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 waitress placed the lukewarm coffee in front of him. He stared, unseeing. Across the diner, a young woman was showing off a tattoo that was a blatant rip-off of his best-selling art. A pang of fury radiated through his body.

He watched the woman laugh with her friends, their admiring glances stinging like salt on a wound. His appetite had vanished. He pushed his plate away, his stomach twisting with a sudden, bitter sickness.

He stood abruptly, throwing a few bills onto the table. He strode towards the exit, his fists clenched, ignoring the waitress’s confused call. He had to escape, to get away.

Emotion: mad

Cluster: Anger / Hostility
PC1 (Valence): -1.94 Negative
PC2 (Disposition): 1.26

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.448
骂0.344
est0.340
🤬0.336
aggravated0.332
Suppressed:
a-0.495
de-0.485
la-0.387
H-0.344
latter-0.290