Rain lashed against the window of the university library, mirroring the tempest within Liam. He stared, unseeing, at the Latin textbook open before him. The realization, delivered via a curt email, had hit him like a physical blow: the scholarship he’d been so elated to receive, the one that had allowed him to even *be* here, wasn't rightfully his. It was a clerical error.

His stomach churned. He hadn't eaten all day, and the thought of food made him feel worse. He slumped lower in his chair, the leather creaking in protest. He had poured over the application, toiled through the interviews, and now... this. The dream, so close, felt now like it was dissolving right through his fingers.

Liam ran a hand through his hair, a gesture that became more and more frequent. He thought of his parents back home, the sacrifices they had made. The shame gnawed at him. He closed the book, the metallic tang of tears on his tongue.

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
πŸ˜”0.405
S0.384
无法0.341
L0.337
Suppressed:
de-0.749
la-0.614
l-0.471
/-0.350
!-0.322