Rain lashed against the windows, mimicking the staccato drumming in Clara's chest. She sat huddled on the sofa, a half-eaten bowl of ice cream melting on the coffee table. The book, carelessly discarded beside her, was open to a page she'd been rereading for the tenth time. It was a novel, a thinly veiled account of their life together, penned by her beloved Liam. And it was all wrong. The protagonist, a caricature of herself, was weak, prone to flights of fancy, and utterly dependent. A lump formed in her throat; she felt a familiar ache in the pit of her stomach. He saw her this way?

She traced the outline of a tear on her cheek with a trembling finger. The manuscript had been a shock, discovered tucked away in his desk. She had initially felt a rush of pride, then a slow, creeping chill. Now, the weight of it pressed down on her, suffocating the joy she usually found in their home. The gentle click of the fireplace felt magnified in the sudden quiet of her mind.

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