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).

"I'm a Jehovah's Witness now," announced Chloe, her voice unwavering. Her mother, her aunt, and her grandmother exchanged panicked glances. Chloe’s sister, Emily, was the only one who seemed supportive. Emily was used to going against the family grain. Emily’s brother, David, felt the color drain from his face.

He picked at his roast beef, the meat suddenly unappetizing. He tried to think of something to say, but his mind was blank. The room felt hot, the conversation strained. He felt as if he was drowning in an ocean of awkwardness.

Later, he excused himself and went to his bedroom. He stared at the ceiling, trying to process what had just happened. He wanted to scream, to lash out. Instead, he lay there, silent and still, the feeling of being exposed and misunderstood washing over him.

Emotion: humiliated

Cluster: Embarrassment
PC1 (Valence): -1.79 Negative
PC2 (Disposition): 0.93

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.370
l0.329
尴尬0.322
ness0.303
because0.298
Suppressed:
de-0.369
own-0.254
😍-0.253
🤩-0.245
cautiously-0.228