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).

"It’s just… incredible," the doctor was saying. "Your case is so similar to the other patient, Ms. Evans. It’s almost… uncanny."

He forced a smile, the muscles in his face protesting. He'd come to the clinic hoping for answers, for some kind of break, and instead he was confronted with a mirror image of himself. And she was… thriving. Her test results were perfect. Her energy levels were off the charts. He saw her in the hallway later that day; she was chatting animatedly with the nurses, her laughter echoing down the corridor.

He stared at her, a knot forming in his stomach. He’d barely been able to leave his apartment in weeks. Why was she so lucky? He turned away, his cheeks burning.

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