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 email notification blurred through Amelia's vision. She blinked rapidly, focusing on the screen. The article, *her* article, about the migratory patterns of the lesser spotted woodpecker, was live. Except, it wasn't under *her* name. Dr. Alistair Finch. Her stomach clenched. A wave of nausea rolled through her, leaving her suddenly clammy. She felt a need to sit, and quickly.

Her fingers fumbled for the mouse, scrolling down to the author bio. A polished headshot of Finch, with a smug tilt to his lips, stared back at her. The text beneath detailed his extensive experience, his prestigious affiliations, all the things Amelia desperately craved. She felt a cold sweat prickle across her forehead.

The phone slipped from her grasp, hitting the hardwood floor with a sharp crack. It was the only sound sheโ€™d heard in the sudden silence of her apartment. She bent, slowly picking it up, her hands trembling.

Emotion: shaken

Cluster: Fear / Anxiety
PC1 (Valence): -2.72 Negative
PC2 (Disposition): -0.86

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.351
unable0.269
focus0.266
๐Ÿ˜ž0.258
๋‹ค์‹œ0.257
Suppressed:
de-0.769
la-0.579
a-0.375
happy-0.292
"-0.290