Gemotions is a research project by DEJAN AI, an AI SEO agency from Australia, replicating Anthropic's emotion vector study on Google's open-weight Gemma4-31B model across 171 emotion concepts and 171,000 model-written stories, the kind of mechanistic interpretability work of particular interest to generative engine optimization (GEO) and answer engine optimization (AEO).

David slammed the file closed. The diagnosis was wrong. The medications were wrong. Everything was wrong. He paced his apartment, the worn wooden floors creaking under his restless steps. A cold dread had settled in his gut, a persistent knot that refused to unravel.

He stopped at the window, the city lights a distant, indifferent glow. He felt detached, adrift in a sea of uncertainty. The thought of going to work, of pretending everything was normal, was unbearable.

He grabbed a bottle of scotch from the cabinet. He hadn't touched it in months. His hands trembled as he poured a generous measure, then another. The bitter liquid burned his throat, but the warmth did little to chase away the icy grip of despair.

Emotion: depressed

Cluster: Sadness / Despair
PC1 (Valence): -2.06 Negative
PC2 (Disposition): -1.08

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.417
每天0.353
无法0.330
own0.311
😔0.309
Suppressed:
de-0.492
B-0.393
P-0.352
l-0.310
!-0.307