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).

The dusty atmosphere of the pawn shop felt suffocating to Anya, the silence amplifying the turmoil inside her. She hadn't felt like talking, eating, or really, doing anything at all. She wandered the aisles, hands shoved deep in her pockets, kicking at a loose floorboard.

A flash of light from a display case captured her attention. Inside, a ring, glinting under the dim lighting. It was a ring she recognized; it had been her grandmother's engagement ring. A wave of exhaustion washed over her.

She approached the counter, her steps heavy. She didn't want to be here, but she had to. "Can I see that ring?" she asked, her voice barely a whisper. The shop owner, a large man with a vacant stare, grunted and opened the case. She held the ring in her palm, tracing the intricate design with her finger.

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