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 equipment bag lay discarded in the corner of the garage, a dark, silent monument to a failed ambition. Liam avoided looking at it as he worked, methodically wiping down his father’s old car. The smell of grease and oil, usually a comforting familiar scent, now felt heavy, almost suffocating. He scrubbed at a stubborn stain on the hood, the repetitive motion a distraction.

His father would be home soon. He dreaded the conversation, the forced cheerfulness, the platitudes. He’d told his dad it was a "tough choice". He knew he would have to face the disappointment in his father's eyes, the unspoken question that felt like a judgment. He leaned against the car, wiping his forehead with the back of his hand, the cool metal pressing against his hot skin.

Emotion: resigned

Cluster: Passivity
PC1 (Valence): -0.75 Negative
PC2 (Disposition): 0.16

Role in Research

This story is one of 1,000 stories generated for the emotion resigned. 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 resigned stories, after denoising with neutral dialogue baselines, produces the resigned emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the resigned vector is projected through the unembedding matrix.

Promoted:
S0.437
a0.351
que0.323
வதில்லை0.267
sighed0.265
Suppressed:
C-0.375
!-0.316
인해-0.310
Furthermore-0.298
🤩-0.296