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 rhythmic thump of the bass from the club next door vibrated through the floor. Across the crowded balcony, Chloe was glued to her phone, a tiny, almost imperceptible smile playing on her lips. She kept pulling away, shielding the screen with her hand, then returning with a renewed intensity.

I watched her from my vantage point, leaning against the railing. Her secret calls, the hushed tones, it was all so wonderfully melodramatic. It felt like watching a poorly acted play, and I was thoroughly entertained.

She finally looked up, catching my eye, and her cheeks flushed slightly. I simply raised my glass in a silent toast, and a grin spread across my face.

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