. People who took anything else seriously were guilty of deep procedural illusions rather than shallow intellectual confusions.

If you think about it, his accounts of AI, evolution, and the human mind make a lot more sense if you see them as outcomes of philosophical construction processes governed by one very simple rule: Only use a building block if it looks mathematically real.

Regardless of what you believe about the reality of things other than mathematically underwritten ones, this is an intellectually powerful move. It is a kind of computational constructionism applied to philosophical inquiry, similar to what Wolfram does with physics on automata or hypergraphs, or what Grothendieck did with mathematics.

It is also far harder to do, because philosophy aims and claims to speak more broadly and deeply than either physics or mathematics.

I think Dennett landed where he did, philosophically, because he was essentially trying to rebuild the universe out of a very narrow admissible subset of the phenomenological experience of it. Mysterian musings didn’t make it in because they could not ride allowable percepts and concepts into the set of allowable construction materials.

In other words, he practiced demiurge phenomenology. Natural philosophy as an elaborate construction practice based on self-given rules of construction.

In adopting such an approach he was ahead of his time. We’re on the cusp of being able to literally do what he tried to do with words — build phenomenologically immersive virtual realities out of computational matter that seem to be defined by nothing more than mathematical absolutes, and have almost no connection even to physical reality, thanks to the seeming buffering universality of Turing-equivalent computation.

In that almost, I think, lies the key to my fundamental disagreement with Dennett, and my willingness to wander in magical realms of thought where mathematically sure-footed angels fear to tread. There are… phenomenological gaps between mathematical reconstructions of reality by energetic demiurges (whether they work with powerful arguments or VR headsets) and reality itself.

The biggest one, in my opinion, is the experience of time, which seems to oddly resist computational mathematization (though Stephen Wolfram claims to have one… but then he claims to have a lot of things). In an indirect way, disagreeing with Dennett at age 20 led me to my lifelong fascination with the philosophy of time.

Where to Next?

It is something of a cliche that over the last century or two, philosophy has gradually and reluctantly retreated from an increasing number of the domains it once claimed as its own, as scientific and technological advances rendered ungrounded philosophical ideas somewhere between moot and ridiculous. Bergson retreating in the face of the Einsteinian assault, ceding the question of the nature of time to physics, is probably as good a historical marker of the culmination of the process as any.

I would characterize Dennett as a late modernist philosopher in relation to this cliche. Unlike many philosophers, who simply gave up on trying to provide useful accounts of things that science and technology were beginning to describe in inexorably more powerful ways, he brought enormous energy to the task of simply keeping up. His methods were traditional, but his aim was radical: Instead of trying to provide accounts of things, he tried to provide constructions of things, aiming to arrive at a sense of the real through philosophical construction with admissible materials. He was something like Brouwer in mathematics, trying to do away with suspect building blocks to get to desirable places only using approved methods.

This actually worked very well, as far as it went. For example, I think his philosophy of mind was almost entirely correct as far as the mechanics of cognition go, and the findings of modern AI vindicate a lot of the specifics. For example, his idea of a “multiple drafts” model of cognition (where one part of the brain generates a lot of behavioral options in a bottom-up, anarchic way, and another part chooses a behavior from among them) is basically broadly correct, not just as a description of how the brain works, but of how things like LLMs work. But unlike many other so-called philosophers of AI he disagreed with, like Nick Bostrom, Dennett’s views managed to be provocative without being simplistic, opinionated without being dogmatic. He appeared to have a Strong AI stance similar to many people I disagree with, but unlike most of those people, I found his views worth understanding with some care, and hard to dismiss as wrong, let alone not-even-wrong.

I like to think he died believing his philosophies — of mind, AI, and Darwinism — to be on the cusp of a triumphant redemption. There are worse ways to go than believing your ideas have been thoroughly vindicated. And indeed, there was a lot Dennett got right. RIP.

Where do we go next with Dennettian questions about AI, minds, and evolution?

Oddly enough, I think Dennett himself pointed the way: demiurge phenomenology is the way. We just need to get more creative with it, and admit magical thinking into the process.

Dennett, I think, approached his questions the way some mathematicians originally approached Euclid’s fifth postulate: Discard it and try to either do without, or derive it from the other postulates. That led him to certain sorts of demiurgical constructions of AI, mind, and evolution.

There is another, equally valid way. Just as other mathematicians replaced the fifth postulates with alternatives and ended up with consistent non-Euclidean geometries, I think we could entertain different mysterian postulates and end up with a consistent non-Dennettian metaphysics of AI, mind, and evolution. You’d proceed by trying to do your own demiurgical constructing of a reality. An alternate reality.

For instance, what happens if you simply assume that there is human “mind stuff” that ends with death and cannot be uploaded or transferred to other matter, and that can never emerge in silico. You don’t have to try accounting for it (no need to mess with speculations about the pineal gland like Descartes, or worry about microtubules and sub-Planck-length phenomena like Penrose). You could just assume that consciousness is a thing like space or time, and run with the idea and see where you land and what sort of consistent metaphysical geometries are possible. This is in fact what certain philosophers of mind like Ned Block do.

The procedure can be extended to other questions as well. For instance, if you think Darwin is not the whole story with evolution, you could simply assume there are additional mathematical selection factors having to do with fractals or prime numbers, and go look for them, as the Santa Fe biologists have done. Start simple and stupid, for example, by applying a rule that “evolution avoids rectangles” or “evolution cannot get to wheels made entirely of grown organic body parts” and see where you land (for the latter, note that the example in Dark Materials trilogy cheats — that’s an assembled wheel, not an evolved one).

But all these procedures follow the basic Dennettian approach of demiurgical constructionist phenomenology. Start with your experiences. Let in an allow-list of percepts as concepts. Add an arbitrarily constructed magical suspicion or two. Let your computer build out the entailments of those starter conditions. See what sort of realities you can conjure into being. Maybe one of them will be more real than your current experience of reality. That would be progress. Perhaps progress only you can experience, but still, progress.

Would such near-solipsistic activities constitute a collective philosophical search for truth? I don’t know. But then, I don’t know if we have ever been on a coherent collective philosophical search for truth. All we’ve ever had is more or less satisfying descriptions of the primal mystery of our own individual existence.

Why is there something, rather than nothing, it is like, to be me?

Ultimately, Dennett did not seem to find that question to be either interesting or serious. But he pointed the way for me to start figuring out why I do. And that’s why I too am a Dennettian.

footnote  1
I found the book in my uncle’s library, and the only reason I picked it up was because I recognized Hofstadter’s name because Godel, Escher, Bach had recently been recommended to me. I think it’s one of the happy accidents of my life that I read The Mind’s I before I read Hofstadter’s Godel, Escher, Bach. I think that accident of path-dependence may have made me a truly philosophical engineer as opposed to just an engineer with a side interest in philosophy. Hofstadter is of course much better known and familiar in the engineering world, and reading him is something of a rite of passage in the education of the more sentimental sorts of engineers. But Hofstadter’s ideas were mostly entertaining and informative for me, in the mode of popular science, rather than impactful. Dennett on the other hand, was impactful.

Wednesday, April 10, 2024

The world of decentralized everything.

Following up on my last post on the Summer of Protocols sessions, I want to pass on (again, to my future self, and possibly a few techie MindBlog readers) a few links to the world of decentralized grass roots everything - commerce, communications, finance, etc.  - trying to bypass the traditional powers and gate keepers in these areas by constructing distributed systems usually based on block chains and cryptocurrencies.  I am trying to learn more about this, taking things in small steps to avoid overload headaches... (One keeps stumbling on areas of world wide engagement of thousands of very intelligent minds.)

Here is a worthwhile read of the general idea from the Ethereum Foundation.

I've described getting into one decentralized context by setting up a Helium Mobile network hotspot, as well as my own private Helium Mobile Cellular account. To follow this up, I pass on a link in an email from Helium pointing to its participation in Consensus24 May 29-31 in Austin TX (where I now live) sponsored by CoinDesk.  At look at the agenda for that meeting gives you an impression of the multiple engagements of government regulatory agencies, business, and crypto-world that are occurring.

Monday, April 08, 2024

New protocols for uncertain times.

I want to point to a project launched by Venkatest Rao and others last year: “The Summer of Protocols.”  Some background for this project can be found in his essay “In Search of Hardness”.  Also,  “The Unreasonable Sufficiency of Protocols”  essay by Rao et al. is an excellent presentation of what protocols are about.  I strongly recommend that you read it if nothing else. 

Here is a description of the project: 

Over 18 weeks in Summer 2023, 33 researchers from diverse fields including architecture, law, game design, technology, media, art, and workplace safety engaged in collaborative speculation, discovery, design, invention, and creative production to explore protocols, boadly construed, from various angles.

Their findings, catalogued here in six modules, comprise a variety of textual and non-textual artifacts (including art works, game designs, and software), organized around a set of research themes: built environments, danger and safety, dense hypermedia, technical standards, web content addressability, authorship, swarms, protocol death, and (artificial) memory.
I have read through through Module One for 2003, and it is solid interesting deep dive stuff.  Module 2 is also available. Modules 3-6 are said to be 'coming soon’  (as of 4/4/24, four months into a year that has Summer of Protocols program 2024 already underway, with the deadline for proposals 4/12/24.)

Here is one clip from the “In Search of Hardness” essay:

…it’s only in the last 50 years or so, with the rise of communications technologies, especially the internet and container shipping, and the emergence of unprecedented planet-scale coordination problems like climate action, that protocols truly came into focus as first-class phenomena in our world; the sine qua non of modernity. The word itself is less than a couple of centuries old.

And it wasn’t until the invention of blockchains in 2009 that they truly came into their own as phenomena with their own unique technological and social characteristics, distinct from other things like machines, institutions, processes, or even algorithms.

Protocols are engineered hardness, and in that, they’re similar to other hard, enduring things, ranging from diamonds and monuments to high-inertia institutions and constitutions.

But modern protocols are more than that. They’re not just engineered hardness, they are programmable, intangible hardness. They are dynamic and evolvable. And we hope they are systematically ossifiable for durability. They are the built environment of digital modernity.”

Friday, April 05, 2024

Our seduction by AI’s believable human voice.

 I want to point to an excellent New Yorker article by Patrick House titled  “The Lifelike Illusion of A.I.”  The article strikes home for me, for when a Chat Bot responds to one of my prompts using the pronoun “I”  I unconsciously attribute personhood to the machine, forgetting that this is a cheap trick used by programmers of large language model to increase the plausibility of responses.

House starts off his article by describing the attachments people formed with the Furby, an animatronic toy resembling a small owl, and Pleo, an animatronic toy dinosaur. Both use a simple set of rules to make the toys appear to be alive. Furby’s eyes move up and down in a way meant to imitate an infant’s eye movements while scanning a parent’s face. Pleo mimes different emotional behaviors when touched differently.
For readers who hit the New Yorker paywall when they click the above link, here are a few clips from the article that I think get across the main points:
“A Furby possessed a pre-programmed set of around two hundred words across English and “Furbish,” a made-up language. It started by speaking Furbish; as people interacted with it, the Furby switched between its language dictionaries, creating the impression that it was learning English. The toy was “one motor—a pile of plastic,” Caleb Chung, a Furby engineer, told me. “But we’re so species-centric. That’s our big blind spot. That’s why it’s so easy to hack humans.” People who used the Furby simply assumed that it must be learning.”
Chung considers Furby and Pleo to be early, limited examples of artificial intelligence—the “single cell” form of a more advanced technology. When I asked him about the newest developments in A.I.—especially the large language models that power systems like ChatGPT—he compared the intentional design of Furby’s eye movements to the chatbots’ use of the word “I.” Both tactics are cheap, simple ways to increase believability. In this view, when ChatGPT uses the word “I,” it’s just blinking its plastic eyes, trying to convince you that it’s a living thing.
We know that, in principle, inanimate ejecta from the big bang can be converted into thinking, living matter. Is that process really happening in miniature at server farms maintained by Google, Meta, and Microsoft? One major obstacle to settling debates about the ontology of our computers is that we are biased to perceive traces of mind and intention even where there are none. In a famous 1944 study, two psychologists, Marianne Simmel and Fritz Heider, had participants watch a simple animation of two triangles and a circle moving around one another. They then asked some viewers what kind of “person” each of the shapes was. People described the shapes using words like “aggressive,” “quarrelsome,” “valiant,” “defiant,” “timid,” and “meek,” even though they knew that they’d been watching lifeless lines on a screen.
…chatbots are designed by teams of programmers, executives, and engineers working under corporate and social pressures to make a convincing product. “All these writers and physicists they’re hiring—that’s game design,” he said. “They’re basically making levels.” (In August of last year, OpenAI acquired an open-world-video-game studio, for an undisclosed amount.) Like a game, a chatbot requires user input to get going, and relies on continued interaction. Its guardrails can even be broken using certain prompts that act like cheat codes, letting players roam otherwise inaccessible areas. Blackley likened all the human tinkering involved in chatbot training to the set design required for “The Truman Show,” the TV program within the eponymous film. Without knowing it, Truman has lived his whole life surrounded not by real people but by actors playing roles—wife, friend, milkman. There’s a fantasy that “we’ve taken our great grand theories of intelligence and baked them into this model, and then we turned it on and suddenly it was exactly like this,” Blackley went on. “It’s much more like Truman’s show, in that they tweak it until it seems really cool.”
A modern chatbot isn’t a Furby. It’s not a motor and a pile of plastic. It’s an analytic behemoth trained on data containing an extraordinary quantity of human ingenuity. It’s one of the most complicated, surprising, and transformative advances in the history of computation. A Furby is knowable: its vocabulary is limited, its circuits fixed. A large language model generates ideas, words, and contexts never before known. It is also—when it takes on the form of a chatbot—a digital metamorph, a character-based shape-shifter, fluid in identity, persona, and design. To perceive its output as anything like life, or like human thinking, is to succumb to its role play.

Wednesday, April 03, 2024

Neurons help flush waste out of our brains during sleep

More information (summarized here) on what is happening in our brains while we sleep is provided by Jiang-Xie et al.,, who show that active neurons can stimulate the clearance of their own metabolic waste by driving changes to ion gradients in the surrounding fluid and by promoting the pulsation of nearby blood vessels.  Here is the Jiang-Xie et al.abstract:

The accumulation of metabolic waste is a leading cause of numerous neurological disorders, yet we still have only limited knowledge of how the brain performs self-cleansing. Here we demonstrate that neural networks synchronize individual action potentials to create large-amplitude, rhythmic and self-perpetuating ionic waves in the interstitial fluid of the brain. These waves are a plausible mechanism to explain the correlated potentiation of the glymphatic flow through the brain parenchyma. Chemogenetic flattening of these high-energy ionic waves largely impeded cerebrospinal fluid infiltration into and clearance of molecules from the brain parenchyma. Notably, synthesized waves generated through transcranial optogenetic stimulation substantially potentiated cerebrospinal fluid-to-interstitial fluid perfusion. Our study demonstrates that neurons serve as master organizers for brain clearance. This fundamental principle introduces a new theoretical framework for the functioning of macroscopic brain waves.

Monday, April 01, 2024

When memories get complex, sleep comes to their rescue

Here I point to a PNAS article by Lutz et al. and a commentary on the work by Schechtman. Here is the Lutz. et al. abstract:


Real-life events usually consist of multiple elements such as a location, people, and objects that become associated during the event. Such associations can differ in their strength, and some elements may be associated only indirectly (e.g., via a third element). Here, we show that sleep compared with nocturnal wakefulness selectively strengthens associations between elements of events that were only weakly encoded and of such that were not encoded together, thus fostering new associations. Importantly, these sleep effects were associated with an improved recall of the complete event after presentation of only a single cue. These findings uncover a fundamental role of sleep in the completion of partial information and are critical for understanding how real-life events are processed during sleep.


Sleep supports the consolidation of episodic memory. It is, however, a matter of ongoing debate how this effect is established, because, so far, it has been demonstrated almost exclusively for simple associations, which lack the complex associative structure of real-life events, typically comprising multiple elements with different association strengths. Because of this associative structure interlinking the individual elements, a partial cue (e.g., a single element) can recover an entire multielement event. This process, referred to as pattern completion, is a fundamental property of episodic memory. Yet, it is currently unknown how sleep affects the associative structure within multielement events and subsequent processes of pattern completion. Here, we investigated the effects of post-encoding sleep, compared with a period of nocturnal wakefulness (followed by a recovery night), on multielement associative structures in healthy humans using a verbal associative learning task including strongly, weakly, and not directly encoded associations. We demonstrate that sleep selectively benefits memory for weakly associated elements as well as for associations that were not directly encoded but not for strongly associated elements within a multielement event structure. Crucially, these effects were accompanied by a beneficial effect of sleep on the ability to recall multiple elements of an event based on a single common cue. In addition, retrieval performance was predicted by sleep spindle activity during post-encoding sleep. Together, these results indicate that sleep plays a fundamental role in shaping associative structures, thereby supporting pattern completion in complex multielement events.