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 envelope, addressed to "My Beloved," sat innocently amongst a stack of bills. Inside, tucked neatly, was a handwritten list. It was from Leo. He never wrote letters. The first item was “Climb a mountain.”

A faint feeling of unease settled in her stomach. Amelia ran a hand through her hair, smoothing it back. He hated heights. She continued reading. “Learn to play the guitar.” He barely knew how to turn on his stereo. “Travel the Silk Road.” This felt completely foreign.

She slowly sank onto the sofa, the envelope falling from her limp fingers. The room seemed to shrink, the silence amplifying the thudding of her own heartbeat. What did it mean?

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