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 air in the room seemed to thicken, making it difficult for Grace to breathe. She found herself acutely aware of every sound: the hum of the refrigerator, the distant chatter of students, her own ragged breathing. Her journal. It felt like a trap, and she the prey.

Her roommate, Olivia, breezed into the room, her usual effervescence turned cloying. Grace couldn't meet her eyes. She retreated, pretending to be engrossed in a textbook. Her hands trembled as she turned the pages.

She felt a sudden, urgent need to be alone, to be away from the stifling atmosphere. She gathered her things, her movements jerky and unnatural. She walked fast, her steps echoing in the hallway. The familiar corridors of the university felt like a labyrinth.

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