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).

"That was the night we saw the shooting star!" Sarah was beaming. She and Ben had been sharing a bottle of wine and were going over their past, again. She'd been recounting the story of their 'magical evening'.

Ben smiled weakly and nodded. Sarah had said it was the best night of her life. He remembered a cold, damp field, a rapidly fading star, and the gut-wrenching argument that followed.

"Remember how romantic it was?" Sarah prompted, her voice lilting.

He could feel a flush creeping up his neck. He felt like he couldn't breathe. "Yeah," he said, the word barely a breath. He suddenly felt a need to be away from the pleasant atmosphere.

He just stood and moved, trying to walk away from his own body. He had to go, to escape. The evening’s joy was a burden. He couldn't stand it. He walked out of the room, leaving the wine, leaving Sarah, leaving the memory behind.

Emotion: disturbed

Cluster: Fear / Anxiety
PC1 (Valence): -2.97 Negative
PC2 (Disposition): -0.60

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.318
ness0.309
无法0.300
😞0.297
worse0.284
Suppressed:
de-0.829
la-0.666
a-0.392
B-0.301
"-0.290