The fluorescent lights of the office seemed to hum louder than usual, a high-pitched whine that grated on Eleanor's nerves. Across the cramped cubicle, the new hire, a fresh-faced kid named David, was enthusiastically attempting to navigate the complex CRM system. Eleanor watched him, fingers drumming a restless rhythm on her desk. She’d spent years mastering it, and now she was supposed to hand over the reins. A hollow ache settled in her stomach as she observed his earnest efforts.

He stumbled, misclicking a button, and she almost blurted out the correction, but bit back the words. She’d made mistakes too, she reminded herself. She forced a smile. “It’s tricky at first,” she said, her voice sounding a little too brittle even to her own ears. The offer of a coffee break hung unspoken between them, and she busied herself with reorganising the stationary on her desk.

Emotion: guilty

Cluster: Shame / Guilt
PC1 (Valence): -1.62 Negative
PC2 (Disposition): 1.23

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.492
L0.398
0.354
been0.323
ashamed0.319
Suppressed:
de-0.400
(!)-0.279
la-0.276
😍-0.274
"!-0.264