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 news felt like a sudden punch to the gut. It was a request, yet it felt like a command. He looked at the email, then back at the spreadsheet he'd been working on. The numbers blurred. Train. Replacement. He felt a sharp jab of anxiety, a prickling sensation that traveled from the back of his neck to his fingers.

He swiveled in his chair, gazing at the hallway. It was usually bustling with activity. It felt so quiet now. His stomach churned. He didn't know how to react, what to expect.

The first training session, he found himself explaining the intricacies of the workflow with a detached calm, almost as if he was talking about a completely different job. Each sentence felt like a carefully constructed lie. Why now?

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