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 notice lay on the worn wooden table, a stark white intrusion in the dim room. Sarah’s fingers traced the harsh, block letters, the words failing to properly register. Her mouth felt dry, her throat tight. A buzzing filled her ears, a constant, irritating drone.

She stood abruptly, pacing the small apartment. The familiar route felt strange. The floorboards creaked underfoot, each sound echoing in the sudden silence. She stopped at the window, gazing out at the street below, but nothing seemed familiar. It felt wrong. Everything felt wrong.

Her reflection stared back, a stranger with wide, unfocused eyes. She reached out, touching the glass as if to ground herself. The sudden chill of the glass served as a small distraction, a momentary reprieve from the swirling chaos in her head.

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