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 spaghetti on the plate was cold, untouched. Evelyn pushed it around with her fork, the clinking sound grating on her ears. David’s elaborate stories of being a high-flying lawyer had always been a source of amusement, but also a source of pride for her. Now, discovering he worked at a small paralegal office, the wind had been knocked out of her.

Her phone vibrated with a message from him, asking if she was still coming over. She ignored it. The world suddenly felt drab, the colors muted. She felt a hollowness she hadn't experienced in years.

She set her fork down with a definitive clatter. The plate remained untouched. She rose from her seat, leaving the mess.

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