The office bulletin board, usually a vibrant collage of team photos and announcements, appeared bleak to David. He pointed out the location of the emergency exits to the new hire, Emily. His usual jovial tone was absent, replaced by a monotone. He felt detached, as if watching a play of which he was no longer a part.

He hadn’t realized how much he’d defined himself by this job. He’d always felt a sense of belonging here, a feeling of being needed. Now, the weight of the change pressed down on him with each syllable.

He rubbed his temples, a dull ache throbbing behind his eyes. He wished he had a good story to tell, something exciting to share, but all that came out was a monotonous recitation of rules.

Emotion: sad

Cluster: Sadness / Despair
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -1.50

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.436
😔0.405
S0.384
无法0.341
L0.337
Suppressed:
de-0.749
la-0.614
l-0.471
/-0.350
!-0.322