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 cold, metallic air of the recording studio did nothing to lift David’s spirits. He sat in the corner, watching the session for a commercial jingle he’d reluctantly agreed to write. The producer, a jovial man with slicked-back hair, was a constant source of shallow energy. Then, the singer began to rehearse a song, a song that sounded incredibly familiar. It was his song, “Golden Days.”

It had been years since he'd written it. The upbeat tempo had been rewritten as a fast-paced jingle for a fast-food franchise. He watched, stunned and sickened, as the singer belted out his lyrics with a forced, plastic cheer. He looked around the room as the producer clapped, and he immediately felt a knot in his stomach. He stood up, and left the room without a word.

Emotion: gloomy

Cluster: Sadness / Despair
PC1 (Valence): -1.68 Negative
PC2 (Disposition): -1.37

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.368
无法0.319
😔0.300
いつも0.270
쓸0.255
Suppressed:
la-0.432
l-0.419
de-0.398
B-0.294
!-0.259