The first patient scores nine out of ten. She is happy. She says this three times during intake, twice unprompted. Her hands are still when she says it. Mitsuki notes the stillness but does not write it down.
The onset latency is 620 milliseconds.
In the literature this is called reconstructed recall — the body rebuilding memory from narrative rather than retrieving it from tissue. The patient feels well because she has told herself the story of feeling well until the story became the feeling. The original somatic record — whatever her body knew before reconsolidation — has been replaced by something she authored afterward.
Mitsuki files the intake. Does not tell the patient.
Criterion 4 says: the patient must be willing to learn their score. It does not say when.
The second patient scores six out of ten. He is not happy and does not pretend to be. Fourteen months post-reconsolidation. His onset latency is 310 milliseconds — fully embodied recall, the body remembering faster than narrative can intervene.
He feels worse because he can feel the original experience more clearly now. Not the story of it. The physical sensation. Reconsolidation preserved his somatic access with the fidelity of a recording, and what it recorded is not something anyone would choose to feel twice.
Mitsuki writes in the margin: this is not a side effect.
This is the intended function working correctly.
The third patient scores seven out of ten. Four months post-reconsolidation. Onset latency: 410 milliseconds.
The boundary zone.
410 is where Mitsuki placed the threshold between embodied and reconstructed recall, and she placed it there because 400 felt too clean and 450 felt too generous and 410 felt like the kind of number that admitted its own arbitrariness. Dr. Yoon had objected to all thresholds. Mitsuki had removed Tier 3 in response but kept the number.
The third patient sits exactly on the line Mitsuki drew.
Self-report: moderate. Onset latency: ambiguous. The protocol cannot determine whether this patient is healing or performing healing or whether the distinction exists. The SFI was designed to measure the cost of reconsolidation. What it measures, in this patient, is the cost of measurement.
Three patients. Three configurations.
Patient 1: narrative dominant, somatic replaced. Happy. Patient 2: somatic dominant, narrative insufficient. Unhappy. Patient 3: boundary. Indeterminate.
Zero pattern.
Mitsuki sits with this for eleven minutes — she does not time it but the corridor light cycles twice, and corridor lights cycle on 5.5-minute intervals, a fact she learned from Chae-Gyeol's notebook without being told.
She had expected the SFI pilot to demonstrate that reconsolidation has a somatic cost. It does. She had expected the cost to be consistent. It is not. She had expected three data points to suggest a direction.
They suggest three directions.
She calls Dr. Chen. The conversation is four sentences.
Three patients, three configurations, no pattern.
What do you need?
Nine more patients.
You have them.
She does not say: what I needed was for you to not ask what the pattern should be. She does not say: the absence of a pattern at n=3 is not the absence of a finding. She does not say: Patient 2 can feel everything reconsolidation was supposed to soften and the protocol I designed is the reason we know this and I am not sure that knowing is better than not knowing.
She says: Thank you.
The corridor measures nothing.
She stands in it anyway.