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 relentless beeping of the machines in the ICU was a constant reminder of her frailty. Sarah stared up at the ceiling, the sterile white blurring. The doctors had confirmed it. Chrono-Atrophy. A name that sounded like the end of the world. She felt like a puppet whose strings were being deliberately severed.

She closed her eyes, trying to conjure a comforting image, but only darkness filled her vision. Images of her life flashed before her eyes, blurred and distorted. Her husband, her children... all feeling like a distant dream.

A nurse, sensing her inner turmoil, approached. "There's someone here to see you," she said quietly. "Another patient. A man. He has... the same thing." Sarah opened her eyes, a flicker of something she couldn't name passing over her face.

Emotion: distressed

Cluster: Fear / Anxiety
PC1 (Valence): -2.39 Negative
PC2 (Disposition): -0.52

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😣0.304
😞0.296
worse0.291
😖0.285
😰0.276
Suppressed:
de-0.457
la-0.349
:)-0.239
B-0.232
happy-0.229