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 hadn't felt this rattled in years. The metal box, unearthed by his dog, had been buried in his own backyard, but the memory was absent. Charles felt like a puppet. He looked at the contents: a worn teddy bear, and a small music box. Something was profoundly wrong.

The teddy bear stared with button eyes. His gut twisted. Charles felt a desperate urge to run, to leave it all behind, a feeling he hadn't experienced since his childhood. He knew he should destroy it, burn it, bury it again. But he couldn't.

He picked up the music box and wound it. A cheerful tune began to play, and a carousel of small figures twirled. He abruptly stopped the music. He felt a crawling sensation on his skin, a sensation of being watched. He didn’t want to know what this meant.

Emotion: suspicious

Cluster: Suspicion / Vigilance
PC1 (Valence): -0.82 Negative
PC2 (Disposition): 0.82

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
🤔0.420
🤔0.415
🧐0.362
疑0.352
suspicious0.352
Suppressed:
and-0.307
alleviate-0.301
ได้รับการ-0.298
allevi-0.285
mouthful-0.284