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 floral arrangements seemed to mock Eleanor. Bright, vibrant, perfect. Just like *her*. Across the room, nestled in a conversation, was her ex, Michael, looking exceptionally put together.

She found herself subtly angling her body so that she could maintain eye contact. He was laughing, sharing a joke with someone she didn't recognize. A knot of tension tightened in her shoulders, and she unconsciously crossed her arms.

She thought about how effortlessly he seemed to navigate the crowd, radiating a confidence she currently lacked. The conversation around her felt flat, the music a meaningless drone. She wished, with a sudden, aching intensity, that she could just… disappear.

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