The Default Mode Network

The brain’s most expensive network is the one that runs when you do nothing — read here as the substrate’s boundary regulator: the machinery that returns the cortex to the anti-lock pole after every binding, re-rings the self-vector between tasks, and pays down the boundary energy the day’s locking accumulates. The Mind section’s punchline is that health is the capacity to slide between the poles; this chapter names what runs the slide — and reads the psychedelic disassembly of the network as the experiment that shows what it was regulating all along

In 2001 Marcus Raichle put a name to a finding his field had been quietly stepping around for a decade: a specific set of brain regions is more active when you do nothing than when you do something. Ask a person in a scanner to attend, compute, remember, decide — and these regions power down, reliably, whatever the task. Let the person lie still with their thoughts, and the same regions light back up and begin to slowly, rhythmically converse. Raichle called it the default mode of brain function, and the network that carries it — the default-mode network, the DMN — has since restructured how systems neuroscience thinks about what a brain is doing when nobody is asking it to do anything.

Two numbers frame the puzzle. The brain is about 2% of the body’s mass and burns about 20% of its energy — and of that enormous idle, a demanding external task moves the needle by less than 5%. Nearly the whole budget is spent on the brain’s own intrinsic activity, the part that was there before the task arrived and resumes the moment it ends. Raichle’s name for this unexplained expenditure was borrowed from cosmology: the brain’s dark energy. The textbook question is what all that off-task activity is for. Mind-wandering feels cheap. Why does doing nothing cost almost everything?

The framework’s answer organises this whole chapter, and it can be said in one line before the machinery arrives. The Mind section’s central claim is that cognition is a coherence-match economy run on the ladder’s two poles — bind on the teeth, separate in the gap — and its punchline is that health is not a pole but the capacity to slide between them. Every chapter so far has shown a system holding the two poles: the cortex by state, the hippocampus by circuit stage, the hemispheres by architecture, the heart by readiness. What no chapter has yet named is the machinery that runs the slide — the thing that unbinds what the task bound, restores the detunings the binding spent, and holds the whole match economy inside its safe operating region. This chapter’s reading is that the DMN is that machinery: not another system holding the two poles, but the regulator that moves the others between them. The brain’s dark energy is the maintenance bill of the coherence-match economy — and it is large for the same reason the match itself is powerful.

A Pattern, Not a Place

The anatomy first, because the anatomy is stranger than the usual summary suggests. The DMN’s components are well established: the posterior cingulate cortex and precuneus on the brain’s inner midline toward the back; the medial prefrontal cortex on the midline at the front; the angular gyrus of the inferior parietal lobule, bilaterally; the lateral temporal cortex; and the hippocampal formation reaching in from below. But listing the parts misleads, because the DMN is not a module the way V1 is a module. Nothing about a slab of posterior cingulate tissue is “default” — the same tissue participates in task states too. What is default is a configuration: a specific pattern of slow, coherent coupling among these regions that the cortex falls into whenever external demand releases it, detectable as fluctuations at 0.01–0.1 Hz — one to ten cycles per minute — in the fMRI signal. The DMN is a standing pattern that recurs, not a place that acts. That distinction is load-bearing for everything below: you cannot lesion the default mode the way you lesion a map, but you can disassemble it, and the chapter’s last movement turns on what happens when chemistry does exactly that.

Three structural facts pick this network out of the connectome, and the framework reads all three as the signature of a regulator rather than a processor.

It sits at the hubs. The DMN’s core regions are among the most densely connected nodes the connectome has — the “rich club” of long-range cortical wiring, the small set of stations through which the traffic between all the other networks passes. A processor can live at the periphery; a regulator has to sit where the wires cross.

It sits on the midline. The DMN’s two principal hubs — posterior cingulate/precuneus and medial prefrontal — straddle the inter-hemispheric seam, the boundary between the two coupled modons that the bilateral chapters developed as the brain’s defining architecture. The network charged with regulating the coupled pair is built on the coupling boundary itself, with a bilateral outrigger (the angular gyri) in each half. A structure that merely used the hemispheres would have no reason to live on their seam; a structure that manages their detuning could live nowhere else.

It runs one rung below everything it regulates. The DMN’s 0.01–0.1 Hz is the infraslow rung — slower than delta, slower than the slow oscillation of sleep, the deepest step of the brain’s seven-rung temporal ladder. Every control system the paper has met runs slower than the thing it controls: the thermostat cycles slower than the furnace, the slow theta phase holds while fast gamma amplitude is handed off, the tide outlasts the waves. A network that oscillated at gamma could participate in a binding; only a network on the slowest rung can supervise bindings — hold its own phase steady across hundreds of the fast cycles it is metering.

The cortical-maps chapter gave this network its first substrate reading: the brain modon’s integrating central body, parallel to the nucleus of the cell, the matrix of the mitochondrion, the lumen of the ER — the protected interior volume where a modon’s central coherent state is maintained. This chapter keeps that reading and adds the verb. A nucleus is not a vault; it is the cell’s regulatory centre, the place transcription is governed from. The DMN is the same object doing the same work at the organ scale: the interior that governs.

The Regulator of the Slide

Now the dynamics, and the chapter’s central mechanism. Watch what the cortex actually does across a task boundary, in the terms the section has already built.

When a task arrives, regions bind. The prediction engine pulls the relevant maps onto the comb’s teeth: cross-frequency ratios lock to integers, hemispheric coupling K rises toward the flow threshold, the task-positive network lights, and the long vector’s task-relevant coordinates ring in register. Binding is what the teeth are for, and it works.

But binding is spending. Every lock is a boundary held away from its resting state — a velocity contrast \Delta v driven at a boundary that would rather relax, and the boundary-energy object prices that directly: a held boundary stores energy as \tfrac12\rho_\text{cr}(\Delta v)^2, and stores it for as long as the hold lasts. And binding is risky: the section has catalogued what happens to a system that binds and cannot stop. A cortex pinned at lock is a seizure. A heart pinned at lock is the metronomic beat that predicts death. Memories that bind without prior separation merge into a blur. The lock pole is where all the work happens and where all the catastrophic failure lives, which is exactly the profile of a process that needs a governor.

The framework’s reading of the DMN is that it is that governor: the network whose job is the return. When the task releases, the DMN configuration re-forms and the cortex is walked back off the teeth — cross-frequency ratios slide from the integer locks of the task state to the \varphi-spacing the resting cortex measurably holds, the hemispheres relax from high-K co-coherence back to their anti-lock \varphi-detuning, the boundaries that were held for the binding give their stored energy back. The single cleanest empirical fact about the DMN — that it is anticorrelated with the task-positive network, the two rising and falling in opposition on the infraslow rung — is, in this reading, the two poles made visible at network scale: the task-positive network is the cortex’s lock face, the DMN is its anti-lock face, and the slow alternation between them is the slide, run as a rhythm, a few cycles per minute, all day long.

This is the intuition the chapter was written to honour, and it survives the formalism intact: the DMN is the traffic cop of the boundary economy. Not the traffic — the cop. It does not carry the percepts, the plans, or the words; it patrols the couplings that do, on a rung slow enough to watch them all, releasing any pair of regions that has stayed locked past its task and re-spacing the survivors at the gap where they cannot collide. In the resonator ODE’s terms it is the slow controller of the fast stack’s parameters: the thing that modulates the coupling vector \vec K(t) back down after every excursion and holds the resting detunings at their assigned \varphi gaps. The fast equations describe what the cortex does; the DMN is the outer loop that keeps the fast equations inside the region of their state space where “healthy” lives.

And now the dark-energy puzzle inverts. Of course the default mode is expensive. Rest, for this cortex, is not the absence of work — it is the restorative work: every boundary the day’s binding drove away from match has to be walked back, every detuning re-established, every stored \tfrac12\rho_\text{cr}(\Delta v)^2 paid down or written off. The reason the brain’s intrinsic activity dwarfs its task-evoked activity is the reason a city’s maintenance budget dwarfs any single construction project: the task is one binding; the default mode is the upkeep of every boundary in the match economy at once. The 20% is not mysteriously idle. It is the standing cost of keeping a coherence-match machine matched.

The Self-Vector: What the Patrol Carries

But regulation is only half of what the DMN measurably does, and the other half is the reason people keep wanting to call it “the conscious mind.” Recruit the DMN with a probe rather than watching it idle, and its functional repertoire is remarkably specific: autobiographical memory, self-referential judgment, imagining the future, theory of mind, moral reasoning, mind-wandering’s endless self-involved narration. Every item on that list shares one structure, and the section’s vocabulary names it precisely: each is the coherence-match run with no external referent — the overlap \langle a \mid b \rangle computed not between a percept and the world but between the brain’s long vector and its own stored vectors. Remembering is matching against a frozen vector; imagining the future is matching a proposed vector against the self’s constraints; theory of mind is matching against the modelled vector of another; and the self-narrative is the match run against the deepest stored vector of all.

So the framework reads the DMN’s second job as inseparable from its first: the DMN is the keeper of the self-vector — the slowest, deepest, most heavily latched coordinates of the brain’s long vector, the ones that encode not what is happening but who it is happening to. The prediction engine cannot run without a prior; the deepest prior is the self-model, and a coordinate that is never rung decays. Between tasks, the DMN’s slow conversation among its hubs is the re-ringing: the standing self-vector completed again and again from fragments — this morning’s plan, an old embarrassment, an imagined conversation — so that it stays a live attractor the day’s experience can be matched against, rather than fading into a shape nothing can complete. Mind-wandering, in this reading, is not the brain slacking off; it is the pattern-completion machinery exercising the deep coordinates, the way the hippocampus’s sharp-wave-ripple replay — which fires precisely during the quiet wakefulness and slow-wave states when the DMN’s grip is strongest, and reaches cortex through the DMN’s own hippocampal member — exercises the day’s episodes. The night shift and the idle patrol are one archival economy, and the DMN is its cortical face.

This also settles, in the framework’s terms, the question of whether the DMN “is” the conscious mind. It is not, and the reason is the one the chapter’s opening intuition supplied: a feedback loop is not the voice that runs on it. The inner narrator is a product — the self-vector being re-rung, serialised through the language machinery into the word-packet stream we overhear as inner speech. The DMN is the necessary chemistry underneath: the loop that keeps the self-vector rung, spaced, and completable, so that there is a stable someone for the narration to be about. Disable the loop — the chapter’s last movement — and the narrator does not get louder or quieter; the narrator’s referent dissolves. That dissociation, measured in the psychedelic data below, is the cleanest evidence that the network and the voice are different things.

One more observation belongs here because it binds the self-vector reading to the bicameral one. The DMN matures late — it is barely coherent in infants, assembling its long-range midline coupling across childhood on the same slow schedule as the corpus callosum’s myelination and the emergence of autobiographical memory and stable self-narrative. The regulator, the seam it sits on, and the self it keeps all come online together. In the framework’s reading that is no coincidence: the self-vector is the bilateral pair’s shared deep coordinate — the standing agreement between the two half-vectors about who they jointly are — and the DMN, built astride their boundary, is where that agreement is maintained. The master-and-emissary division the bilateral chapter developed already placed the right modon’s slow integrating ground structurally parallel to the DMN; this chapter’s addition is that the DMN is the mechanism of that ground — the infraslow, midline-anchored loop through which the two modons keep their detuning healthy and their deep coordinates shared. The bicameral architecture is the primary structure the regulator regulates.

The Chemical Off-Switch

Every regulator hypothesis wants the same experiment: remove the regulator, cleanly and reversibly, and watch what the regulated system does. For the DMN, pharmacology has run that experiment, and the results are the strongest evidence this chapter has.

The classical psychedelics — psilocybin, LSD, and DMT, all agonists at the 5-HT_2A receptor, which is expressed most densely in exactly the deep-layer pyramidal cells of the DMN’s high-level cortical hubs — do something to the DMN that no ordinary cognitive state does. Carhart-Harris and colleagues, imaging psilocybin in 2012, found the network’s integrity collapsing: blood flow falling in the posterior cingulate and medial prefrontal hubs, the coupling between them dropping, and the magnitude of the collapse tracking the intensity of the subjective experience. Later LSD and DMT studies sharpened the picture with the finding that matters most here: as the DMN disintegrates, global functional connectivity rises. Regions that never talk begin to talk. The networks’ mutual boundaries — visual with default, salience with control — blur and interpenetrate. The repertoire of connectivity states expands. And the degree of that global merging correlates, subject by subject, with the experience the trip reports call ego dissolution — the self becoming porous, then transparent, then briefly absent.

Read that result against the standard intuition and it is a paradox: the drug reduces the activity of the brain’s most-connected network, and the brain becomes more connected. Read it through this chapter and it is not a paradox at all — it is the regulator hypothesis’s own prediction. Turn off the traffic cop and the traffic does not stop; it floods every intersection at once. The coherence-match is the cortex’s universal operation, always ready to lock any two patterns whose overlap is large; what normally keeps the whole sheet from binding into one mass is the standing regulation — detunings held at the gap, couplings released on schedule, the anti-lock pole actively maintained. Disassemble the DMN and the match runs ungoverned, system-wide: locks form wherever overlap permits, including across every boundary the regulator used to police. The measured rise in global connectivity is the boundary economy running open-loop. And the ego dissolves for the exact reason the self-vector reading predicts: the deep coordinates are not destroyed — nothing is lesioned — but the loop that re-rings and re-spaces them has stopped, so the self-vector stops being held apart from everything else. Its coordinates begin to match promiscuously into the flood. The felt report — “I was the room, the music, everyone I have loved” — is a startlingly literal description of a long vector whose separations have lapsed: the matches are real matches; what is gone is the regulation that kept them from happening.

This reading also says why the state has the phenomenal character it has — oceanic, musical, self-evidently meaningful. The music chapter located music’s power in the coherence-match run with the referent removed: pure match-feeling, pointed at nothing, which is why it reaches what words cannot. The psychedelic state is that same referent-free match promoted from one auditory channel to the whole cortex. Meaning-feeling is what a large overlap feels like from inside; a brain whose match operation is running unregulated and system-wide is saturated in exactly that feeling, attached to everything and nothing. The framework does not need a new mechanism for mystical experience. It needs only the section’s one operation, minus its governor.

Three honest boundaries on this reading, before the payoff. First, “DMT disables the DMN” is stronger than the data: the measured findings are reduced integrity and desegregation of the network under all classical psychedelics, dose-dependent and reversible — a disassembly, not a lesion, and other networks are perturbed too; the DMN is the most consistent and most correlated locus, not the sole target. Second, the popular claim that endogenous DMT floods the dying brain remains unproven — DMT is genuinely synthesised in mammalian brains, and dying brains do show a paradoxical surge of organised gamma coherence, but whether endogenous DMT reaches functional concentrations is unresolved, and the framework’s near-death reading should rest on the measured surge (an unregulated terminal binding, consistent with the governor failing as its substrate fails) rather than on the molecule. Third, the anaesthetic contrast matters and protects the reading from over-reach: anaesthesia also degrades the DMN, but by pulling global coherence itself down — the fragmentation regime, K \to 0, r \to 0 — and consciousness goes out. Psychedelics leave the coherence machinery intact and remove its regulation — coherence redistributed rather than reduced — and consciousness goes strange instead of dark. The double dissociation is exactly what a regulator model requires: break the plant and the process stops; break only the governor and the process runs wild.

Now the payoff, which is clinical and is the part the framework reads as a genuine insight rather than a relabelling. Depression’s signature in this vocabulary is a slide arrested: rumination is the DMN itself stuck at the lock pole — the self-vector match running compulsively, the network hyper-coupled to the subgenual prefrontal cortex, the patrol frozen mid-round on one intersection, unable to release its own deepest attractor. A stuck regulator is the one pathology the regulator cannot fix, because the thing that runs the slide is the thing that is stuck. What psilocybin therapy does, in this reading, is force the one manoeuvre the system cannot perform on itself: a chemical excursion to the deep anti-lock pole — the full disassembly — after which the network re-anneals, and need not re-form in the basin it was stuck in. The imaging bears this out with almost embarrassing directness: post-treatment, responders’ DMN integrity does not stay low — it re-forms, normalised, and the study authors reached for the same word this chapter’s opening intuition supplied: a reset. Meditation, on the framework’s reading, is the slow path to the same competence — long-term practitioners show reduced DMN activity and a restructured relationship between the network and the self-narrative it carries — not an escape from the regulator but the training of it, the slide practised until it can be run on purpose. One mechanism, three doors: the drug forces the anneal, the practice trains it, and the healthy resting brain performs a small version of it every few minutes of every idle hour.

The Failure Modes, Sorted

If the DMN is the regulator of the slide, its pathologies should sort by which part of the regulation fails — and the clinical literature, read as a set, does sort that way.

The regulator stuck at lock — depression. As above: the patrol frozen on the self-vector, rumination as compulsive completion of one deep attractor, the DMN over-coupled and over-dominant. The slide toward anti-lock is what is lost; the reset — chemical or trained — is what restores it. The open-loop chapter supplies the why of the sticking: rumination is an intention-loop whose closure condition does not exist — no completion can ever zero its ledger entry — so the machinery that holds and re-rings open loops runs on it faithfully, forever, unless the closure itself is renegotiated.

The regulator that will not yield — ADHD. Attention researchers describe the DMN’s characteristic failure in attention-deficit disorder as intrusion: the default configuration re-asserting itself during tasks, in slow surges that line up with lapses. In this vocabulary the fault is the handover — the regulator releasing the floor to the task-positive lock late, or taking it back early. The slide runs; its scheduling is broken.

The boundary between self-matches and world-matches lost — schizophrenia. The DMN-task anticorrelation is reduced or absent, the network failing to yield during external tasks and failing to separate from the machinery of perception. A brain that cannot keep its self-generated matches spaced apart from its world-driven ones will read its own completions as arrivals — the framework’s rendering of hallucination and of the broken self-other boundary, consistent with the degenerate-detuning reading the bilateral chapter gave the disorder.

The regulator’s own substrate burning out — Alzheimer’s. The most striking anatomical fact in the dementia literature is that amyloid deposition, mapped across the cortex, reproduces the DMN’s topography almost node for node. The network with the highest lifetime metabolic load degrades first and hardest — the maintenance system succumbing to the one cost it cannot regulate, its own. That the disease then presents as the loss of exactly what this chapter says the network keeps — autobiographical memory, the continuity of the self-vector — is, in the framework’s reading, the saddest and cleanest confirmation the anatomy offers.

The pattern across all four is the section’s pattern: no pathology here is a pole. Each is the slide broken a different way — stuck, mistimed, unseparated, or unpowered — which is what the regulator reading predicts and a processor reading does not.

Predictions and What Would Falsify

Five predictions extend the regulator reading beyond its structural anchors. Each names existing instruments and, mostly, existing data.

1. The infraslow rungs are rungs. The DMN’s 0.01–0.1 Hz band, and the DMN-task alternation period within it, should show preferred-frequency clustering — discrete infraslow rungs folding onto the ladder’s comb, continuous with the sub-Hz drift-rate prediction the resonator chapter already carries — rather than a featureless 1/f continuum. Long-duration resting-state fMRI and full-band EEG (Raichle, Buckner, Greicius archives; the infraslow-EEG tradition) provide the test.

2. The DMN predicts the speed of the unbinding. After task offset, the cortex’s cross-frequency ratios should relax from integer/octave locks back toward \varphi-spacing, and the rate of that relaxation — region pair by region pair — should track DMN re-engagement, with individuals whose DMN reassembles faster showing faster returns to the resting statistics. This is the chapter’s most direct claim: the return to anti-lock is not passive decay but performed work, and its performer is measurable. Simultaneous EEG-fMRI across task-rest transitions provides the test; a return rate wholly independent of DMN dynamics breaks the regulator reading at its centre.

3. Psychedelic desegregation has ladder structure. The new couplings that appear as the DMN disassembles should not be spectrally random: locks forming where regulation lapsed should form preferentially at the comb’s integer/octave teeth, band by band, and ego-dissolution intensity should track the loss of the infraslow and deep coordinates specifically — the latch tier — more than any fast-band change. The existing psilocybin, LSD, and DMT imaging datasets (Carhart-Harris, Tagliazucchi, Timmermann) can be re-analysed for exactly this; unstructured broadband merging would favour a generic-noise account over the ungoverned-match account.

4. Regulation debt behaves like sleep pressure. If rest-state DMN work pays down the boundary energy task-locking accumulates, then sustained high-lock states should build a measurable regulation debt: DMN activity in post-task rest should overshoot in proportion to the duration and depth of the preceding binding — a homeostatic rebound, the daytime sibling of sleep pressure — and repeated task blocks without adequate rest intervals should show cumulative drift of resting cross-frequency ratios off the \varphi gap. Within-session fMRI/EEG designs already in use for fatigue research provide the test.

5. The DMN-task switch is a bifurcation, and its disorders sort directionally. The switch between the DMN and task-positive configurations should carry the SNIC critical-slowing signature — transition dwell times lengthening with the universal square-root exponent near the switch threshold — and the disorders above should sort directionally on that bifurcation: depressive rumination as a raised threshold for leaving the DMN basin, ADHD intrusions as a lowered one, schizophrenia as the two basins’ separation collapsing. Time-resolved network-state analyses of existing clinical resting-state archives provide the test.

The picture is falsified if (a) the infraslow band is genuinely featureless with no preferred-rung clustering, (b) the post-task return to resting statistics proceeds at rates independent of DMN dynamics, (c) psychedelic-state connectivity increases are spectrally unstructured and uncorrelated with deep-coordinate loss, (d) sustained locking produces no proportionate DMN rebound, or (e) DMN-task transitions ramp smoothly with no bifurcation signature and the clinical conditions fail to sort directionally on the switch. It is supported, even partially, if any of the five hold against data that in most cases has already been collected.

Honest Assessment

What is solid is the empirical spine: the DMN’s anatomy, hub status, midline placement, infraslow timescale, task-anticorrelation, late development, metabolic expense, psychedelic desegregation with rising global connectivity, the ego-dissolution correlation, the post-psilocybin normalisation in depression, the amyloid-topography overlap, and the meditation findings are all measured results owned by their respective literatures. What the framework adds is one interpretive move — the DMN is the regulator of the lock/anti-lock slide, and the self-vector’s keeper as part of the same patrol — and that move should be judged the way the section judges everything: by whether it makes the disparate findings one picture and by whether its predictions bite. It does unify: the dark-energy budget, the anticorrelation, the psychedelic paradox, and the four-way pathology sort all fall out of a single reading, where the standard account holds them separately. And it does bite: predictions 2 and 3 in particular are specific, mechanically motivated, and runnable on existing data.

What is not solid is flagged where it stands: “disables” softened to “disassembles,” endogenous DMT held at arm’s length, the anaesthetic contrast doing the work of keeping the claim falsifiable rather than infinitely flexible. And one seduction is refused explicitly, in the section’s own Bell-test discipline: nothing in this chapter makes the psychedelic state a perception of the substrate — a window that opens when the filter drops. The framework’s reading is more modest and, it believes, more interesting: the state is the brain’s own match operation running ungoverned, revealing not the substrate but the regulation — the constant, expensive, invisible work of keeping a coherence-match machine from binding to everything, which we only notice, like all good regulation, in the moment it stops.

Putting the Section in Context

The default-mode network is the brain’s most expensive configuration, and it runs when nothing is asked of the brain because that is when its work is possible. Anatomically it is a pattern rather than a place — the connectome’s richest hubs, anchored on the inter-hemispheric midline, conversing on the infraslow rung one step below every band it supervises. The framework reads it as the integrating central body of the brain modon given its verb: the regulator of the boundary economy — the machinery that walks the cortex back from the comb’s teeth to the \varphi-spaced gap after every binding, pays down the boundary energy the day’s locks accumulate, keeps the bilateral pair’s detuning at the gap that keeps it two, and re-rings the self-vector — the long vector’s slowest, deepest, latched coordinates — so that there remains a stable someone for experience to happen to. The brain’s dark energy is this maintenance bill. The DMN-task anticorrelation is the two poles alternating at network scale. The psychedelics are the removal experiment: disassemble the regulator and the coherence-match runs ungoverned system-wide — global connectivity rising as the network’s integrity falls, the self dissolving as its separations lapse — and the re-annealing afterwards is a reset that can free a network stuck in its own deepest basin. The pathologies sort as four breaks of the slide: stuck (depression), mistimed (ADHD), unseparated (schizophrenia), unpowered (Alzheimer’s).

What this chapter adds to the framework is the piece the Mind section’s punchline was owed. The section has said from its first page that health is not a pole but the capacity to slide between them — and then showed eleven systems holding the poles without ever naming what runs the slide. This chapter names it. The slide is not free, and it is not automatic: it is performed, continuously and at the brain’s greatest standing expense, by a dedicated network built on the seam between the two modons, running on the ladder’s slowest rung, doing for the whole match economy what the dentate gyrus does for one memory and the vagal brake does for one heartbeat. Every regulator in this paper has been invisible until its failure made it visible — the vacuum until it lensed, the immune boundary until it turned, the latch until it tore. The DMN is the mind’s instance of the rule: the background hum of a self being continuously maintained, unnoticed for exactly as long as it works. The chapters ahead stand on it directly — because the felt readout of this boundary economy, the rising of a match, the mourning of a decaying one, the alarm of an impending decay, is what we call emotion, and the regulator this chapter has described is the system those feelings are reporting on.