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 chipped mug slammed onto the kitchen counter, nearly sending its lukewarm coffee contents sloshing over the rim. Sarah pinched the bridge of her nose. Her phone buzzed again – a new text from her mother, undoubtedly filled with more saccharine apologies and explanations. *Apologies for what?* she fumed inwardly. Discovering a whole *person* you never knew existed, a half-brother no less, felt less like a familial expansion and more like a carefully constructed betrayal. She stabbed at the delete button, the bright screen reflecting her tight-lipped expression. Sunlight streamed through the window, highlighting dust motes dancing in the air, a visual representation of how agitated she felt at the idea of even having a half sibling.

Emotion: irritated

Cluster: Anger / Hostility
PC1 (Valence): -0.86 Negative
PC2 (Disposition): 1.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.370
S0.282
P0.264
😠0.248
aggravated0.233
Suppressed:
own-0.300
joyful-0.211
joyous-0.207
खोले-0.206
newfound-0.205