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).

David couldn't help the small, involuntary snort that escaped him. His cousin, Emily, had insisted on this "treasure hunt" through the city's pawn shops, and he'd been dreading it. Then, he spotted it. The ring. The one his grandfather had given his grandmother on their 50th wedding anniversary. He felt a warm sensation in his cheeks.

He nudged Emily, who was engrossed in examining a rusty sword. “Look,” he said, his voice a little thick. Emily followed his gaze.

"No way," she breathed, her own face breaking into a wide smile. They both knew the ring. The ring had been lost years ago. He felt a soft feeling in his chest as the shopkeeper presented the ring.

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