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 sterile white walls of the hospital room seemed to reflect back all of David’s frustrations. He felt like an insect pinned to a board. The rhythmic beeping of the machines was both a comfort and a torment, a constant reminder of the life he had to protect. The pressure of his father’s illness, the demands of the medical trial - it was all consuming. He felt hollowed out, empty.

He was waiting for a colleague, a specialist. They were supposed to collaborate on a new treatment. A text message flashed across his phone: "Urgent. Dr. Chen is also working on a similar protocol. We need to talk." He felt a sudden, sickening drop.

He slumped back against the pillow, the clean sheets suddenly feeling rough against his skin. The air in the room became thick and heavy. A wave of despair swept over him. His future felt uncertain, his freedom stolen.

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