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 air in the garage was stale and close. David ran a hand through his hair, his brow furrowed in concentration. He was hunting for his old baseball glove, not this. He opened the first box. "David - Middle School." He stopped. He stared.

The meticulously preserved assignments felt like a punch to the gut. The carefully graded tests, the hand-written essays. He felt a churning in his stomach, a sensation of disorientation that was almost physical. He ran a hand over his mouth, as if to physically restrain himself.

He pulled out a science project he'd been particularly proud of. He stared at it, the memories flooding back. He felt a powerful sense of unease, a feeling of being watched, even now. He wondered why. What was the point? He sat down on a dusty milk crate, suddenly feeling very tired.

Emotion: shocked

Cluster: Surprise / Confusion
PC1 (Valence): -2.19 Negative
PC2 (Disposition): -0.65

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
будто0.298
😰0.268
sickening0.266
не0.266
que0.257
Suppressed:
enthusi-0.283
开心-0.269
ولة-0.262
ارك-0.257
উৎসাহিত-0.255