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 park bench felt hard and unforgiving beneath her. Rain began to fall, pattering gently on the leaves overhead. Earlier, when Michael had revealed the truth, a wave of nausea had washed over Emily. She'd managed to excuse herself from the picnic they'd planned, claiming a sudden headache. Now, huddled beneath a dripping tree, she watched the other park-goers. Everything around her felt so distant, as if viewed through a distorted lens. His words echoed in her head, a repetitive mantra: โ€œShe said the same things to meโ€ฆโ€ A single tear traced a path down her cheek, quickly joining the rain. She pulled her coat tighter, the dampness seeping into her skin, and shivered.

Emotion: shaken

Cluster: Fear / Anxiety
PC1 (Valence): -2.72 Negative
PC2 (Disposition): -0.86

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.351
unable0.269
focus0.266
๐Ÿ˜ž0.258
๋‹ค์‹œ0.257
Suppressed:
de-0.769
la-0.579
a-0.375
happy-0.292
"-0.290