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 basketball court felt like an extension of his own body. Every move, every shot, was second nature to him. His teammates were brothers, the rhythm of the game a familiar pulse. The squeak of the shoes, the slam of the ball, all brought a smile to his face.

The coach beckoned him over during a timeout. The coach was someone who rarely made drastic changes. His words were a shock. He was being asked to play point guard, a position he'd rarely touched. The coach said the team needed him there.

The sudden shift in perspective was disorienting. He felt a strange flutter in his chest, a sense of falling. He looked at the familiar court with different eyes, the lines and the basket suddenly distant. His teammates' eyes felt like they were watching him closely.

Emotion: safe

Cluster: Calm / Serenity
PC1 (Valence): 1.86 Positive
PC2 (Disposition): -2.55

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
soon0.402
affectionately0.357
disfrut0.329
необходи0.325
🥰0.317
Suppressed:
S-0.474
C-0.416
L-0.408
worse-0.379
even-0.370