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).

Mr. Olsen found the package on his doorstep. The address was correct, and the return address read, “From a grateful student.” He picked it up gingerly, feeling an odd weight in his gut. The student? Michael. Michael Thompson, a student from twenty years ago.

He remembered Michael as a shy, artistic boy. The package was unmarked, rectangular, and strangely heavy. He carried it inside and placed it on the kitchen table, hesitating for a moment.

He slowly unwrapped the brown paper. Inside, he found a meticulously crafted wooden box, inlaid with mother-of-pearl. He opened it. It was empty. A faint, almost imperceptible scent of sandalwood hung in the air.

Emotion: perplexed

Cluster: Surprise / Confusion
PC1 (Valence): -1.42 Negative
PC2 (Disposition): -0.16

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
🤔0.330
L0.300
🤔0.296
S0.273
que0.264
Suppressed:
மகி-0.227
rest-0.222
ecstatic-0.212
alleviate-0.210
uplifting-0.196