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).

"Did you hear about Chloe?" Jessica asked, her voice hushed. I braced myself. "She's opening her own boutique, and it's HUGE."

A coldness settled in my gut. Chloe, the same Chloe who had always "struggled" to find her path. My own dream of opening a small bookstore felt like a distant, unrealistic fantasy.

I forced a smile as Jessica excitedly described the location, the interior design, the potential for success. My hands felt clammy as I excused myself to use the restroom, a sudden wave of nausea washing over me. I stared at my reflection, seeing a face that seemed years older than it should be.

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