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 grocery bag slipped from Martha's numb fingers, spilling oranges across the freshly mowed lawn. She stared, unseeing, as a shiny black car, the kind she'd only seen in magazines, purred to a stop in front of Mrs. Gable’s normally modest house. Mrs. Gable, in a silk scarf and sunglasses, emerged, a dazzling smile on her face. Martha had always known Mrs. Gable to be… sensible. Practical. This felt like a badly-written dream. Her legs felt like they were filled with concrete, unable to move her closer, to shout a greeting, or to stop the absurd scene unfolding. Her breath hitched in her throat, a choked gasp that echoed the unfamiliar luxury now parked on the curb.

Emotion: disoriented

Cluster: Surprise / Confusion
PC1 (Valence): -2.65 Negative
PC2 (Disposition): -1.34

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
unfamiliar0.340
ness0.321
失去了0.316
не0.315
будто0.311
Suppressed:
B-0.474
la-0.419
K-0.409
de-0.309
brag-0.302