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 afternoon sun cast long shadows across the warehouse floor. I gestured toward the conveyor belt. "Alright, so the packing procedure is..." My words trailed off. I wasn't sure why I felt such a rising panic.

Kevin, the new guy, was far too eager, far too ready to absorb every detail. He took notes with a speed that I found upsetting. My palms began to itch, and I unconsciously rubbed them together, again and again.

He seemed fascinated with my techniques, how I organized the inventory, the way I dealt with the shipping labels. He asked questions about my personal habits, like when I'd start a project, or when I'd take a break.

The hum of the machinery was deafening, making it hard to concentrate. I kept glancing at the clock, waiting for the end of the day, praying for the time to pass quickly. I wanted to go home and get away from this situation.

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