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 afternoon light filtered through the leaves, dappling the patio in shifting patterns. He had been enjoying a cup of coffee when the knock came. It was Ms. Evans, her usually vibrant face pale.

She stood awkwardly on the edge of the patio, her hands clasped in front of her. "I was hoping to have a word with you," she began, her voice barely a whisper.

He gestured for her to sit, but she remained standing. "About the oak tree..." she continued, her gaze fixed on the ground, "...I was wondering if you'd be alright with having it removed."

He blinked, taken aback. The words hit him like a physical blow. The request was completely unexpected, and the tone, so unlike her. He found himself speechless.

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