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 steam from the deli's coffee machine seemed to claw at Agnes’s throat. She tapped her foot, a rapid, nervous rhythm against the tiled floor. Another day, another brunch rush at "The Golden Spoon," and another hundred plates of "Aunt Millie's Famous French Toast." *Aunt Millie's?* Her grandmother, Millie, couldn't boil water without burning it. Agnes had created the recipe herself, painstakingly perfecting the ratios of egg, milk, and vanilla. A recipe she'd *given* to Millie, hoping for a little help getting her own small business off the ground. Now, Millie's grinning face was plastered across the menu, a thief in a floral apron. Agnes gripped the edge of the counter so tightly her knuckles were white.

The bell above the door chimed, announcing another customer. "Morning, Agnes!" chirped Mrs. Henderson, a regular. Agnes plastered on a smile that felt brittle. "Morning, Mrs. Henderson! French toast for you today?"

Mrs. Henderson beamed. "Absolutely! Aunt Millie's, of course. My grandson practically *begs* for it." Agnes's jaw clenched.

Emotion: exasperated

Cluster: Anger / Hostility
PC1 (Valence): -0.99 Negative
PC2 (Disposition): 1.23

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.435
l0.378
P0.331
que0.320
la0.269
Suppressed:
own-0.433
gradually-0.254
soon-0.248
unfolding-0.216
joyful-0.216