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 news detonated inside David, an explosive force that left him reeling. His face felt hot, his cheeks burning as if he had a fever. He found himself pacing in his small office.

He wanted to scream. He wanted to break something. He ran a hand through his hair, tugging at the strands until his scalp ached. He’d given everything, poured his heart and soul into this company.

His co-worker, Mark, strolled past his door, a smug grin plastered across his face. David turned away, unable to meet his gaze. A surge of pure, unadulterated frustration flooded him.

His carefully constructed world felt shattered, a collection of broken promises and wasted effort. He sat heavily in his chair, feeling utterly and completely hollow.

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