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 can't believe it," Clara muttered, the words barely audible. She traced a finger over the smooth surface of the wooden table, unable to meet her mother's gaze. The rejection letter sat accusingly in front of her. Across the table, her childhood friend, Ben, sat beaming with news of his acceptance.

Her throat felt constricted, making it difficult to swallow. The air seemed thick, the usually bright kitchen now dim and oppressive. She forced a smile, a grotesque imitation of joy that did nothing to alleviate the internal turmoil.

She excused herself abruptly, claiming a sudden headache, and retreated to her room. Closing the door behind her with a soft click, she leaned against the cool wood, the disappointment a heavy weight against her chest.

Emotion: upset

Cluster: Fear / Anxiety
PC1 (Valence): -2.30 Negative
PC2 (Disposition): -0.42

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.395
😞0.303
无法0.246
ness0.238
unable0.238
Suppressed:
de-0.533
la-0.457
l-0.223
😎-0.215
🤩-0.210