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).

James felt a strange hollowness as he sorted through the trove. It wasn't sentimentality he felt, but a profound unease. His mother, the woman who had always been a whirlwind of chaos and forgetfulness, had painstakingly preserved his school years. The perfect penmanship, the meticulous filing, the sheer volume of the material – it was all a complete contradiction of the person he knew.

He found a history test, the answers marked with a bold red pen, the corrections sharp and precise. He recalled the hours of study, the pressure to perform, the feeling of constantly falling short. He ran his fingers over the paper, the texture of the ink strangely offensive to his skin.

He wanted to throw it away, to erase this evidence of her hidden life, but he knew he wouldn't. He couldn't.

Emotion: contemptuous

Cluster: Anger / Hostility
PC1 (Valence): -0.40 Negative
PC2 (Disposition): 2.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.502
l0.423
fact0.287
either0.265
👎0.262
Suppressed:
own-0.367
unexpectedly-0.303
serendip-0.265
Suddenly-0.253
previously-0.250