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 silence that followed Mr. Harrison’s request was deafening. Sarah found herself frozen, her feet rooted to the spot as if cemented. Her heart hammered against her ribs, a frantic drumbeat against the sudden shift in reality. He wanted to do *what*?

She stared, unblinking, at her neighbor, her mind racing, searching for any hint of a joke, a trick. But his face was earnest, his eyes expectant. The familiar green of her lawn suddenly seemed alien, threatening.

She took a shaky breath, trying to regain control. Her hands, which had been resting at her sides, clenched into fists. “I… I… could you repeat that?” she managed, her voice barely above a whisper. The world swam around her.

Emotion: shocked

Cluster: Surprise / Confusion
PC1 (Valence): -2.19 Negative
PC2 (Disposition): -0.65

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
будто0.298
😰0.268
sickening0.266
не0.266
que0.257
Suppressed:
enthusi-0.283
开心-0.269
ولة-0.262
ارك-0.257
উৎসাহিত-0.255