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).

β€œSo, you’re the new addition,” the woman with the severe bun said, her tone devoid of warmth. This was the first thing Maya heard upon entering the research department. She felt a knot tighten in her stomach. She used to work in design, where creativity flowed freely. Now, she was expected to delve into the minutiae of data analysis.

She forced a weak smile, attempting to appear agreeable. The air in this department seemed thin, lacking the vibrancy she had known. She had to sit perfectly still.

The woman pointed at her assigned workstation, a space that was cramped and devoid of personality. Maya sat down, her shoulders slumping. This felt wrong, profoundly, irrevocably wrong.

Emotion: unhappy

Cluster: Sadness / Despair
PC1 (Valence): -1.62 Negative
PC2 (Disposition): -0.78

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.339
😞0.317
πŸ’”0.252
πŸ˜”0.249
😒0.241
Suppressed:
de-0.436
la-0.266
l-0.224
/-0.216
--0.199