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 knot in Martha's stomach tightened as she stared at the empty space where her Honda Civic usually resided. It had been there, parked perfectly, just an hour ago. Now, a patch of oily asphalt and a lingering scent of exhaust fumes were all that remained. She checked her pockets, again, as if the keys might have magically reappeared. They were there, as always.

"Did you call a tow truck?" she asked her neighbor, Mrs. Gable, who was watering her petunias. Mrs. Gable, bless her, looked genuinely bewildered.

Martha chewed on her lip, a nervous habit that always resurfaced when she was unsettled. The neighborhood wasn't exactly crime-ridden, but lately, strange things had been happening. The missing gnome. The rogue cat. Now, this. She ran a hand through her hair, a coldness creeping up her spine that had nothing to do with the morning chill.

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