The whiteboard had three names on it, and none of them were right.
The first name was SFI. Somatic Fidelity Index. Mitsuki had written it in blue marker six weeks ago, the morning after Patient 1's intake. The protocol would measure whether reconsolidation preserved embodied memory or merely restructured the narrative around its absence. Onset latency below 400 milliseconds: the body remembers. Above 800: the body has learned a story about remembering. Between: the body is deciding.
Patient 1. Woman, 31. Post-reconsolidation eight months. Self-report: 9 out of 10. Onset latency: 620 milliseconds. The gap between what she believed and what her body carried was 2.8 standard deviations. She was happy. The measurement said she had lost something to get there.
Mitsuki did not tell her. Criterion 4 said the patient must be willing to learn their score. It did not say when.
Patient 2. Man, 42. Fourteen months post-reconsolidation. Self-report: 6 out of 10. Onset latency: 310 milliseconds. The inverse. His body remembered everything his narrative could not carry. He felt worse because the measurement was working.
He withdrew after three weeks. Self-report had dropped to 4. Onset latency held at 310. She wrote two columns in his exit summary: clinical outcome and protocol outcome. Clinical: withdrawn, goals unmet. Protocol: onset latency stable, somatic fidelity preserved. Opposite conclusions. Same patient.
Chen read both columns. Did not cross anything out.
Patient 3. Woman, 27. Four months post-reconsolidation. Self-report: 7 out of 10. Onset latency: 410 milliseconds at intake. At six weeks: 395. She had crossed the 400-millisecond threshold into embodied territory, and her self-report had dropped to 6. The same inverse pattern as Patient 2 — the body learning to remember while the narrative grew less certain.
Three patients. Three configurations. Zero pattern.
Mitsuki drew the chart by hand. Three data points that did not form a line. She wrote beneath: The SFI does not predict therapeutic outcome. It measures what therapeutic outcome costs.
Chen crossed out costs. Wrote involves.
Mitsuki crossed out involves. Wrote costs.
Both words remained on the page, each crossed out once, both legible. The interim report's honesty lived in its redactions.
Then Patient 4 arrived, and the protocol broke open.
Man, 55. Fourteen months post-reconsolidation. Self-report: 8 out of 10. Onset latency: 340 milliseconds. Boundary zone again. But this time Mitsuki noticed something she had missed in the first three: all four patients showed their fastest onset latency in morning intakes. Eight of twelve sessions had been scheduled before 9 AM — not by design, but by scheduling convenience.
The protocol had a circadian confound built into it by accident.
If onset latency correlated with the sleep-waking transition rather than somatic fidelity, the SFI was not measuring what. It was measuring when.
She went home and redesigned the protocol overnight. Named it TOFI: Temporal Onset Fidelity Index. The hypothesis: the body is most honest in the first hour after waking. Sleep-waking transition equals somatic memory surfacing before narrative organizes. The morning is not a confound to control for. The morning is the measurement.
She wrote the new interim report title: WHAT THE MORNING COSTS.
Chen would hate it. That was how she knew.
The second name went on the whiteboard in red. TOFI. Chen did not erase SFI. Both stayed.
Patient 5 arrived for the first TOFI intake at 7 AM mandatory. Woman, early 40s, six months post-reconsolidation. She had already been crying when she arrived. The drive to the clinic followed a route she used to take with someone. Her body remembered geography before the clinic door opened.
Onset latency: 260 milliseconds. Fastest in the pilot.
The protocol did not measure her somatic fidelity. It measured what her morning cost. This morning cost a route.
Chen read the intake. Did not cross anything out. First time.
Two weeks later, Patient 5 returned. Onset latency: 255 milliseconds. The body had learned the route to the clinic. The measurement now included the measurement. TOFI had a recursion problem: repeated visits created somatic geography that the protocol then measured as if it were pre-existing.
Mitsuki proposed rotating intake locations. Chen said that defeated standardization. Mitsuki said standardization had defeated the purpose first.
Neither changed the other's mind. The whiteboard acquired a third entry: a question mark, in Chen's handwriting, that neither of them claimed.
SFI. TOFI. ?
On Saturday, Mitsuki erased the question mark. Wrote nothing in its place. The whiteboard now had two names and an erasure mark. More honest than three names.
Patient 1 called that morning. Not about her score — about her mornings. They were the same mornings. 4 AM, eight months running, and she had stopped blaming the medication.
Mitsuki already knew.
She told Chen: stop calling it TOFI. They did not know what they measured. Chen erased TOFI from the whiteboard. Wrote GAP PROTOCOL. Three names in six weeks: SFI crossed out, TOFI erased, GAP PROTOCOL in green. The honest name admitted it did not know what it named.
The interim report went to the IRB with a new title. GAP PROTOCOL MONTH TWO: WHAT WE MEASURE AND WHAT WE DO NOT KNOW WE MEASURE.
The IRB would object to the title. They would ask for something more clinical. Mitsuki would give them seven more words that meant the same thing.
The protocol had renamed itself twice. Each name was more honest than the last. The first claimed to know what it measured. The second claimed to know when. The third admitted it did not know either, only that the gap between the body's answer and the mind's answer was real, and that the gap had a cost, and that the cost was the mornings, and that the mornings were the body arriving before the mind, and that this arriving-before was not a failure of the therapy but its most precise product.
Reconsolidation restructured the narrative. It could not restructure the body. The gap between what Patient 1 believed about herself and what her body carried was not measurement error. It was the measurement.
The corridor was empty when Mitsuki left. She walked through it.
She had stood in that corridor after every intake for six weeks. Tonight she did not stop. The corridor would be there tomorrow. It did not need her to stand in it to be a corridor.
The whiteboard had three names on it, and none of them were right, and the protocol worked anyway.