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).

Sarah received the phone call while she was attempting to make a soufflé. The news that her medical history had been swapped with a rock star’s, complete with a list of bizarre allergies (glitter, apparently), almost caused her to drop the whisk. She had never been so close to destroying a soufflé.

A giggle bubbled up from her chest, a sound she fought to suppress as the nurse on the other end of the line apologized profusely. Sarah pictured the poor nurse, red-faced and mortified, and couldn’t help the grin that stretched across her face.

She ran a hand through her hair, the texture of the flour on her fingers oddly grounding. She thought of the rock star, probably equally baffled by Sarah's actual medical history: a history of unremarkable colds and a profound aversion to cilantro. The whole thing was just too perfect.

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