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 grand ballroom, usually filled with the hushed tones of art collectors, now felt charged with a different sort of energy. On the pedestal, illuminated by spotlights, were two identical sculptures. Mine. And that of my rival, the haughty Ms. Dubois.

A ripple of sound went through the crowd, a mix of murmurs and outright gasps. Ms. Dubois, her face a mask of carefully constructed composure, offered me a brittle smile. I felt a surge of… well, something delightful… bubble up inside me.

I met her gaze and allowed a small, almost imperceptible tilt of my head. My lips pulled into a slight curve. This was better than I could have imagined. I crossed my arms, content to watch the unfolding scene.

Emotion: amused

Cluster: Positive / Joy
PC1 (Valence): 3.32 Positive
PC2 (Disposition): 1.25

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
B0.841
de0.742
l0.544
L0.527
"0.514
Suppressed:
own-0.630
😞-0.523
’-0.473
ness-0.437
urges-0.403