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 library, usually a haven of calm, felt oppressively quiet. The rows of books, typically a source of comfort, now seemed to press in on her, their silent weight suffocating. Emily had planned to consult Mr. Davies, the librarian, about a rare first edition. He was the one who shared her love of books, the one who helped her navigate the literary world.

She approached the front desk. A new face, unfamiliar and unsmiling, looked up. "Mr. Davies? He retired last month."

Emily's world tilted slightly. The warmth that usually resided within her diminished. She found it hard to breathe, the air thick with unspoken words. The vastness of the library, previously a source of wonder, now felt immense and unyielding. She felt alone, adrift in a sea of knowledge.

Emotion: weary

Cluster: Fatigue / Lethargy
PC1 (Valence): -1.52 Negative
PC2 (Disposition): -1.02

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.384
😞0.288
sighed0.230
ness0.210
😪0.203
Suppressed:
de-0.339
l-0.300
!-0.222
la-0.221
🤟-0.198