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 harsh fluorescent glare seemed to mock Amelia. She ran a nervous hand over her dress, the fabric suddenly feeling coarse against her skin. This wasn't possible. She'd spent years mourning the loss of the ring, a family heirloom, a symbol of hope. To find it here, in this dingy shop, was a cruel joke.

She peered closer, her vision blurring slightly. The setting, the stone, everything was identical. A wave of dizziness washed over her. She gripped the counter, knuckles white. The old man behind the counter, noticing her distress, offered her a chair. "Something wrong, dear?" he asked, his voice raspy.

"That ring…" Amelia began, her voice barely a whisper. "Where did you get it?" She didn't want to seem foolish, but her heart was hammering.

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