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).

Rain hammered against the windows of the antique shop. Charles ran a finger along the dusty display case, tracing the outlines of a chipped porcelain doll. The doll seemed to mirror his own current state; fractured and incomplete. His wife, Margaret, had always been meticulous with their finances, a detail he admired. Until today.

He had unearthed the account ledger hidden in the back of her desk. Enough to buy a small cottage, twice over. He felt a dull throb behind his eyes, a weight pressing down on his chest.

“Something catch your eye, dear?” Margaret’s voice, cheerful as ever, sliced through the quiet of the shop. He turned, forcing a smile that felt brittle. “Just admiring the antiques, love.”

Emotion: sullen

Cluster: Sadness / Despair
PC1 (Valence): -1.51 Negative
PC2 (Disposition): -0.22

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.404
😞0.282
’0.254
:(0.233
无法0.215
Suppressed:
de-0.539
!-0.276
la-0.263
🤩-0.225
que-0.216