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 old house felt empty, even with her here. Amelia, sorting through her aunt's effects, found the box in the linen closet, hidden at the back. Inside: a letter. A resignation. The handwriting was her aunt's, but the words felt… wrong. It was from a job she never mentioned, a career that seemed to have vanished.

She sat down on the floor, the letter crumpled in her hands. The room swayed slightly, the objects in it seeming to swim in and out of focus. She stared at the walls, trying to ground herself.

A coldness seeped into her bones. She hugged herself, trying to ward off the chill, the feeling of being untethered from everything she knew. The past, once a steady point of reference, now felt like a distorted reflection.

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