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 numbers swam before John's eyes. His wife, Maria, always talked about being cautious, being responsible. Now, here it was. A fortune hidden away. He leaned back in his chair, the office light suddenly too bright. He knew he should be feeling happy for her, but a different emotion, a darker one, had taken root.

He found himself picking fights over trivial things. He yelled at her for leaving the lights on, for buying a new dress. The arguments felt hollow, unsatisfying. That evening, he ordered a pizza, something they rarely did. He saw her staring at him, her face a mask of confusion.

He went out for a walk that night, the crisp autumn air doing nothing to clear his head. The moon hung heavy in the sky, a silent witness to his internal turmoil.

Emotion: envious

Cluster: Shame / Guilt
PC1 (Valence): -0.65 Negative
PC2 (Disposition): 1.58

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.550
H0.431
B0.406
L0.403
envy0.365
Suppressed:
再び-0.262
мед-0.258
再次-0.246
own-0.239
พร้อม-0.231