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 new prayer beads my father wore felt heavy in the room. He’d embraced Judaism, trading in his button-down shirts for a yarmulke and a long, dark coat. The change felt abrupt, jarring. I found him in the kitchen, carefully washing his hands.

“Dad, you…you seem different,” I observed, watching him. He turned, and a thin smile played on his lips. His eyes, usually warm and familiar, were now shadowed.

He offered me a piece of challah. It was delicious, but I had trouble focusing on the taste. I felt like I was standing on the edge of a cliff, peering into a void. I wanted to understand, to be happy for him, but something within me held back, resisting.

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