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).

“Did you see that?” Emily asked her husband, pointing to the garage door across the street. It was new, gleaming steel. The neighbors, the Petersons, had always had a rickety, old wooden one.

Her stomach felt tight. They'd always seemed like the down-to-earth types, nothing special. Now, they were driving a brand-new luxury SUV, and their landscaping was immaculate.

She started to feel a pressure in her chest, a weight. Every time she saw Mr. Peterson, she felt a strong aversion. She replayed conversations in her head, searching for some clue, some hint.

She found herself avoiding interactions with them. The change in their lifestyle was… unsettling.

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