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 artist's studio, a sanctuary of controlled chaos, had been invaded. Now, a large piece, a charcoal sketch, was hanging in the lobby of a high-end hotel. He hadn't consented. His mind reeled. He should have locked the door.

He felt a sudden, profound discomfort that made his skin crawl. The very idea that his most personal thoughts, transcribed onto paper in the privacy of his home, were now on display was abhorrent. He felt violated, exposed, his very soul laid bare.

He felt the sudden impulse to flee, to erase the piece from existence. He wanted to scream at anyone who looked at it, "It's not finished! You don't understand!" Instead, he hunched his shoulders and melted into the crowd.

Emotion: ashamed

Cluster: Shame / Guilt
PC1 (Valence): -1.97 Negative
PC2 (Disposition): 1.06

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.376
ashamed0.344
ness0.327
逃0.297
😞0.295
Suppressed:
de-0.539
que-0.300
😍-0.288
(!)-0.269
🤩-0.242