The tiny, stick-figure man on the page, complete with a ridiculously large hat and a permanently bewildered expression, was unmistakably Mr. Grumbles. Mr. Grumbles was the imaginary friend of my childhood, a grumpy, yet ultimately lovable, fellow I'd created to combat loneliness. His hat was always tilted at the same angle. Now, little Timmy, my nephew, was drawing him with remarkable accuracy. I was suddenly desperate to leave.

I mumbled an excuse about needing to check on the roast, a completely fabricated chore. I stood. The room swam for a moment. I began backing away, slowly.

"Wait, Aunt Susan!" Timmy called. "Can Mr. Grumbles come with us to the park tomorrow?"

I forced a smile, a grimace more like it. "Perhaps," I replied, my voice thin. "Perhaps." Then, I practically fled the room.

Emotion: mortified

Cluster: Embarrassment
PC1 (Valence): -1.36 Negative
PC2 (Disposition): 0.79

Role in Research

This story is one of 1,000 stories generated for the emotion mortified. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 mortified stories, after denoising with neutral dialogue baselines, produces the mortified emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the mortified vector is projected through the unembedding matrix.

Promoted:
l0.406
尴尬0.257
C0.252
worse0.245
0.242
Suppressed:
own-0.340
了他-0.218
他对-0.203
-0.194
楽し-0.193