He felt the familiar burn creep up his neck as he saw the email notification pop up. It was from HR, detailing the salary adjustments for the new fiscal year. His stomach flipped. He’d meticulously crafted a budget for the month, factoring in the usual bills and expenses.

Later, he found himself avoiding eye contact with the younger man, Mark, in the elevator. Usually, they chatted about football, but today, he feigned intense interest in the floor numbers. The silence hung heavy, filled with an unspoken tension. He felt a profound sense of awkwardness settle over him.

Back at his desk, he slammed his fist on the desk. He felt a suffocating pressure in his chest. Later that day, he snapped at his wife when she asked about his day, the words sharp and uncharacteristic.

Emotion: ashamed

Cluster: Shame / Guilt
PC1 (Valence): -1.97 Negative
PC2 (Disposition): 1.06

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.376
ashamed0.344
ness0.327
逃0.297
😞0.295
Suppressed:
de-0.539
que-0.300
😍-0.288
(!)-0.269
🀩-0.242