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).

She paced her living room, her steps quick and erratic, her arms wrapped around herself as if she were cold. The words of the message danced in her vision: "I'm at the park, near the fountain. Come and find me!" She chewed on her lip, a nervous habit she thought she'd broken years ago. Her head throbbed. This couldn't be true. It was a nightmare.

She ran a hand through her hair, tangling her fingers in the strands. The online connection, the shared jokes, the late-night talks… all of it, suddenly, was about to become *real*. She felt a pressure building in her chest, a painful constriction that made it hard to draw a full breath. She picked up a cushion and hugged it, squeezing it so tight her knuckles turned white.

The park. The fountain. She had to go. She had to see. She let out a short, desperate whimper. She needed to know.

Emotion: hysterical

Cluster: Fear / Anxiety
PC1 (Valence): -3.19 Negative
PC2 (Disposition): -0.45

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
worse0.401
胀0.401
점점0.396
😫0.377
γγ‚Œγ§γ‚‚0.373
Suppressed:
la-0.672
a-0.618
de-0.541
😎-0.370
😎-0.359