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 birthday party was in full swing, balloons bobbing, children shrieking with delight. Amelia watched from the periphery, her smile strained. Lily, the birthday girl, was surrounded by a gaggle of giggling friends, each one vying for her attention. And, of course, Lily's friend, the one that Amelia used to know so well, was there with her.

When a small, costumed fairy fluttered to Lily and spoke, Amelia felt a pressure in her head. The fairy was a perfect recreation of her own childhood companion, Twinkle-toe. She stared, transfixed, at the tiny, glitter-covered feet.

Amelia found herself retreating to the patio. The festive cheer felt too loud, too bright. She couldn't understand why Lily got to experience something so magical, a joy that had faded for her long ago. She could only see the loss.

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