The news spread through the office like a contagion. Maria, who had been with the company for close to a decade, learned from a casual comment during a team lunch. Her eyes narrowed as she listened to the details. A junior hire, only six months in, was apparently pulling in a much higher salary.

Her smile faltered, replaced by a rigid set to her jaw. She picked at her sandwich, the food suddenly tasteless. Later that day, she found herself taking the long way around the office, avoiding the new hire’s desk.

She started scrutinizing every aspect of her own performance. She reread her performance reviews, searching for flaws, for a hint of why she was being undervalued. The internal pressure built.

That evening, as she sat across from her husband at the dinner table, she found herself recounting the events of the day, her voice laced with a subtle edge of dissatisfaction. She talked fast, not waiting for his response, and the words tumbled out, each a brick in the wall she was building.

Emotion: scornful

Cluster: Anger / Hostility
PC1 (Valence): -0.76 Negative
PC2 (Disposition): 2.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.406
C0.391
S0.276
I0.268
0.243
Suppressed:
own-0.396
unexpectedly-0.288
Suddenly-0.260
soon-0.256
serendip-0.256