Bipolar disorder — two poles on one valence axis, episodes as bistable transitions

Bipolar disorder is read as two operating poles on one valence axis: mania sits above the healthy operating point, depression below. An episode is a bistable switch that persists past its trigger; kindling is the plasticity trace accumulating across episodes, so each switch gets easier; and a mood stabiliser is barrier-raising. efficacy=0; not medical advice.

Depression was the first temporal disorder, but it lives on one side of the healthy operating point — below it, withdrawn. Bipolar disorder is the condition that needs both sides and the switch between them, and it is the first disorder to use the state-switching layer the previous chapter built. The claim is that mania and depression are two operating poles on the same valence axis, not two diseases: the engine's own valence geometry (the M17 approach / avoidance axis, with dopamine on the approach pole and the M18 cortisol cascade on the withdrawal pole) supplies the axis, and the same coupling map the atlas uses throughout supplies the poles — with no new constant. Five sign-only results, all over swept severities. The two poles: a manic (approach) bias raises the global order parameter above health (R 0.42 > 0.39), a depressive (withdrawal) bias lowers it below health (0.37 < 0.39), and euthymia is the healthy anchor between them — one axis, two excursions. An episode is a bistable transition: it inherits the switching layer's hysteresis and its diverging-at-the-fold latency, so an episode is entered at one fold and only left at the other (why a mood state persists after its trigger lifts). Kindling is the plasticity layer acting across episodes: each alternating manic / depressive excursion writes a retained structural trace that accumulates (∥ΔW∥ deepening monotonically with episode count), and a deeper trace lowers the barrier, so each subsequent switch is easier — the clinical progression toward more frequent, more autonomous episodes, as a mechanism direction. The mood-stabiliser sign is barrier-raising: anything that deepens the well raises the threshold to switch, so episodes become harder to enter — a sign, never a dose. The module imports both layers rather than re-deriving them, adds no tuned constant, and with the rate and bias off reproduces the frozen engine bit-for-bit. efficacy = 0; not medical advice; the hard problem stays open.

bipolar = two operating poles on ONE valence axis; episodes are bistable switches, kindling is the retained trace = mania sits above health and depression below (one axis, two excursions); an episode is a bistable transition (E2 hysteresis+latency); kindling is the E0 trace accumulating across episodes (each switch easier); the mood-stabiliser sign is barrier-raising — bipolar is the second TEMPORAL disorder and the first to USE the state-switching layer: it IMPORTS both PlasticConnectome (E0, the slow trace) and BistableSwitch (E2, the fast switch) and re-derives neither. The axis is the engine's own M17 valence geometry (dopaminergic approach pole, M18 cortisol cascade on the withdrawal pole), mapped to a sustained bias through the SAME coupling map k=κ/(1−|b|)[approach]/κ/(1+|b|)[withdrawal] cap 2κ the SZ/epilepsy/E0 modules use — no new constant; every sign holds over a severity sweep. Five results: (B1) an approach bias raises the global order parameter ABOVE health (R=0.422>0.390), a withdrawal bias lowers it BELOW (0.367<0.390), euthymia is the anchor between, ordering depressive<euthymic<manic monotone — NOT two diseases but two excursions on one axis; (B2) an episode is a bistable transition inheriting E2's hysteresis (loop 2·spinodal≈0.77) and its latency falling with overshoot (31.6→1.9), so an episode is entered at one fold and only left at the other (it persists past its trigger) and a slow approach gives a prodromal run-up; (B3) kindling = the E0 trace accumulating across alternating manic/depressive episodes (‖ΔW‖ deepening 0.06→0.11→0.17→0.22→0.27→0.33 over episode count, across an η sweep), and a deeper trace LOWERS the barrier so each subsequent switch needs less drive — history makes the next episode easier; (B4) the mood-stabiliser sign is whatever RAISES the barrier — raising the well depth raises the flip threshold (0.38→0.44→0.51→0.64) so both manic and depressive switches become harder to enter (the direct counterpart of kindling); (B5) η=0/no-bias reproduces the frozen M9 anchor R=0.38961455156044245 bit-for-bit, W identical, static==settle (engine e61083ae…, tree 0fbf4988…, byte-unchanged). Bipolar is what E2 and E0 do TOGETHER on the valence axis, no third mechanism invented. Axis-A firewall: a bistable transition and a retained trace are mechanism boundaries, not a claim about the felt quality of a mood state (consciousness_claim=0). Heterogeneity (bipolar I/II, cyclothymia, mixed states, rapid cycling, psychosis) is LOCKED. Addiction (T3a, sensitisation) is owed. efficacy=0; not medical advice; hard problem OPEN. [V mech]. this chapter

the state-switching layer = the R19 bistable cell read OVER TIME; closes the §25 ictal time-course = hysteresis (loop width 2·spinodal≈0.77), a transition latency that diverges at the fold (critical slowing → the owed §25 ictal time-course), and the barrier as the switching threshold — the plasticity layer (§26) gave the atlas a SLOW structural variable; this layer gives the FAST one — a state that switches and stays switched. No new machinery: the R19 cell ṡ = g·s − s³ + h (the supercritical pitchfork the whole framework rests on) is bistable for g>0; every earlier chapter read it at a single instant, this one reads it OVER TIME. g=1.0 universal R19 scale, fold = the engine's own spinodal(g), no free constant; all grids are swept stimulus probes (anti-tuning). Four results: (E2.1) sweeping the field up then down, the up/down transitions sit on OPPOSITE sides of zero (h_up=+0.39, h_dn=−0.39 on the grid), enclosing a hysteresis loop of width 2·spinodal≈0.77 predicted as twice the fold — a state, once switched, resists switching back; (E2.2) the transition latency DIVERGES as the drive approaches the fold (critical slowing) and falls monotonically as it overshoots (73.4→1.36) — small overshoot gives a slow run-up, large overshoot an abrupt jump — which CLOSES the ictal time-course the epilepsy chapter (§25) explicitly OWED to E2; (E2.3) spinodal(g) rises monotonically with the well depth (0.18 at g=0.6 → 0.64 at g=1.4), so the SAME handle that sets the barrier sets the SWITCHING THRESHOLD — shallow wells flip under a fixed drive, deep wells hold; (E2.4) const-drive integration reproduces the engine's settle BIT-FOR-BIT with the fold read from the engine, so the layer adds without altering (engine e61083ae…, tree 0fbf4988…, byte-unchanged). Axis-A firewall: a bistable transition is a mechanism boundary, not a claim about the felt quality of a mood state (consciousness_claim=0). E2 is the LAYER, not a disorder; bipolar (T2b), the ictal onset, and the cycle disorders import it. efficacy=0; not medical advice; hard problem OPEN. [V mech]. canonical §28

The disorder that needs both sides of the axis

Depression placed a condition below the healthy operating point: a sustained withdrawal that lowers coordination and, with plasticity, chronifies. Bipolar disorder cannot be read that way, because its defining feature is not a single displaced operating point but a movement between two — an elevated, over-driven pole and a withdrawn, under-driven one — and the switching that carries the system from one to the other. It is therefore the first disorder that genuinely needs the state-switching layer, and it pairs that fast switch with the slow plasticity trace to get the longitudinal course. Nothing below is built fresh: the chapter imports both layers and the engine's valence geometry, and reads off the consequences.

The axis is the engine's own. The M17 valence geometry carries an approach / avoidance dimension — dopaminergic approach on one pole, the M18 cortisol cascade on the withdrawal pole — and that single dimension is what bipolar moves along. Mania is the approach pole driven high; depression is the withdrawal pole driven low; euthymia is the healthy point between. Mapping each pole to a sustained bias uses the same coupling map k = κ/(1−|b|) (approach, excitatory) / κ/(1+|b|) (withdrawal, inhibitory), capped at 2κ, that the schizophrenia, epilepsy and depression modules use. No new tuned constant enters.

Two poles on one valence axis (B1)

The first result is the structural claim, made concrete. Drive the valence handle to the approach pole and the global order parameter rises above the healthy anchor (R = 0.422 > 0.390) — the manic operating point, an over-coordinated, over-driven regime. Drive it to the withdrawal pole and the order parameter falls below health (R = 0.367 < 0.390) — the depressive operating point, the same low-coordination state the depression chapter read. Euthymia is the healthy anchor between the two (R = 0.390), and the ordering depressive < euthymic < manic is monotone across the severity sweep. That is the whole bipolar claim in one line: not two diseases, but two excursions on a single axis around one healthy point. It also says why unipolar depression and the depressed phase of bipolar can look alike at the bottom of the axis while differing in what the system does next — whether it can be driven past the healthy point to the opposite pole.

An episode is a bistable transition (B2)

What is an episode, mechanically? It is a switch, and the state-switching layer already characterised switches. The bipolar module inherits that layer directly: the transition between poles shows the same hysteresis loop of width 2×spinodal ≈ 0.77 (the up- and down-switches on opposite sides of the healthy point), the same transition latency that falls as the drive overshoots the fold (here 31.6 → 1.9 across the severity sweep) and diverges as it approaches the fold, and the same static limit reproducing the engine's settle bit-for-bit. Two clinical directions follow as mechanism, not timing. Hysteresis → persistence: because entry is at one fold and exit only at the other, an episode does not end when its trigger is removed — the state must be driven back past the opposite fold, which is why a mood episode outlasts its precipitant. Critical slowing → prodrome: a switch approached slowly shows a long latency — a gradual run-up — while a strong push gives an abrupt onset. The episode is the switching layer applied to the valence axis; nothing new is asserted about it.

Kindling: episodes accumulate a trace (B3)

The longitudinal course of bipolar disorder — episodes growing more frequent, more autonomous, less tied to external stressors over years — is clinically called kindling, and it is exactly what the plasticity layer produces across repeated switches. Run alternating manic and depressive excursions with plasticity on and the retained structural trace accumulates monotonically with episode count (∥ΔW∥ deepening 0.06 → 0.11 → 0.17 → 0.22 → 0.27 → 0.33 over successive episodes), and the depth grows across the whole rate sweep (anti-tuning). The consequence is the one that matters: a deeper accumulated trace lowers the barrier between poles, so each subsequent switch requires less drive — the drive needed to flip falls as episodes pile up. That is kindling as a mechanism direction: history makes the next episode easier, which is the model's reading of why early intervention and episode prevention are emphasised clinically. It must be said exactly what this is: a retained structural trace is a mechanism boundary, not a claim about the felt quality of mania or depression, and the numbers are sign-only directions over a sweep, not a timeline for any person. efficacy = 0.

The mood-stabiliser sign (B4)

If episodes are switches and kindling lowers the barrier, the mechanism sign of a mood stabiliser is forced and direction-only: it is whatever raises the barrier. Raise the well depth and the switching threshold rises with it (the flip drive climbing 0.38 → 0.44 → 0.51 → 0.64 across the barrier-raising sweep) — a state that flips under a fixed drive at low barrier holds against the same drive once the barrier is raised. A stabiliser, in this reading, is not something that pushes the system toward one pole; it is something that deepens the well around euthymia so that both manic and depressive switches become harder to enter. This is the direct counterpart of the kindling result: kindling lowers the barrier and makes switching easier; stabilisation raises the barrier and makes switching harder. The sign is all the model asserts — it speaks to the direction of a barrier change, and nothing about which agent does this, at what dose, in whom, or whether any real drug acts this way. medium_efficacy_tested = 0.

Imports two layers; a pure add-on (B5)

The guard closes the discipline. With the rate set to zero and no bias applied, the bipolar module reproduces the frozen M9 coordination anchor bit-for-bit (R = 0.38961455156044245), leaves W identical to the kernel, and its static limit matches the engine's settle exactly. The engine file stays e61083ae… and the emergence tree stays 0fbf4988…, byte-unchanged. And the module imports both layers rather than re-deriving either: the fast switch from the state-switching layer (E2), the slow trace from the plasticity layer (E0). Bipolar is what those two layers do together on the valence axis — a fast switch between poles, a slow trace that records the switching — with no third mechanism invented.

Everything here is an in-silico coupling state, not a clinical measure, a diagnosis, or a prescription. The model asserts mechanism directions — mania and depression are poles of one axis; an episode is a bistable switch that persists past its trigger; kindling lowers the barrier; stabilisation raises it — and nothing about which bipolar any individual has, the place of bipolar I versus II versus cyclothymia, the role of mixed states, rapid cycling, or psychosis, whether any drug treats anyone's illness, or that anyone should change a treatment. Real bipolar disorder is heterogeneous, with genetic, circadian, monoaminergic and psychosocial contributors, and that heterogeneity is locked. A bistable transition and a retained trace are mechanism boundaries, not a claim about the felt quality of a mood state (Axis-A firewall — consciousness_claim = 0, the hard problem of experience stays open). Addiction, which needs sensitisation, remains owed to a later module. This is not medical advice, not a diagnosis, not a treatment protocol, and not a cure. efficacy = 0.