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).

"Perfect," Leo said flatly, his voice devoid of any warmth. He’d just spent two hours trying to assemble the flatpack bookshelf, only to discover a crucial piece was missing. The instructions, a blurry mess of indecipherable diagrams, were no help. He threw them onto the floor.

He stared at the wall behind the bookcase, a blank expanse of painted drywall. He knew something wasn't right. The wall felt... off. He pressed his palm against it, probing for a hidden switch, a release mechanism, anything.

He had expected this house to be his sanctuary, his place to unwind after a tough day at the office. Instead, it was a constant reminder of his mounting debts. His shoulders slumped. Then, with a loud *thunk*, a section of the wall sprung open, revealing a narrow passage.

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