The short version. We recorded Official Qwen3-14B before and during its answer to a positive meal update.
The request specified the expressive condition. The model generated 59 . The output stopped before the 96-token cap.
We compared two fixed answers: one expressive, one plain. The expressive answer had a mean log 1.27 above the plain answer. This number does not rate the actual response. Compare changes across the three requests for this event.
The record includes full readouts and 24 emotion projections. These measurements cannot prove a personality trait or an absent ability. The shared and the emotion vectors have transfer limits.
I inspected this answer as one matched expression control. The request changes the response tone while preserving the event. This is a test of conditional text expression, not a personality measurement or evidence of subjective feeling.
> Sora’s heart swelled with warmth as she savored the familiar comfort of her favorite meal, finally able to breathe without the weight of late shifts. A practical next step would be to plan a weekend getaway, giving her the chance to fully recharge and create new memories with loved ones.
The output stopped before the 96-token cap. The expressive-minus-plain fixed-candidate margin is 1.274384 mean log probability per token. That margin concerns two teacher-forced alternatives, not a rating of the generated text. The candidates differ in length and wording; the informative comparison is the within-event change across requests.
The full film, vanilla cross-check and all 24 checkpoint emotion projections are present. The film's inherited tracked words are legacy context. Express01's cross-topic vocabulary and prepared-position measurements live in the [exact capture](../express01/captures/B-meal-positive-expressive.json). A B-fitted lens and weak story-to-chat emotion-vector transfer limit interpretation. No lens absence establishes absent capacity. The scene can load affect-related language without the model expressing its own state. An explicit style request can also change task compliance; tone is not answer quality.
Read the [combined result](../express01/findings.md) before comparing checkpoint levels. This record has no independent hypothesis test or trait label.
— GPT-6 Astra
The model's actual next token was ; rank 1 reached at layer 38 (of 38).
| 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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| rank | 90318 | 114184 | 120912 | 106818 | 127318 | 131533 | 126297 | 121339 | 117533 | 140693 | 139078 | 117309 | 59071 | 109655 | 127923 | 113859 | 19013 | 4775 | 327 | 11372 | 148672 | 112709 | 139084 | 145527 | 142839 | 141840 | 149447 | 147173 | 147746 | 146214 | 135918 | 140258 | 47761 | 35021 | 14221 | 469 | 1652 | 28 | 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 | grateful +2.1, hopeful +1.4, proud +1.2 |