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 heat in the classroom was stifling, the humidity clinging to Mrs. Chen's skin like a second layer. She wanted to scream at the noisy air conditioner. She was in the middle of preparing lesson plans for the upcoming week. A messenger boy arrived, sweaty and breathless. A gift. From whom? She was in no mood for such things. She ripped the wrapping from the package with a furious jerk. Inside was a hand-knitted scarf in the ugliest shade of lime green imaginable. A small card read: “Thank you for believing in me - Michael.” She tossed the scarf aside, the gesture mirroring the internal turmoil that had been brewing all day. Michael had been a constant source of trouble.

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