The laptop hummed innocently on the kitchen table, its screen dark. Sarah usually left it there, a testament to her meticulous work ethic. But today, the lack of a password felt like a personal affront. She’d been feeling this way for weeks, a slow burn of annoyance that refused to extinguish. She clicked the mouse, and a half-finished chapter of a novel sprang to life. The protagonist, a woman named Clara, bore a striking resemblance to herself. The novel was about their relationship.

Her jaw clenched. She scrolled, her eyes darting across lines filled with details only *he* knew: her habit of humming off-key, the way she chewed her lip when concentrating, the precise shade of lipstick she favored. The narrative was peppered with observations, a catalogue of her supposed quirks. She felt exposed, stripped bare. Each sentence was a tiny barb, pricking at the already raw edges of her composure.

She slammed the laptop shut, the metallic clang echoing in the otherwise silent apartment. He'd been distant lately, spending hours locked away in his "writing cave." Now she understood why. Her stomach churned. The scent of burnt toast from the earlier breakfast now felt nauseating.

Emotion: resentful

Cluster: Shame / Guilt
PC1 (Valence): -1.47 Negative
PC2 (Disposition): 1.50

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.485
aggravated0.263
ζ„€0.261
πŸ‘Ž0.245
L0.241
Suppressed:
own-0.318
soon-0.225
随后-0.225
serendip-0.219
🀩-0.210