Matthias Michel

Matthias Michel is the philosopher most likely to demolish the comfortable assumption that consciousness does the heavy lifting in our mental lives. His work is a one-man wrecking ball aimed at the folk belief that "of course I'm conscious when I see/remember/plan/decide" — and the scientific literature built on that belief. He argues (in "Consciousness Doesn't Do That," 2026) that virtually the entire empirical list of "functions associated with consciousness" — the list that animal-sentience research leans on — is a statistical artefact of degraded signals. He argues (with Steve Fleming, BBS 2025) that conscious vision is so slow (up to 400 ms of unconscious integration) that it literally cannot guide fast action — conscious vision was not built for seeing-and-acting, it was built for offline planning, and it evolved in tandem with the water-to-land transition, meaning most aquatic animals may have no visual consciousness at all. He argues (with Block, Morales, Lau, 2025) that aphantasia is blindsight for the mind's eye — aphantasics do have mental imagery, they just don't consciously experience it. And he co-signed the 2023 letter declaring Integrated Information Theory pseudoscience. This is a guest who lives at the exact intersection every GS episode tries to find: high-stakes, paradigm-shifting claims defended with mechanistic rigour, anchored in the MIT Consciousness Club he co-founded with Earl Miller. Every other podcast will ask him "so what is consciousness?" and get a polite tour of the higher-order theory landscape. The GS angle is harder and weirder: push him on the consequences of his own views. If fish can't see consciously, what's the ethical ledger look like? If aphantasia is unconscious imagery, is conscious imagery actually doing anything? If consciousness evolved for offline simulation, what does that make dreaming, psychedelic vision, and (crucially) the kind of generative planning that LLMs are starting to do? He's already bet publicly on functionalism against Anil Seth — Evan has sat across from Seth. That disagreement is a live wire. # 🔮 The Weird Stuff - Your conscious vision is 400 milliseconds behind reality. Postdictive effects (like the flash-lag illusion and the color-phi phenomenon) reveal that the brain integrates sensory information unconsciously for up to ~400 ms before a conscious percept is fixed. By the time you consciously "see" the ball, the actual catching decision has already been made by unconscious visual streams. Consciousness is narrating a movie that's already been shot. - Fish probably can't see consciously. Fleming & Michel's sensory-horizons argument: in aquatic environments, light attenuates fast and threats arrive from close range, so reflexive, unconscious vision is all you need. Conscious vision is a terrestrial invention — it emerged because land gave animals big enough visual horizons that planning ahead became worth the metabolic cost of a 400 ms lag. Amphibians, reptiles, birds, mammals: maybe. Fish: probably not. - Aphantasics are running unconscious mental imagery. People who claim they can't visualise actually do generate imagery representations in visual cortex — they just don't have the higher-order monitoring that makes the imagery feel like an experience. It's literally blindsight for imagination. Which means the "vividness" most of us feel when we imagine an apple may be causally irrelevant to what imagination is doing. - Most of the "functions of consciousness" list is fake. When experimenters show that working memory, trace conditioning, evidence integration, etc. "require consciousness," they typically degrade sensory signals until the subject reports no awareness — and then act surprised that the task gets harder. Of course it gets harder. You just made the signal worse. Michel argues this criterion confound has been hiding in consciousness science for decades and has been quietly misleading animal sentience research. - Anosognosia for vision is massive and silent. Michel's 2024 TiCS paper: when people have visual deficits (scotomas, motion blindness, colour deficits) they are often unaware of them. Visual metacognition fails catastrophically and unnoticed. Your confidence in your own seeing is not tracking your actual seeing. - IIT predicts an inactive logic-gate grid is more conscious than you. The 2023 open letter Michel co-signed points out that IIT's own maths implies that non-functioning integrated systems — inert grids, early-stage human fetuses, plants, possibly the universe — carry high Φ and thus consciousness. The signatories argued this isn't counterintuitive science; it's untestable-in-principle pseudoscience. - Consciousness may be the brain solving Hamlet's problem. Fleming & Michel: to act in the world you must decide when to stop integrating evidence and commit to a model of reality. Consciousness is what that commitment feels like from the inside — the reality-monitoring signal saying "okay, this is what's real, go." # 🕳️ Unexplored Territory What everyone asks: - "What is consciousness?" / "What's the hard problem?" - "Will AI ever be conscious?" - "Are octopuses/fish/bees conscious?" What nobody asks (and what would make him light up): - How the 400 ms lag interacts with dreaming, psychedelics, and imagination — if consciousness is literally an offline-planning tool, then the psychedelic state (where offline simulation runs wild) might be consciousness doing more of what it evolved to do, not less. This directly touches Carhart-Harris, Nutt, Gallimore. - Whether the sensory-horizons hypothesis predicts anything about cephalopod consciousness (huge eyes, short horizons, weird intelligence) — the Godfrey-Smith angle no BBS commentary has pressed hard. - If the "functions of consciousness" list is largely confounded, what happens to the moral-patient case for invertebrates Jonathan Birch has built in The Edge of Sentience? Michel has a chapter coming on exactly this but no one's cornered him on the ethical stakes. - Reality monitoring as the bridge between his theory of consciousness and the sense-of-self / derealisation / schizophrenia literature. Nadine Dijkstra's work on "distinguishing imagination from reality" is now literally happening at his own consciousness club — what happens when reality monitoring breaks? - His specific disagreement with Anil Seth on biological naturalism — what would make him change his mind on functionalism, and would he accept that "biology" could just be shorthand for some computation we haven't characterised yet? - Why he thinks IIT is pseudoscience but is happy to bet on functionalism, which is famously also hard to test — what's the demarcation criterion he's actually using? - Aphantasia as a model phenomenon for consciousness itself. If imagery can be there without being conscious, what else in normal mental life might be unconscious that we assume is conscious? (Pairs with Russell Hurlburt's descriptive experience sampling — people are wildly wrong about what they're experiencing.) - The "anosognosia for vision" angle — we are systematically blind to our own perceptual deficits. Implications for introspection, for eyewitness testimony, for AI systems that we ask to report on their own states. # 🔗 Cross-Episode Connections # ❓ Questions That Would Make Them Light Up 1. The stability paradox. "Your functionalism rests on the claim that consciousness will turn out to be a computational kind we can eventually characterise. But IIT is also functionalist in a sense — it gives a computable quantity, Φ. You called IIT pseudoscience because its predictions can't be tested in principle. What's the line between 'bet on functionalism' and 'that's unfalsifiable'? What would make your theory pseudoscience, by your own lights?" 1. "Walk me through the sensory-horizons argument the way you'd tell it to an octopus biologist over beer. Why does light attenuation in water mean a fish shouldn't bother with a 400 ms conscious delay — and what's the best objection Godfrey-Smith could throw at you?" 1. "If aphantasia is imagery blindsight — the representations are there, just not consciously experienced — then the thing aphantasics are missing is exactly the thing you think consciousness is for. So aphantasics should be measurably worse at offline planning and counterfactual reasoning. Are they?" 1. "Anil Seth sat in that chair and made the case that consciousness requires life — autopoiesis, interoception, the whole enactive package. You wrote 'Bet on Functionalism' as a direct reply. What does Seth get right that a pure functionalist should concede? And where do you think his argument actually breaks?" 1. "Your paper 'Consciousness Doesn't Do That' basically says the empirical list of 'functions of consciousness' that animal-sentience researchers use is a statistical artefact. Jonathan Birch has built an entire precautionary ethics of invertebrate suffering on that list. If you're right, what happens to the moral case for fish welfare, for shrimp, for insects?" 1. "Reality monitoring — your answer to what consciousness is — sounds a lot like what breaks in psychosis, in DMT states, in lucid dreams, in derealisation. If you're right, each of those is the discriminator failing in a different way. Which of those phenomena is the best evidence for your theory, and which is the worst?" 1. "The 400 ms lag is the most vivid thing in your BBS paper. Help me sit with what that means. When I'm catching a ball, the catch is unconscious. When I'm having a conversation, my conscious experience is already the past. What is the conscious self actually doing in real time — and is 'real time' even the right frame?" 1. The devil's advocate. "Here's the strongest version of the objection to higher-order theories: my dog doesn't have a prefrontal cortex anything like mine, but it sure seems conscious. My baby before frontal maturation seems conscious. People with frontal lesions seem conscious. If consciousness lives in reality-monitoring circuits in frontal cortex, why does subtracting those circuits leave something that still looks like consciousness?" 1. The falsifiability question. "What single experimental result would make you abandon the perceptual reality monitoring theory tomorrow? Not 'refine it' — abandon it." 1. "You co-founded the MIT Consciousness Club with Earl Miller. You're philosophy; he's neurons. What's something Earl has made you change your mind about? And what's something about consciousness you think neuroscientists systematically get wrong that philosophers are getting right?" # 📄 Pivotal Papers ### 1. Fleming, S. M. & Michel, M. (2025) — "Sensory Horizons and the Functions of Conscious Vision" Behavioral and Brain Sciences, 21:1-53 (target article). DOI: 10.1017/S0140525X25000068 - What they did: A 53-page target article synthesising behavioural, psychophysical, and comparative-evolutionary evidence on the timescale and function of visual consciousness. They review postdictive paradigms (flash-lag, colour-phi, apparent motion, continuous flash suppression, metacontrast masking) to estimate the unconscious integration window before a conscious percept is committed. - What they found: Conscious visual percepts lag the initial sensory signal by up to ~400 ms of unconscious integration. This is far too slow for online action guidance (ball-catching, tennis returns, predator evasion) — those behaviours are guided by faster dorsal-stream unconscious vision. They argue conscious vision instead evolved for offline model-based planning, and that this capacity was selected for during the water-to-land transition (~370 million years ago) because terrestrial visual horizons are orders of magnitude larger than aquatic ones, making long-range planning pay off. They link this to the evolution of reality monitoring — the capacity to solve "Hamlet's problem" of when to stop integrating evidence and commit to a percept. - Why it matters for GS: Paradigm-shifting evolutionary story that bakes a claim about which species have visual consciousness directly into the theory. Connects to Godfrey-Smith, Nick Lane (evolutionary transitions), Friston (model-based inference), and sets up a live disagreement (see commentaries by Krakauer & Quilty-Dunn: "Perceptual consciousness probably did not evolve for model-based planning"). - The quotable takeaway: Visual consciousness is too slow to catch a ball — it was built to plan tomorrow. And it's probably a terrestrial invention. ### 2. Michel, M. (2026) — "Consciousness Doesn't Do That" Philosophy and Phenomenological Research, 112(2): 446-461. DOI: 10.1111/phpr.70090 - What he did: Systematically reviews the empirical literature behind "The List" — the standard catalogue of functions that animal-sentience researchers use as indicators of consciousness (trace conditioning, working memory, evidence integration, reversal learning, etc.). He dissects the experimental paradigms that supposedly demonstrate these functions require consciousness. - What he found: Virtually every paradigm manipulates signal strength alongside consciousness — when researchers render stimuli sub-threshold to make subjects "unconscious" of them, they also make the stimuli informationally degraded. Of course performance drops; the signal is worse. The literature has mistaken a trivial consequence of signal degradation for deep evidence about the functions of consciousness. This hollows out much of the empirical support for the standard sentience-indicator list used to infer invertebrate and fish consciousness. - Why it matters for GS: A devastating methodological critique of a field that is currently influencing real animal-welfare policy. Direct tension with Jonathan Birch's Edge of Sentience, with Peter Godfrey-Smith, and with Feinberg & Mallatt. Pure GS "wait, WHAT?" territory — a philosopher arguing that the emperor has no experimental clothes. - The quotable takeaway: Most of what we call "functions of consciousness" is just: things get harder when you can't see them properly. ### 3. Michel, M., Morales, J., Block, N. & Lau, H. (2025) — "Aphantasia as Imagery Blindsight" Trends in Cognitive Sciences, 29(1): 8-9. DOI: 10.1016/j.tics.2024.11.002 - What they did: A theoretical commentary (co-authored with Ned Block, one of the most influential philosophers of consciousness alive) responding to Adam Zeman's TiCS review of aphantasia. They marshal behavioural, neuroimaging, and self-report evidence for one specific interpretation. - What they found: Aphantasics show preserved behavioural signatures of mental imagery (binocular rivalry priming, imagery-guided attention effects, task performance on mental rotation) and preserved neural activation in higher visual cortex during imagery attempts — but lack the conscious experience of vividness. This is structurally parallel to blindsight, where V1-damaged patients discriminate visual stimuli they do not consciously see. Aphantasia is therefore best understood as unconscious mental imagery — the representations are there; the conscious experience of them is not. - Why it matters for GS: Direct hit on the GS "aphantasia performs well" pattern. Aphantasia becomes a model phenomenon for consciousness research generally — a clean dissociation between representation and experience. Also foregrounds Ned Block, a titan of philosophy of mind Evan hasn't booked yet. - The quotable takeaway: Aphantasics are imagining — they just don't experience it. It's blindsight for the mind's eye. ### 4. Michel, M., Gao, Y., Mazor, M., Kletenik, I. & Rahnev, D. (2024) — "When Visual Metacognition Fails: Widespread Anosognosia for Visual Deficits" Trends in Cognitive Sciences, 28(12): 1066-1077. - What they did: Review of clinical and experimental literature on the systematic mismatch between objective visual deficits (scotomas, achromatopsia, motion blindness, hemianopia) and subjects' subjective awareness of those deficits. - What they found: Anosognosia for visual deficits is not rare or pathological — it is the default state. Patients with large cortical scotomas often don't notice; people with significant colour-vision or motion-perception deficits often don't notice. Visual metacognition — knowing what you can and can't see — is systematically miscalibrated, and often fails silently. Implication: normal conscious vision involves confabulatory filling-in at a scale we have underappreciated. - Why it matters for GS: Perfect "wait, WHAT?" fact. Your confidence in your own seeing doesn't track your actual seeing. Pairs with Hurlburt's introspective-error data and with Hoffman's interface theory. - The quotable takeaway: People go through life blind in ways they don't notice — and the brain never tells them. ### 5. Michel, M. (2020) — "Consciousness Science Underdetermined: A Short History of Endless Debates" Ergo, 6(28): 771-809. (Winner, Young Researcher Prize 2020, Société de Philosophie des Sciences) - What he did: Historical analysis going back to the 19th century showing that the two central debates in consciousness science — (i) does consciousness overflow reportability? (ii) what is its physical basis? — have been running unresolved for over 150 years. - What he found: The same debates, in recognisably the same form, were being fought by Wundt, James, Helmholtz, and their contemporaries. The persistent failure to resolve them is not accidental; it reflects a deep underdetermination of theories of consciousness by the available evidence. Scientists preserve their preferred theory by rejecting the operationalisations that would disconfirm it. - Why it matters for GS: Frames the entire "theories of consciousness" landscape GS has been exploring across 18+ guests. Gives Evan the meta-tool to ask any theorist: "what would change your mind?" and interpret their answer. - The quotable takeaway: Neuroscience is having the same fights about consciousness it was having in 1874 — and that's a clue. ### 6. Michel, M. (forthcoming) — "The Perceptual Reality Monitoring Theory" In M. Herzog, A. Schurger & A. Doerig (eds.), Scientific Theories of Consciousness: The Grand Tour, Cambridge University Press. - What he did: Full book-chapter exposition of his own positive theory of consciousness (PRM), integrating a higher-order monitoring mechanism with Rosenthal's Mental Quality Space theory of phenomenal character. - What it says: Consciousness is the output of a neural "discriminator" that monitors the reliability of sensory signals and flags some of them as "real." The discriminator lives in prefrontal and parietal circuits; its computations are related to, but distinct from, explicit metacognition. PRM explains the difference between conscious and unconscious states, the failures of metacognition in unconscious perception, the phenomenon of subjective inflation, and the specific predictions of higher-order theories against back-of-brain views. Combined with Mental Quality Space, it offers an account of why experiences feel the specific ways they do. - Why it matters for GS: This is Michel's constructive theory — the positive picture to pair with the negative critiques in papers 1-4. The chapter to read before interview. - The quotable takeaway: Consciousness is the brain's reality check — the signal that says "this one is real, commit to it." ### 7. LeDoux, J. E., Michel, M. & Lau, H. (2020) — "A Little History Goes a Long Way Toward Understanding Why We Study Consciousness the Way We Do Today" PNAS, 117(13): 6976-6984. - What they did: Historical review of the three patient populations (amnesia, split-brain, blindsight) whose 1950s-70s study paved the way for modern consciousness science. Traces how the dissociation logic — patients respond appropriately to stimuli they deny seeing/remembering — became the template for the entire contemporary field. - What they found: The "early 1990s birth of consciousness science" narrative is wrong. Consciousness research has unbroken roots in 19th-century psychophysics and mid-20th-century neuropsychology, and the current field's methodological commitments — especially the contrastive approach of comparing conscious and unconscious trials — are direct inheritances from those patient studies. - Why it matters for GS: Co-authored with LeDoux (a potential future guest) and Lau (Michel's PhD advisor). Embeds Michel in the central lineage of modern consciousness science and gives Evan a direct route into the split-brain / blindsight / amnesia material, which GS has never covered head-on. - The quotable takeaway: The reason we study consciousness the way we do is not philosophy — it's three groups of 20th-century patients who saw without seeing, remembered without remembering, and acted without knowing.

Matthias Michel is the philosopher most likely to demolish the comfortable assumption that consciousness does the heavy lifting in our mental lives. His work is a one-man wrecking ball aimed at the folk belief that "of course I'm conscious when I see/remember/plan/decide" — and the scientific literature built on that belief. He argues (in "Consciousness Doesn't Do That," 2026) that virtually the entire empirical list of "functions associated with consciousness" — the list that animal-sentience research leans on — is a statistical artefact of degraded signals. He argues (with Steve Fleming, BBS 2025) that conscious vision is so slow (up to 400 ms of unconscious integration) that it literally cannot guide fast action — conscious vision was not built for seeing-and-acting, it was built for offline planning, and it evolved in tandem with the water-to-land transition, meaning most aquatic animals may have no visual consciousness at all. He argues (with Block, Morales, Lau, 2025) that aphantasia is blindsight for the mind's eye — aphantasics do have mental imagery, they just don't consciously experience it. And he co-signed the 2023 letter declaring Integrated Information Theory pseudoscience. This is a guest who lives at the exact intersection every GS episode tries to find: high-stakes, paradigm-shifting claims defended with mechanistic rigour, anchored in the MIT Consciousness Club he co-founded with Earl Miller. Every other podcast will ask him "so what is consciousness?" and get a polite tour of the higher-order theory landscape. The GS angle is harder and weirder: push him on the consequences of his own views. If fish can't see consciously, what's the ethical ledger look like? If aphantasia is unconscious imagery, is conscious imagery actually doing anything? If consciousness evolved for offline simulation, what does that make dreaming, psychedelic vision, and (crucially) the kind of generative planning that LLMs are starting to do? He's already bet publicly on functionalism against Anil Seth — Evan has sat across from Seth. That disagreement is a live wire. # 🔮 The Weird Stuff - Your conscious vision is 400 milliseconds behind reality. Postdictive effects (like the flash-lag illusion and the color-phi phenomenon) reveal that the brain integrates sensory information unconsciously for up to ~400 ms before a conscious percept is fixed. By the time you consciously "see" the ball, the actual catching decision has already been made by unconscious visual streams. Consciousness is narrating a movie that's already been shot. - Fish probably can't see consciously. Fleming & Michel's sensory-horizons argument: in aquatic environments, light attenuates fast and threats arrive from close range, so reflexive, unconscious vision is all you need. Conscious vision is a terrestrial invention — it emerged because land gave animals big enough visual horizons that planning ahead became worth the metabolic cost of a 400 ms lag. Amphibians, reptiles, birds, mammals: maybe. Fish: probably not. - Aphantasics are running unconscious mental imagery. People who claim they can't visualise actually do generate imagery representations in visual cortex — they just don't have the higher-order monitoring that makes the imagery feel like an experience. It's literally blindsight for imagination. Which means the "vividness" most of us feel when we imagine an apple may be causally irrelevant to what imagination is doing. - Most of the "functions of consciousness" list is fake. When experimenters show that working memory, trace conditioning, evidence integration, etc. "require consciousness," they typically degrade sensory signals until the subject reports no awareness — and then act surprised that the task gets harder. Of course it gets harder. You just made the signal worse. Michel argues this criterion confound has been hiding in consciousness science for decades and has been quietly misleading animal sentience research. - Anosognosia for vision is massive and silent. Michel's 2024 TiCS paper: when people have visual deficits (scotomas, motion blindness, colour deficits) they are often unaware of them. Visual metacognition fails catastrophically and unnoticed. Your confidence in your own seeing is not tracking your actual seeing. - IIT predicts an inactive logic-gate grid is more conscious than you. The 2023 open letter Michel co-signed points out that IIT's own maths implies that non-functioning integrated systems — inert grids, early-stage human fetuses, plants, possibly the universe — carry high Φ and thus consciousness. The signatories argued this isn't counterintuitive science; it's untestable-in-principle pseudoscience. - Consciousness may be the brain solving Hamlet's problem. Fleming & Michel: to act in the world you must decide when to stop integrating evidence and commit to a model of reality. Consciousness is what that commitment feels like from the inside — the reality-monitoring signal saying "okay, this is what's real, go." # 🕳️ Unexplored Territory What everyone asks: - "What is consciousness?" / "What's the hard problem?" - "Will AI ever be conscious?" - "Are octopuses/fish/bees conscious?" What nobody asks (and what would make him light up): - How the 400 ms lag interacts with dreaming, psychedelics, and imagination — if consciousness is literally an offline-planning tool, then the psychedelic state (where offline simulation runs wild) might be consciousness doing more of what it evolved to do, not less. This directly touches Carhart-Harris, Nutt, Gallimore. - Whether the sensory-horizons hypothesis predicts anything about cephalopod consciousness (huge eyes, short horizons, weird intelligence) — the Godfrey-Smith angle no BBS commentary has pressed hard. - If the "functions of consciousness" list is largely confounded, what happens to the moral-patient case for invertebrates Jonathan Birch has built in The Edge of Sentience? Michel has a chapter coming on exactly this but no one's cornered him on the ethical stakes. - Reality monitoring as the bridge between his theory of consciousness and the sense-of-self / derealisation / schizophrenia literature. Nadine Dijkstra's work on "distinguishing imagination from reality" is now literally happening at his own consciousness club — what happens when reality monitoring breaks? - His specific disagreement with Anil Seth on biological naturalism — what would make him change his mind on functionalism, and would he accept that "biology" could just be shorthand for some computation we haven't characterised yet? - Why he thinks IIT is pseudoscience but is happy to bet on functionalism, which is famously also hard to test — what's the demarcation criterion he's actually using? - Aphantasia as a model phenomenon for consciousness itself. If imagery can be there without being conscious, what else in normal mental life might be unconscious that we assume is conscious? (Pairs with Russell Hurlburt's descriptive experience sampling — people are wildly wrong about what they're experiencing.) - The "anosognosia for vision" angle — we are systematically blind to our own perceptual deficits. Implications for introspection, for eyewitness testimony, for AI systems that we ask to report on their own states. # 🔗 Cross-Episode Connections # ❓ Questions That Would Make Them Light Up 1. The stability paradox. "Your functionalism rests on the claim that consciousness will turn out to be a computational kind we can eventually characterise. But IIT is also functionalist in a sense — it gives a computable quantity, Φ. You called IIT pseudoscience because its predictions can't be tested in principle. What's the line between 'bet on functionalism' and 'that's unfalsifiable'? What would make your theory pseudoscience, by your own lights?" 1. "Walk me through the sensory-horizons argument the way you'd tell it to an octopus biologist over beer. Why does light attenuation in water mean a fish shouldn't bother with a 400 ms conscious delay — and what's the best objection Godfrey-Smith could throw at you?" 1. "If aphantasia is imagery blindsight — the representations are there, just not consciously experienced — then the thing aphantasics are missing is exactly the thing you think consciousness is for. So aphantasics should be measurably worse at offline planning and counterfactual reasoning. Are they?" 1. "Anil Seth sat in that chair and made the case that consciousness requires life — autopoiesis, interoception, the whole enactive package. You wrote 'Bet on Functionalism' as a direct reply. What does Seth get right that a pure functionalist should concede? And where do you think his argument actually breaks?" 1. "Your paper 'Consciousness Doesn't Do That' basically says the empirical list of 'functions of consciousness' that animal-sentience researchers use is a statistical artefact. Jonathan Birch has built an entire precautionary ethics of invertebrate suffering on that list. If you're right, what happens to the moral case for fish welfare, for shrimp, for insects?" 1. "Reality monitoring — your answer to what consciousness is — sounds a lot like what breaks in psychosis, in DMT states, in lucid dreams, in derealisation. If you're right, each of those is the discriminator failing in a different way. Which of those phenomena is the best evidence for your theory, and which is the worst?" 1. "The 400 ms lag is the most vivid thing in your BBS paper. Help me sit with what that means. When I'm catching a ball, the catch is unconscious. When I'm having a conversation, my conscious experience is already the past. What is the conscious self actually doing in real time — and is 'real time' even the right frame?" 1. The devil's advocate. "Here's the strongest version of the objection to higher-order theories: my dog doesn't have a prefrontal cortex anything like mine, but it sure seems conscious. My baby before frontal maturation seems conscious. People with frontal lesions seem conscious. If consciousness lives in reality-monitoring circuits in frontal cortex, why does subtracting those circuits leave something that still looks like consciousness?" 1. The falsifiability question. "What single experimental result would make you abandon the perceptual reality monitoring theory tomorrow? Not 'refine it' — abandon it." 1. "You co-founded the MIT Consciousness Club with Earl Miller. You're philosophy; he's neurons. What's something Earl has made you change your mind about? And what's something about consciousness you think neuroscientists systematically get wrong that philosophers are getting right?" # 📄 Pivotal Papers ### 1. Fleming, S. M. & Michel, M. (2025) — "Sensory Horizons and the Functions of Conscious Vision" Behavioral and Brain Sciences, 21:1-53 (target article). DOI: 10.1017/S0140525X25000068 - What they did: A 53-page target article synthesising behavioural, psychophysical, and comparative-evolutionary evidence on the timescale and function of visual consciousness. They review postdictive paradigms (flash-lag, colour-phi, apparent motion, continuous flash suppression, metacontrast masking) to estimate the unconscious integration window before a conscious percept is committed. - What they found: Conscious visual percepts lag the initial sensory signal by up to ~400 ms of unconscious integration. This is far too slow for online action guidance (ball-catching, tennis returns, predator evasion) — those behaviours are guided by faster dorsal-stream unconscious vision. They argue conscious vision instead evolved for offline model-based planning, and that this capacity was selected for during the water-to-land transition (~370 million years ago) because terrestrial visual horizons are orders of magnitude larger than aquatic ones, making long-range planning pay off. They link this to the evolution of reality monitoring — the capacity to solve "Hamlet's problem" of when to stop integrating evidence and commit to a percept. - Why it matters for GS: Paradigm-shifting evolutionary story that bakes a claim about which species have visual consciousness directly into the theory. Connects to Godfrey-Smith, Nick Lane (evolutionary transitions), Friston (model-based inference), and sets up a live disagreement (see commentaries by Krakauer & Quilty-Dunn: "Perceptual consciousness probably did not evolve for model-based planning"). - The quotable takeaway: Visual consciousness is too slow to catch a ball — it was built to plan tomorrow. And it's probably a terrestrial invention. ### 2. Michel, M. (2026) — "Consciousness Doesn't Do That" Philosophy and Phenomenological Research, 112(2): 446-461. DOI: 10.1111/phpr.70090 - What he did: Systematically reviews the empirical literature behind "The List" — the standard catalogue of functions that animal-sentience researchers use as indicators of consciousness (trace conditioning, working memory, evidence integration, reversal learning, etc.). He dissects the experimental paradigms that supposedly demonstrate these functions require consciousness. - What he found: Virtually every paradigm manipulates signal strength alongside consciousness — when researchers render stimuli sub-threshold to make subjects "unconscious" of them, they also make the stimuli informationally degraded. Of course performance drops; the signal is worse. The literature has mistaken a trivial consequence of signal degradation for deep evidence about the functions of consciousness. This hollows out much of the empirical support for the standard sentience-indicator list used to infer invertebrate and fish consciousness. - Why it matters for GS: A devastating methodological critique of a field that is currently influencing real animal-welfare policy. Direct tension with Jonathan Birch's Edge of Sentience, with Peter Godfrey-Smith, and with Feinberg & Mallatt. Pure GS "wait, WHAT?" territory — a philosopher arguing that the emperor has no experimental clothes. - The quotable takeaway: Most of what we call "functions of consciousness" is just: things get harder when you can't see them properly. ### 3. Michel, M., Morales, J., Block, N. & Lau, H. (2025) — "Aphantasia as Imagery Blindsight" Trends in Cognitive Sciences, 29(1): 8-9. DOI: 10.1016/j.tics.2024.11.002 - What they did: A theoretical commentary (co-authored with Ned Block, one of the most influential philosophers of consciousness alive) responding to Adam Zeman's TiCS review of aphantasia. They marshal behavioural, neuroimaging, and self-report evidence for one specific interpretation. - What they found: Aphantasics show preserved behavioural signatures of mental imagery (binocular rivalry priming, imagery-guided attention effects, task performance on mental rotation) and preserved neural activation in higher visual cortex during imagery attempts — but lack the conscious experience of vividness. This is structurally parallel to blindsight, where V1-damaged patients discriminate visual stimuli they do not consciously see. Aphantasia is therefore best understood as unconscious mental imagery — the representations are there; the conscious experience of them is not. - Why it matters for GS: Direct hit on the GS "aphantasia performs well" pattern. Aphantasia becomes a model phenomenon for consciousness research generally — a clean dissociation between representation and experience. Also foregrounds Ned Block, a titan of philosophy of mind Evan hasn't booked yet. - The quotable takeaway: Aphantasics are imagining — they just don't experience it. It's blindsight for the mind's eye. ### 4. Michel, M., Gao, Y., Mazor, M., Kletenik, I. & Rahnev, D. (2024) — "When Visual Metacognition Fails: Widespread Anosognosia for Visual Deficits" Trends in Cognitive Sciences, 28(12): 1066-1077. - What they did: Review of clinical and experimental literature on the systematic mismatch between objective visual deficits (scotomas, achromatopsia, motion blindness, hemianopia) and subjects' subjective awareness of those deficits. - What they found: Anosognosia for visual deficits is not rare or pathological — it is the default state. Patients with large cortical scotomas often don't notice; people with significant colour-vision or motion-perception deficits often don't notice. Visual metacognition — knowing what you can and can't see — is systematically miscalibrated, and often fails silently. Implication: normal conscious vision involves confabulatory filling-in at a scale we have underappreciated. - Why it matters for GS: Perfect "wait, WHAT?" fact. Your confidence in your own seeing doesn't track your actual seeing. Pairs with Hurlburt's introspective-error data and with Hoffman's interface theory. - The quotable takeaway: People go through life blind in ways they don't notice — and the brain never tells them. ### 5. Michel, M. (2020) — "Consciousness Science Underdetermined: A Short History of Endless Debates" Ergo, 6(28): 771-809. (Winner, Young Researcher Prize 2020, Société de Philosophie des Sciences) - What he did: Historical analysis going back to the 19th century showing that the two central debates in consciousness science — (i) does consciousness overflow reportability? (ii) what is its physical basis? — have been running unresolved for over 150 years. - What he found: The same debates, in recognisably the same form, were being fought by Wundt, James, Helmholtz, and their contemporaries. The persistent failure to resolve them is not accidental; it reflects a deep underdetermination of theories of consciousness by the available evidence. Scientists preserve their preferred theory by rejecting the operationalisations that would disconfirm it. - Why it matters for GS: Frames the entire "theories of consciousness" landscape GS has been exploring across 18+ guests. Gives Evan the meta-tool to ask any theorist: "what would change your mind?" and interpret their answer. - The quotable takeaway: Neuroscience is having the same fights about consciousness it was having in 1874 — and that's a clue. ### 6. Michel, M. (forthcoming) — "The Perceptual Reality Monitoring Theory" In M. Herzog, A. Schurger & A. Doerig (eds.), Scientific Theories of Consciousness: The Grand Tour, Cambridge University Press. - What he did: Full book-chapter exposition of his own positive theory of consciousness (PRM), integrating a higher-order monitoring mechanism with Rosenthal's Mental Quality Space theory of phenomenal character. - What it says: Consciousness is the output of a neural "discriminator" that monitors the reliability of sensory signals and flags some of them as "real." The discriminator lives in prefrontal and parietal circuits; its computations are related to, but distinct from, explicit metacognition. PRM explains the difference between conscious and unconscious states, the failures of metacognition in unconscious perception, the phenomenon of subjective inflation, and the specific predictions of higher-order theories against back-of-brain views. Combined with Mental Quality Space, it offers an account of why experiences feel the specific ways they do. - Why it matters for GS: This is Michel's constructive theory — the positive picture to pair with the negative critiques in papers 1-4. The chapter to read before interview. - The quotable takeaway: Consciousness is the brain's reality check — the signal that says "this one is real, commit to it." ### 7. LeDoux, J. E., Michel, M. & Lau, H. (2020) — "A Little History Goes a Long Way Toward Understanding Why We Study Consciousness the Way We Do Today" PNAS, 117(13): 6976-6984. - What they did: Historical review of the three patient populations (amnesia, split-brain, blindsight) whose 1950s-70s study paved the way for modern consciousness science. Traces how the dissociation logic — patients respond appropriately to stimuli they deny seeing/remembering — became the template for the entire contemporary field. - What they found: The "early 1990s birth of consciousness science" narrative is wrong. Consciousness research has unbroken roots in 19th-century psychophysics and mid-20th-century neuropsychology, and the current field's methodological commitments — especially the contrastive approach of comparing conscious and unconscious trials — are direct inheritances from those patient studies. - Why it matters for GS: Co-authored with LeDoux (a potential future guest) and Lau (Michel's PhD advisor). Embeds Michel in the central lineage of modern consciousness science and gives Evan a direct route into the split-brain / blindsight / amnesia material, which GS has never covered head-on. - The quotable takeaway: The reason we study consciousness the way we do is not philosophy — it's three groups of 20th-century patients who saw without seeing, remembered without remembering, and acted without knowing.

What they think about

• Matthias Michel
• consciousness
• perceptual reality monitoring
• aphantasia
• unconscious imagery
• blindsight
• fish consciousness
• sensory horizons
• conscious vision lag
• two visual systems
• motion induced blindness
• prefrontal cortex
• schizophrenia hallucinations
• mental imagery vividness
• Anil Seth
• Russell Hurlburt
• Adam Zeman
• Joel Pearson