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).

He stared out the window, the neighbor’s shadow already falling across the precious strip of sunlight on his property. Mark felt a prickly heat rise in his chest. His breathing became shallow, each inhale a struggle. The garden, his pride, his escape, now threatened. The idea of sharing it, of someone else’s presence disrupting the carefully constructed order, felt like a violation.

“Just a little space,” the neighbor had said, her voice like wind chimes – irritatingly cheerful. He’d plastered on a smile, a mask he wore with increasing difficulty. Inside, his thoughts were a frantic scramble, each one a sharp, jagged edge. He found himself picking at the skin around his thumb, the sting a small, insignificant distraction from the overwhelming pressure.

Emotion: tormented

Cluster: Fear / Anxiety
PC1 (Valence): -3.33 Negative
PC2 (Disposition): -0.70

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
无法0.406
😞0.399
每一次0.378
無法0.369
S0.353
Suppressed:
la-0.803
de-0.755
a-0.407
B-0.404
l-0.354