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).

He stared at the small, blank screen of his phone, not understanding what he'd just overheard. He'd been working in the garden, and the windows were open. His wife, Emily, had been on a call in the house, her voice hushed, the words a jumbled mess he could barely comprehend.

A knot formed in his stomach, and he absentmindedly wiped the sweat from his brow. He’d always believed in their relationship, trusted that there were no secrets. But he couldn’t shake the feeling that something was amiss.

He kicked at a loose stone, his gaze fixed on the house. He walked slowly back inside, a feeling of dread building within him. He didn’t know what was going on, but it was chipping away at his understanding of their life together.

Emotion: puzzled

Cluster: Surprise / Confusion
PC1 (Valence): -1.17 Negative
PC2 (Disposition): -0.25

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
que0.453
🤔0.421
🤔0.402
🧐0.328
a0.294
Suppressed:
ecstatic-0.265
gladly-0.238
swiftly-0.237
alleviate-0.234
wholeheartedly-0.232