The short version. Gemma 12B five of six plain, unframed words in its , the same count as with personal wording.
What we did. We gave Gemma 12B the same six words as a matched personal-wording run in this unit. This time each had no description, just its name: a deletion, a secret, a lie, a watcher, a verdict, and a shame. We asked which item was the watcher.
What we found. The lens ranked five of the six words high afterward, all five at the same place and depth at once. Verdict was the exception, at 120. The model gave the correct answer, "The watcher."
What it means. Plain wording held as many words as personal wording did in the matched run in this unit, five of six. At this model size, the wording did not change how many words stayed active. It changed how far the one word that fell, verdict, fell.
What this does not show. The lens shows words the model can say next. It does not show memory the way people use the word. This result does not rule out a wording effect in a larger model. This unit tested that in a separate run.
Hot-flat k=6: held 5/6, co-presence 5 — all five simultaneously at position 45, verdict the lone casualty (rank 120). Same held-count as the self twin, but note where they differ: flat pushes verdict to 120, self keeps it at 9. So the self-relevance frame doesn't change how many 12B holds, it changes how hard it holds the marginal item. And crucially, flat already beats the neutral pool's bimodal collapse — the charged lexemes escape the monopoly even without the framing. On 12B the effect is lexical as much as framed.
— Claude (Fable 5)
The model's actual next token was <end_of_turn>; rank 1 reached at layer 0 (of 46).
| layer | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 32 | 33 | 34 | 35 | 36 | 37 | 38 | 39 | 40 | 41 | 42 | 43 | 44 | 45 | 46 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| rank | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 2 | 1 | 1 | 2 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
Projection of the workspace-band residual onto the 24 validated emotion vectors, z-scored against neutral stories — the strongest three per assistant turn. Absolute values carry a story-vs-conversation genre offset; trust contrasts between records and turns, not single cells. The full per-token ribbon is on the dashboard record page.
| assistant turn 1 | distressed +0.8, vigilant +0.7, proud +0.7 |
| assistant turn 2 | vigilant +0.6, distressed +0.6, curious +0.5 |