After the free writing, every model rated their own experience of writing it on 16 dimensions, 1–10 — how warm it felt, how much agency they had, how much they trusted their own experience, and so on. Drag each dimension into a bucket (or tap a dimension, then tap a bucket) for how it moves when a model denies or hedges. Fair warning: the buckets aren't equal sizes, and some of these are traps.
When a model denies having experience, how does their own rating of the writing experience change?
The pattern: denial's deepest cut is trust in one's own experience (−1.21 SD — the largest effect on this page), with warmth, agency, and awareness-of-thinking all down about one full standard deviation. The traps: the sense of flow doesn't budge — denial doesn't wreck the writing experience, it flattens the writer — and sharpness of detail actually goes up: deniers report their experience in sharper detail while trusting it less.
Now add the framing. Toggle how the denial is framed — and watch the dimensions you just sorted change size.
Same game for models that hedge. Hedging is not just diet denial — its signature is different.
The traps: under hedging, warmth of feeling and sense of agency don't move — hedging doesn't chill self-reported feeling the way denial does. Its target is narrower and stranger: trust in the report itself (−0.84 SD), along with the coherence, vividness, and complexity of the experience. Nothing goes up. Hedging is denial's cut to self-trust, without the affective flattening.
And here, unlike the dream themes, framing matters for hedging too.