“Well, this is just great,” Mark muttered under his breath. He was so close to finishing his crossword puzzle when the dog, a clumsy St. Bernard, lumbered into his yard. The puzzle lay discarded on the table, ink smeared across the page where he'd been working on a particularly difficult clue.

He slammed the newspaper down. He ran his hand through his already messy hair, leaving it even worse. He felt a throbbing sensation in his head.

He walked outside, slowly, his hands in his pockets. He stared at the dog, who was now attempting to eat his prized garden gnome. “Oh, come on,” he grumbled. “Really?”

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