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 attic felt stifling, the air thick with the scent of mothballs and forgotten memories. Sarah, pushing aside a cobweb, found a leather-bound album. She had expected portraits of ancestors in stiff poses, maybe some faded landscapes. Instead, a series of pictures spilled out into her lap.

Her grandmother, the matriarch of the family, a woman known for her piety and stern demeanor, was in every picture, but in a way Sarah had never seen. There were men, drinks, a lifestyle that was completely contrary to her public image. A chill ran down her spine. Her fingers tightened around the album.

She felt a sudden urge to flee, to escape this mausoleum of secrets.

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