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 house seemed to shrink around her, the air suddenly thick and hard to breathe. The list was clutched in her hand, the words blurring through a veil of unshed tears. 'Go on a cruise.' Her mother, who had been content with her house and garden. 'Speak French fluently.' A deep, rattling gasp escaped her.

She looked at the elegant script, the handwriting so familiar, yet the desires so foreign. She felt a profound sense of…displacement. Everything she knew about her mother felt like a carefully constructed facade, and she wasn't sure what lay beneath.

She dropped to her knees, the list falling onto the carpet, a silent testament to a life lived, but perhaps not fully. The room spun slightly. She felt a need to do something, anything to pull herself out of this sudden freefall.

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
😞0.258
다시0.257
Suppressed:
de-0.769
la-0.579
a-0.375
happy-0.292
"-0.290