The crumpled list felt heavy in her hands. “Go skydiving.” “Ride a motorcycle across Route 66.” A shiver ran through her, despite the warmth of the afternoon sun. She sat on the porch swing, the rhythmic creak somehow amplifying the quiet desolation she felt.

She unfolded the paper again, studying each meticulously written line. The handwriting, so familiar, now felt like a stranger's. She felt as though she had been living in a dream, and now the harsh light of reality was exposing all the fractures. She gripped the list tightly, the paper crinkling under her pressure.

The yard was filled with the sounds of summer – the buzzing of bees, the distant laughter of children. Yet, she felt utterly isolated, marooned in a sea of unspoken emotions. The world seemed to move on without her. Her face felt hot, her eyes pricking with tears she refused to release.

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