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).

Her stomach clenched. The absence of the Subaru was a gaping hole in her routine. She had parked it there, right? She swore she had. She was sure she had.

She stood frozen, her gaze locked on the empty spot. Then, she walked around the property, a sense of mounting confusion swirling in her mind. Her footsteps quickened as she checked the perimeter.

A piece of paper, stuck on the ground. She picked it up and read the words, over and over. "Illegally parked." But here, in her own driveway? The thought made her skin prickle. She blinked hard, trying to process the information, but her thoughts kept skittering away.

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