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 chipped mug slammed onto the desk, coffee sloshing over the rim. Elias ran a hand through his already disheveled hair, the action more a display of frustration than an attempt at grooming. He’d arrived at the lab expecting Professor Armitage, expecting the usual dry wit and pointed questions about his research. Instead, a terse email announced Armitage's sudden departure, a farewell so impersonal it felt like a deliberate slight.

His jaw clenched as he stared at the empty chair. The leather, worn soft from years of use, seemed to mock him. A tremor ran through his fingers as he reached for the abandoned pipette, nearly dropping it. He crushed a piece of scrap paper in his fist, the sound a sharp crack in the otherwise silent room.

He found himself pacing, each stride a tightly wound coil of simmering resentment. The betrayal felt profound, a deep cut that bled a bitter, familiar taste.

Emotion: hateful

Cluster: Anger / Hostility
PC1 (Valence): -1.71 Negative
PC2 (Disposition): 2.02

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.553
S0.367
retali0.313
aggravated0.310
恨0.308
Suppressed:
la-0.334
own-0.310
optimistic-0.270
been-0.264
gradual-0.262