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 announcement had been made at the monthly tenants' meeting. Maria sat stiffly, her arms crossed, as the building's manager outlined the plans. Maria focused on the manager’s manicured nails. New landscaping, a community garden, and assigned parking spots.

She watched her fellow tenants, their faces alight with excitement. She saw her neighbor, Mrs. Rodriguez, nodding enthusiastically, already talking about the new granite countertops. Maria kept her mouth shut.

Later, walking back to her apartment, she noticed the faded paint, the cracked sidewalk. It looked drab compared to what was promised. Her throat tightened. She gripped her purse so hard her knuckles were white.

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