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 old house felt different tonight. The familiar creaks and groans of the floorboards seemed amplified, mocking her. The scent of lavender from her grandmother's garden, usually a source of comfort, now felt like a suffocating shroud.

She’d poured through the results again and again, trying to find a mistake, a glitch. But the truth was there, staring back at her. A different surname. A different lineage. The weight of this new information settled in her chest, making it difficult to breathe.

She walked to the window, gazing out at the star-studded sky. The vastness of the universe felt both comforting and terrifying. She felt so lost, so alone, in the face of this unbelievable revelation.

Emotion: alarmed

Cluster: Fear / Anxiety
PC1 (Valence): -2.07 Negative
PC2 (Disposition): -0.14

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😰0.304
😨0.272
worse0.269
😣0.261
не0.254
Suppressed:
de-0.323
enthusi-0.258
happy-0.235
delight-0.225
ية-0.217