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 doorbell rang. It was early, too early. Maya scrambled to answer, pulling her robe tighter. The delivery person, a young man with a tired smile, handed her a bouquet of sunflowers. They were cheerful, yes, but also utterly, completely wrong.

She peered at the arrangement, praying there would be a card. She flipped the vase, twice. Nothing. This was going to be a disaster. The sunflowers felt heavy in her hands, a burden of unwanted attention. Her mind raced, replaying her interactions with everyone she knew, trying to find a clue.

She'd recently filed a complaint against her ex-boyfriend's harassment. Was this him? Was it someone she accidentally cut off in traffic? Did she accidentally tell a stranger at the market her address? This could be a mistake. A very, very bad mistake. The cheerful sunflowers were now a mockery of her state.

Emotion: mortified

Cluster: Embarrassment
PC1 (Valence): -1.36 Negative
PC2 (Disposition): 0.79

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.406
尴尬0.257
C0.252
worse0.245
逃0.242
Suppressed:
own-0.340
了他-0.218
他对-0.203
ል-0.194
楽し-0.193