He stared at the overflowing boxes, his teeth grinding audibly. The sheer volume of their belongings was astounding. Half-eaten bags of chips, broken picture frames, and boxes of faded photographs. His lip curled in disdain.

He began to pace the narrow space, his steps quick and agitated. He wanted to shout, to rail against their lack of consideration. He ran a hand through his hair, his fingers digging into his scalp. He felt a throbbing sensation in his temples.

He snatched up a chipped coffee mug, examining it as if it had personally offended him. He threw it across the room. The mug shattered against the far wall, the sound strangely satisfying. He felt the tension ease slightly, but the underlying unease remained.

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