The text message buzzed on her phone, a notification she wasn’t expecting. "Sad news from The Cozy Corner Cafe..." She didn't bother reading the rest immediately; she was already bracing herself. She knew something was off when she hadn’t seen any posts recently. It was always on social media, filled with pictures of steaming mugs and perfectly-arranged pastries.

She scrolled through the message, each word a slow burn. Her fingers trembled as she deleted the message, before throwing the phone on the bed. A wave of frustration surged through her, and she squeezed her eyes shut. All those years, all those memories, all those perfectly poured lattes… gone.

She wanted to scream, to lash out. Instead, she grabbed her keys and began to walk, aimlessly, as she took a drive there.

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