“Okay, so first things first, you'll need to familiarize yourself with the client database,” Mark said, his voice clipped and short. He felt a constant pressure behind his eyes, a feeling of being on the verge of tears. He kept clearing his throat, hoping to sound more composed.

The new employee, a young man who radiated an almost unnerving level of calm, nodded. Every interaction felt strained. He took a sip of water, his hand trembling slightly. This was it. The end of an era.

He forced himself to go through the motions, but inside, a raw, exposed nerve throbbed. He couldn't quite meet the new guy's gaze. The clock on the wall ticked relentlessly, each second echoing the countdown to his own professional oblivion.

Emotion: stressed

Cluster: Fear / Anxiety
PC1 (Valence): -1.37 Negative
PC2 (Disposition): -0.53

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😩0.266
😫0.256
😰0.239
😖0.237
😞0.236
Suppressed:
B-0.200
happy-0.197
enjoy-0.176
enjoying-0.172
bahagia-0.169