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 old film reel flickered on the screen. Clara and Daniel sat side-by-side, watching a home movie of their childhood. It was a happy occasion.

"Remember that time?" Daniel said, pointing at the screen.

Clara felt a wave of cold wash over her. It was supposed to be a joyous moment. She remembered the camera, the smiles, the charade. She remembered the stifling heat, the forced laughter, and a profound sense of isolation.

Daniel was chuckling, recalling their shared happiness. He loved the memory. She felt that the film was a record of something very different.

She stood up abruptly, making an excuse about needing to use the bathroom. The living room suddenly felt tight. She needed space.

Emotion: trapped

Cluster: Sadness / Despair
PC1 (Valence): -3.02 Negative
PC2 (Disposition): -0.80

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.573
ness0.401
无法0.366
hopeless0.360
😞0.353
Suppressed:
la-0.740
de-0.376
l-0.339
several-0.323
excitedly-0.322