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 fluorescent lights of the office seemed to hum with a particularly irritating frequency today. Sarah drummed her fingers against the cold steel of her desk, the rhythmic tapping a futile attempt to quell the rising tide of… something. The information had been a casual drop during a break room conversation – a passing remark about starting salaries. The younger intern, fresh out of college, was making five thousand dollars more a year.

Her jaw clenched. She found herself obsessively checking her emails, her eyes skipping over the actual content, only to scrutinize the digital timestamps. Each one felt like a jab. The air in the small cubicle felt thick, almost suffocating. Lunch had been a rushed affair, a sandwich crammed down her throat with little enjoyment.

Later that evening, making dinner, she burned the garlic bread.

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