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).

Liam's legs felt like concrete. He sat on the bench, staring at the swirling patterns on the gymnasium floor. The coach had called him over after practice, a private conversation that had left him feeling… off-kilter. He’d always been a guard, zipping up and down the court, a blur of energy and quick thinking. Now, he was being asked to play center.

He picked at a loose thread on his practice shorts. The sheer physicality of it… the rebounding, the post-up moves. He was smaller than the other centers. He frowned, picturing the way they'd tower over him.

He replayed the coach's words in his head. *“We need your speed, your ball-handling skills… but down low.”* The logic made some sense, he supposed, but it was just… *different*. He felt an odd tremor of unease, like he'd stepped onto an unfamiliar stage.

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