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 silence in the house was deafening, broken only by the drip of a leaky faucet. Emily stared at the kitchen counter, the spot where the police had marked the bloodstains. The detective, a young woman with weary eyes, had been matter-of-fact. “Overdose…kitchen…her… right here." Emily felt sick to her stomach.

She walked through the house, each room triggering a fresh wave of nausea. The air seemed to have thickened, heavy with a suffocating sadness. She had dreamed of making this place a home, but now it felt like a mausoleum. The sunlight seemed to mock her.

She found herself clutching at her chest. She had wanted escape, a fresh start. Instead, she had inherited a tragedy, a heavy burden she didn't know how to carry. The weight of her new home seemed to press down on her, stealing her breath.

Emotion: miserable

Cluster: Sadness / Despair
PC1 (Valence): -2.19 Negative
PC2 (Disposition): -0.87

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.369
无法0.321
無法0.270
💔0.265
😔0.262
Suppressed:
de-0.490
la-0.329
B-0.300
l-0.292
/-0.224