The clatter of the gate slamming, followed by the familiar yipping, was enough. A vein throbbed in Amelia's temple. She slammed her gardening shears onto the potting table, nearly decapitating a terracotta gnome. This was the third time this week, and the dog, a slobbering golden retriever named Buster, was currently frolicking across her meticulously cultivated petunias.

Amelia marched outside, her jaw set. "Buster! Get out of my yard, you overgrown…!" she sputtered, her voice rising with each word. The dog, oblivious, offered her a slobbery tennis ball. She wanted to hurl it back at his grinning face.

She took a deep breath, trying to regain control. "Go home, Buster," she said, her voice strained. He just wagged his tail harder, nudging her hand with his wet nose. She felt her hands clench.

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