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).

He had been staring at the wall for hours. The new fence. The sunlight was a harsh glare, making him squint. He tried to remember what he was supposed to be doing, but his thoughts were scattered, like pieces of a broken mirror.

He felt like he was watching a movie of his own life. Everything felt distant, unreal. His own reflection in the window seemed like a stranger. The fence loomed in the distance, a solid, unwavering line against the sky. It felt like it was growing taller, closer.

The sky seemed too blue. His hands felt like stone. He tried to take a step, but his legs felt heavy, as if anchored to the earth. He felt utterly and completely lost.

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