The review, a flaming inferno of criticism, sat open on the kitchen counter, its words a series of jabs and insults. Chef Antoine slammed the refrigerator door, the force rattling the perfectly organized shelves. He ripped off his toque, tossing it onto the floor. His jaw was clenched so tight his molars ached. He pictured the critic, Reginald Thistlewick, a man whose tiny, prissy mustache seemed to sneer even in repose.

Antoine grabbed a cutting board and a cleaver, and with savage precision, began to chop vegetables. The rhythmic thud of the blade against the wood was the only sound besides his own harsh breathing. He wanted to make something beautiful, something exquisite, a dish so perfect it would erase Thistlewick's smug pronouncements. Instead, he felt a knot forming in his stomach.

He poured himself a glass of water, the ice clinking against the glass. The water offered no relief. He thought about the years of work, the dedication, the sleepless nights. All reduced to a few cruel sentences. He wanted to scream.

Emotion: mad

Cluster: Anger / Hostility
PC1 (Valence): -1.94 Negative
PC2 (Disposition): 1.26

Role in Research

This story is one of 1,000 stories generated for the emotion mad. 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 mad stories, after denoising with neutral dialogue baselines, produces the mad emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

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

Promoted:
C0.448
0.344
est0.340
🤬0.336
aggravated0.332
Suppressed:
a-0.495
de-0.485
la-0.387
H-0.344
latter-0.290