An interactive model of the periaqueductal grey — the PAG from here on. Four columns of midbrain, wrapped around a channel of fluid. Textbooks catalogue it as defensive: pain modulation, freezing, fleeing. Following the affective neuroscience of Panksepp and Solms, this treats it as generative instead — the place affect is produced, with the cortex scaling what it finds rather than originating it. A contested position, and the one everything here is built on.
Pick a situation below, then make the one move it asks for. Background, figures and the research behind this →
This is a model, not a clinical instrument and not advice. The directions come from published work; the numbers are the author's, and the sections marked Hypothesis are proposals rather than findings. Nothing here diagnoses anything, and nothing here is a reason to change what you are doing about your health or anyone else's.
Hearing a voice is not in itself a disorder — a significant minority of people hear voices without distress or diagnosis, and what predicts difficulty is the relationship, not the presence (Longden et al., 2014). This model takes that seriously: presence and distress are separate readouts, and hosting moves one without moving the other. That is a claim about relationship, not a claim that anyone should manage without support they want or need.
On the cultural material: the distinction Sapolsky draws is load-bearing and easy to lose. Schizotypal traits finding a social role is not the same as schizophrenia being valuable, and the model should not be read as romanticising distress or scarcity of care.
On SSRIs: chronic serotonergic antidepressant use attenuates the subjective effects of psychedelics, and that attenuation can persist for a period after stopping (Becker et al., 2022; Gukasyan et al., 2023). The model shows the direction, not a timeline, and it is not guidance about anyone's medication — those are decisions for a person and their prescriber.
On being told to rethink your pain: this deck does not say that pain is in the mind, or that reframing it is useless. Nociceptive input is its own control here, treated as real and independent. What it does say is that the descending system sets the gain on that input, that cortical reframing scales what is already running rather than reaching the conditions producing it, and that an instruction to reframe often arrives attached to a program, a review and a compliance requirement — a demand landing on a register the reframing does not touch. People whose pain is produced under marginalisation are being addressed at the wrong level, not being unreasonable.
On medication generally: these sliders are mechanisms, not regimens. Serotonergic and catecholaminergic tone appear here as parameters because they have columnar consequences, not because the model has a view about what anyone should take. The evidence for managing blunting — dose change, switching, augmentation — is observational and case-based rather than trial-grade, so the phase behaviour modelled here is a proposal about shape, not a prediction about a person.
The model is also deliberately diagnosis-agnostic. There are no controls for autism, ADHD or any other category, because the columnar evidence does not support category-level claims and because a parameter named after a diagnosis invites people to read a verdict about themselves off a model. Tone, gain, demand and channel are the level this operates at; how that maps onto a life is the person's to say.
On drug combinations: serotonin syndrome is the caveat people expect here, and it is the weaker one. With classic psychedelics the risk is largely theoretical and the documented SSRI interaction is attenuation rather than toxicity — and overstating it pushes people to stop antidepressants abruptly, which carries its own risk. The combinations with real signal are different. Lithium: seizures in 47% of 62 online reports of lithium taken with a classic psychedelic, against none of 34 lamotrigine reports (Nayak et al., 2021) — self-selected reports rather than a trial, but a stark contrast. MAOIs, including the harmala alkaloids in ayahuasca. Serotonin releasers such as MDMA, where serotonin toxicity concern genuinely belongs. None of this is prescribing advice.
On the 5-HT2A section: psychosis-spectrum vulnerability is one of the few points of broad agreement in the caution literature on psychedelics. If that is live for you or someone you work with, treat that section as describing a mechanism, not suggesting a practice.
Inferential personae and figurative images run on two different channels. How the two registers differ →
The named figures below come from documented practice, not from this model — see the glossary. Some arrivals carry propositional content — they say things, address you, take a position. A voice has the arrival characteristics of the image (not composed, not under executive control, costs nothing to produce) and the content characteristics of the persona. Neither register's usual move works on it: you cannot argue it away, because it did not come by that channel, and you cannot leave it merely ambient, because it makes claims. The third operation is hosting — a persona whose object is an arrival rather than a demand. It changes the relationship, not the presence.
Pick a situation. It sets the whole deck at once and then asks you to change one thing yourself — that single move is the demonstration. Watch the readouts on the right and the line under the diagram as you make it. Each one has a short spoken guide that paces you through it — switch it on with the control above if you would like one.
Measured against a quiet waking morning — the state Reset the deck returns you to. Each input is returned to its baseline setting one at a time, the model is re-run, and whichever return moves the column furthest is named. So up and down mean relative to that morning rather than to zero, and what is named is the largest single difference, not the only one. Where second place is close, both are named.
It is here because a column's label is a claim about tissue, not a verdict on a person. The dorsomedial column rising under a slow breath and rising under threat are the same number for entirely different reasons, and the deck should say which.
Gamma in somatosensory and prefrontal cortex tracks subjective pain intensity (Gross et al., 2007; Schulz et al., 2015) — Evidenced, though a 2025 review finds the effect small in some paradigms. Alpha slows and falls when pain is present, and peak alpha frequency has been proposed as a marker of pain sensitivity (Furman et al., 2018) — Evidenced, and actively contested: several 2025 papers fail to replicate it as a predictor. Both fall together under 5-HT2A agonism, which desynchronises broadband — 1–50 Hz posteriorly, 8–100 Hz frontally (Muthukumaraswamy et al., 2013). That is the one place on this deck where nothing is traded for anything; it is simply withdrawn.
What is not claimed: that the PAG has an alpha or a gamma state. Scalp EEG records from the surface and reads cortex; it does not see the midbrain, so there is no columnar band evidence to have. This readout sits in the cortical layer for that reason, and it moves when the cortical layer moves. Treat it as a second face of cortical precision rather than a new mechanism — the composition is the author's, the components are not.
And a refusal worth stating: there are no tones here, and there will not be. Driving a band with sound is the part of this literature that has been most thoroughly colonised by people selling things.
The named figures on this deck are not invented for it. Most are documented style-practices from the author's research with people living with chronic pain under structural marginalisation, where figurative composition was the operative work rather than decoration on it. Participant names are the pseudonyms used in that study. A few are general figures, and one is a team motif — each entry says which.
“I would prefer not to.”
The formula is the whole figure. It is agrammatical, and that is the point: it neither accepts the demand nor refuses it, so there is nothing to argue with and no refusal to escalate against. Deleuze’s reading is that it hollows out a zone of indiscernibility between the preferred and the non-preferred, leaving the demand with no terms it could be restated in. Bartleby is not a rebel; the employer cannot even be properly angry.
That is why this figure sits where it does on the deck: highest suspension of an institutional demand, and almost the cheapest to hold, because it is one move rather than continual cognition. The cheapest is not the weakest. It carries a permanent dorsal tilt it never loses — it reads worse-to-come.
Meeting the demand by running faster. Partial suspension at the highest cost here, and it closes the lateral column while it runs. Degrades precisely when demand is sustained.
Urgency without suspension. Expensive, raises the gate, and barely touches the demand. The figure that costs most and returns least.
Reworks the terms of eligibility instead of contesting them. Substantial suspension at moderate cost, and the only persona routing to lateral and ventrolateral columns at once.
Licensed nonsense and licensed cunning. The fool is cheap and refuses the obligation to already know; the trickster suspends more, costs more, and keeps some dorsolateral arousal going. Included because they are widely held and sit informatively against the documented ones.
Accountable for another’s state. Attends outward and forecloses the carer’s own play; suspends almost none of their own demands; expensive. Unpractised it fails under load while the caring continues, which is the mechanical shape of burnout rather than a failing of the person.
Practitioner and participant compose together in a shared activity rather than through helper-helped relations. The activity is the third term mediating the relation. Routes toward the practitioner’s own lateral column instead of away from it, and costs a third of what The Carer costs.
A persona whose object is an arrival rather than a demand. It lowers the position an arrival occupies without reducing its presence. Barely suspends an institutional demand — a hosted relationship with a voice does little about an assessment regime.
The bush stone-curlew heard across Brisbane at 2am. Taken as teacher rather than symbol: embodied knowledge of survival without dispensation. Routes to lateral and ventrolateral columns.
The same source figure as Red Queening, in the other register. Ambient, costless, and it routes toward the lateral column rather than closing it.
Eight arms, no head office. Shifting roles, relaxed open awareness, attuning to others, no formal hierarchy, play. Routes hardest to the lateral column of any image here — and it holds the arrangement that Composing through a third needs, so that persona costs markedly less to run while this image is going. A motif a team genuinely shares does this; one person privately entertaining it barely does.
Red Queening and the Rain-soaked Red Queen come from one source and behave in opposite directions. Run as a persona, the Red Queen is expensive, raises the threat gate and shuts the lateral column — it is a figure you perform against a demand. Primed as an image, she arrives, costs nothing, and opens the lateral column instead. The figure is not the register. What decides the effect is which channel it runs on — and a figure that exhausts you in one may be a resource in the other.
The burnout literature runs a depletion logic: affective resources flow from practitioner to service user, exhaust, and are replenished by withdrawal — so the remedies are self-care and boundaries, and the failure reads as the practitioner's. The Carer on this deck is that model: it routes away from the practitioner's own lateral column and survives only by being practised at its own expense. Composing through a third is the alternative this research proposes. The work happens in a shared activity, the practitioner is a participant in it rather than a supply line into it, and their lateral column is engaged rather than foreclosed. Run both under the same load and the difference is not how much someone cares. It is the structure of the caring — which makes the collapse a property of the arrangement rather than a deficiency in the person inside it.
These two registers mostly do not touch: an image reaches no demand, and no amount of figurative atmosphere will make an eligibility review go away. There is one interaction, and it runs the other way. An image that already embodies an arrangement makes the persona of that arrangement cheaper to hold, because the surroundings are doing maintenance the executive would otherwise do alone. Le Poulpe and Composing through a third are the clearest pair: no head office, shifting roles, nobody fixed in the helper position. The Cautious Curlew does something similar for Hosting the arrival. This is a cost effect, not a suspension — the domain limit holds. And it scales with sharing specifically, not with private practice — the two are separate sliders, so you can hold everything else and move only the sharing. A motif held between people does roughly three times what one person entertaining it alone does, which is the social function of image sharing stated as a quantity rather than a sentiment.
A figure held representationally — as an idea about a body rather than something a body does — loses this. That is how a powerful figuration can become inert, and it is why these entries describe operations rather than meanings.
Reproduces a published direction or gating relation. The magnitudes are the author's. What is tested: that a figurative construction recruits μ-opioid-dependent descending inhibition; that disclosure does not abolish it; that persona adoption regulates without engaging cognitive control; that practice reduces demand on a domain-general control network; that dreaming is dopaminergically instigated and dissociable from REM, depends on forebrain pathways active while the executive is offline, and that primed imagery enters dream content and alters post-sleep performance.
Extends that work. Not tested: that inferential personae operate as low-cost modulators on the cortico-PAG channel, with persona-specific columnar routing; that practised dreamwork widens access to a columnar route; the routing of dream imagery onto specific PAG columns. The per-persona cost values are the author's construction, set to make the load dissociation legible rather than to estimate anything.
On REBUS specifically: it is a leading model rather than a settled result, and competing accounts hold that psychedelics increase rather than relax precision at some hierarchical levels. The columnar consequences modelled here are not established. The context-dependence shown — that the threat gate, not the agonist, selects which column expresses liberated drive — follows from the gating relations above rather than from direct measurement.
Bouchet, C. A., & Ingram, S. L. (2020). Cannabinoids in the descending pain modulatory circuit: Role in inflammation. Pharmacology & Therapeutics, 209, 107495.
Becker, A. M., Holze, F., Grandinetti, T., Klaiber, A., Toedtli, V. E., Kolaczynska, K. E., … Liechti, M. E. (2022). Acute effects of psilocybin after escitalopram or placebo pretreatment in a randomized, double-blind, placebo-controlled, crossover study in healthy subjects. Clinical Pharmacology & Therapeutics, 111(4), 886–895.
Barrett, L. F. (2017). How emotions are made: The secret life of the brain. Houghton Mifflin Harcourt.
Bobeck, E. N., Chen, Q., Morgan, M. M., & Ingram, S. L. (2022). Cellular and circuit diversity determines the impact of endogenous opioids in the descending pain modulatory pathway. Frontiers in Systems Neuroscience, 16, 963812.
Burnet, P. W. J., Eastwood, S. L., Lacey, K., & Harrison, P. J. (1995). The distribution of 5-HT1A and 5-HT2A receptor mRNA in human brain. Brain Research, 676(1), 157–168.
Cambridge University Press. (2023). Apathy associated with antidepressant drugs: A systematic review. Acta Neuropsychiatrica, 35(6), 317–326.
Carhart-Harris, R. L., & Friston, K. J. (2019). REBUS and the anarchic brain: Toward a unified model of the brain action of psychedelics. Pharmacological Reviews, 71(3), 316–344.
Chein, J. M., & Schneider, W. (2012). The brain's learning and control architecture. Current Directions in Psychological Science, 21(2), 78–84.
Deleuze, G. (1997). Bartleby; or, the formula. In Essays critical and clinical (D. W. Smith & M. A. Greco, Trans., pp. 68–90). University of Minnesota Press. (Original work published 1993)
Dreha-Kulaczewski, S., Konopka, M., Joseph, A. A., Kollmeier, J., Merboldt, K.-D., Ludwig, H.-C., Gärtner, J., & Frahm, J. (2018). Respiration and the watershed of spinal CSF flow in humans. Scientific Reports, 8, 5594.
Dreha-Kulaczewski, S., Joseph, A. A., Merboldt, K.-D., Ludwig, H.-C., Gärtner, J., & Frahm, J. (2015). Inspiration is the major regulator of human CSF flow. Journal of Neuroscience, 35(6), 2485–2491.
Eippert, F., Bingel, U., Schoell, E. D., Yacubian, J., Klinger, R., Lorenz, J., & Büchel, C. (2009). Activation of the opioidergic descending pain control system underlies placebo analgesia. Neuron, 63(4), 533–543.
Furman, A. J., Meeker, T. J., Rietschel, J. C., Yoo, S., Muthulingam, J., Prokhorenko, M., Keaser, M. L., Goodman, R. N., Mazaheri, A., & Seminowicz, D. A. (2018). Cerebral peak alpha frequency predicts individual differences in pain sensitivity. NeuroImage, 167.
Gross, J., Schnitzler, A., Timmermann, L., & Ploner, M. (2007). Gamma oscillations in human primary somatosensory cortex reflect pain perception. PLoS Biology, 5(5), e133.
Muthukumaraswamy, S. D., Carhart-Harris, R. L., Moran, R. J., Brookes, M. J., Williams, T. M., Errtizoe, D., Sessa, B., Papadopoulos, A., Bolstridge, M., Singh, K. D., Feilding, A., Friston, K. J., & Nutt, D. J. (2013). Broadband cortical desynchronization underlies the human psychedelic state. Journal of Neuroscience, 33(38), 15171–15183.
Faull, O. K., Jenkinson, M., Clare, S., & Pattinson, K. T. S. (2015). Functional subdivision of the human periaqueductal grey in respiratory control using 7 tesla fMRI. NeuroImage, 113, 356–364.
Guattari, F. (2011). The machinic unconscious: Essays in schizoanalysis (T. Adkins, Trans.). Semiotext(e). (Original work published 1979)
Gloveli, N., Simonnet, J., Tang, W., Concha-Miranda, M., Maier, E., Dvorzhak, A., Schmitz, D., & Brecht, M. (2023). Play and tickling responses map to the lateral columns of the rat periaqueductal gray. Neuron, 111(19), 3041–3052.
Graeff, F. G., & Zangrossi, H. (2010). The dual role of serotonin in defense and the mode of action of antidepressants on generalized anxiety and panic disorders. Central Nervous System Agents in Medicinal Chemistry, 10(3), 207–217.
Gukasyan, N., Griffiths, R. R., Yaden, D. B., Antoine, D. G., & Nayak, S. M. (2023). Attenuation of psilocybin mushroom effects during and after SSRI/SNRI antidepressant use. Journal of Psychopharmacology, 37(7), 707–716.
Heinricher, M. M., & Ingram, S. L. (2024). The "in's and out's" of descending pain modulation from the rostral ventromedial medulla. Neuroscience Bulletin.
Horowitz, A. H., Esfahany, K., Gálvez, T. V., Maes, P., & Stickgold, R. (2023). Targeted dream incubation at sleep onset increases post-sleep creative performance. Scientific Reports, 13, 7319.
Husen, M. (1996). Det fælles tredje [The common third]. Kultur & Pædagogik, 218–232.
Iwasaki, M., Lefevre, A., Althammer, F., Clauss Creusot, E., Łąpieś, O., Petitjean, H., … Charlet, A. (2023). An analgesic pathway from parvocellular oxytocin neurons to the periaqueductal gray in rats. Nature Communications, 14, 1066.
Kaptchuk, T. J., Friedlander, E., Kelley, J. M., Sanchez, M. N., Kokkotou, E., Singer, J. P., … Lembo, A. J. (2010). Placebos without deception: A randomized controlled trial in irritable bowel syndrome. PLoS ONE, 5(12), e15591.
Johnson, S. B., et al. (2022). Activity in a prefrontal–periaqueductal gray circuit overcomes behavioral and endocrine features of the passive coping stress response. Proceedings of the National Academy of Sciences, 119(44), e2210783119.
Keay, K. A., & Bandler, R. (2001). Parallel circuits mediating distinct emotional coping reactions to different types of stress. Neuroscience & Biobehavioral Reviews, 25(7–8), 669–678.
Krakow, B., Hollifield, M., Johnston, L., Koss, M., Schrader, R., Warner, T. D., … Prince, H. (2001). Imagery rehearsal therapy for chronic nightmares in sexual assault survivors with posttraumatic stress disorder: A randomized controlled trial. JAMA, 286(5), 537–545.
López, F. A., Childress, A., Adeyi, B., Dirks, B., Babcock, T., Scheckner, B., Lasser, R. A., Shepski, J., & Arnold, V. (2017). ADHD symptom rebound and emotional lability with lisdexamfetamine dimesylate in children aged 6 to 12 years. Journal of Attention Disorders, 21(1), 52–61.
Longden, E., Read, J., & Dillon, J. (2014). Emerging perspectives from the hearing voices movement: Implications for research and practice. Schizophrenia Bulletin, 40(Suppl. 4), S285–S294.
Lee, J. J., & Miller, S. E. (2013). A self-care framework for social workers: Building a strong foundation for practice. Families in Society, 94(2), 96–103.
Nayak, S. M., Gukasyan, N., Barrett, F. S., Erowid, E., Erowid, F., & Griffiths, R. R. (2021). Classic psychedelic coadministration with lithium, but not lamotrigine, is associated with seizures: An analysis of online psychedelic experience reports. Pharmacopsychiatry, 54(5), 240–245.
Maslach, C. (1986). Burnout: The cost of caring. Prentice-Hall.
Mehnert, J., Tinnermann, A., Basedau, H., & May, A. (2024). Functional representation of trigeminal nociceptive input in the human periaqueductal gray. Science Advances, 10(12), eadj8213.
Moser, J. S., Dougherty, A., Mattson, W. I., Katz, B., Moran, T. P., Guevarra, D., … Kross, E. (2017). Third-person self-talk facilitates emotion regulation without engaging cognitive control: Converging evidence from ERP and fMRI. Scientific Reports, 7, 4519.
Sapolsky, R. M. (2003). Belief and biology [Lecture]. Stanford University.
Panksepp, J., Herman, B., Conner, R., Bishop, P., & Scott, J. P. (1978). The biology of social attachments: Opiates alleviate separation distress. Biological Psychiatry, 13(5), 607–618.
Ribeiro, S. (2021). The oracle of night: The history and science of dreams (D. Hahn, Trans.). Pantheon.
Schulz, E., May, E. S., Postorino, M., Tiemann, L., Nickel, M. M., Witkovský, V., Schmidt, P., Gross, J., & Ploner, M. (2015). Prefrontal gamma oscillations encode tonic pain in humans. Cerebral Cortex, 25(11), 4407–4414.
Skog, T. D., et al. (2024). A prefrontal→periaqueductal gray pathway differentially engages autonomic, hormonal, and behavioral features of the stress-coping response. Journal of Neuroscience, 44(46), e0844242024.
Solms, M. (2021). The hidden spring: A journey to the source of consciousness. W. W. Norton.
Solms, M. (2000). Dreaming and REM sleep are controlled by different brain mechanisms. Behavioral and Brain Sciences, 23(6), 843–850.
Tinoco Mendoza, F. A., Hughes, T. E. S., Robertson, R. V., Crawford, L. S., Meylakh, N., Macey, P. M., Macefield, V. G., Keay, K. A., & Henderson, L. A. (2023). Detailed organisation of the human midbrain periaqueductal grey revealed using ultra-high field magnetic resonance imaging. NeuroImage, 266, 119828.
Vanderah, T. W., Suenaga, N. M. H., Ossipov, M. H., Malan, T. P., Lai, J., & Porreca, F. (2001). Tonic descending facilitation from the rostral ventromedial medulla mediates opioid-induced abnormal pain and antinociceptive tolerance. Journal of Neuroscience, 21(1), 279–286.
von Wernsdorff, M., Loef, M., Tuschen-Caffier, B., & Schmidt, S. (2021). Effects of open-label placebos in clinical trials: A systematic review and meta-analysis. Scientific Reports, 11, 3855.
White, R. E., Prager, E. O., Schaefer, C., Kross, E., Duckworth, A. L., & Carlson, S. M. (2017). The "Batman effect": Improving perseverance in young children. Child Development, 88(5), 1563–1571.
Zhang, Y.-Y., Liu, Y., Zhao, R., … Chen, X.-J. (2021). Social touch-like tactile stimulation activates a tachykinin 1–oxytocin pathway to promote social interactions. Neuron, 109(24), 4051–4067.