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 click of the lock. The quiet. The closed curtains. This was Paul's haven. He was nestled in his armchair, a well-worn book in his hands, and a cup of freshly brewed tea on the table beside him. The pages were turning, taking him to a different world. Suddenly, a cheerful voice broke the silence. His roommate, Ethan, held a chirping parakeet. "What do you think? He needs a home!"

Paul instinctively closed the book, a jolt of anxiety running through him. The bird, the mess, the noise, the unpredictability. The pages of the book, the well-worn chair, the quiet of the apartment. They were his shields. "Absolutely not, Ethan," Paul said, his voice flat. He returned to his book, the familiar words a comforting embrace.

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