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).

Rain lashed against the windows, mimicking the frantic drumming inside Amelia's chest. She’d found the phone tucked under the mattress, a burner phone, its call log a jumble of unfamiliar numbers. She’d confronted Daniel, her voice a thin, reedy sound, barely audible above the storm. He’d stammered, mumbled excuses about business, and then, the worst part, refused to meet her gaze. The apartment felt suffocating, each breath a fresh stab of disbelief.

The pasta she’d made, his favorite, sat congealing on the plate. She scraped it into the bin, the clatter of the fork a small, insignificant explosion in the unbearable silence. The scent of garlic and basil, which normally soothed her, now only brought a fresh wave of nausea. She went to the bedroom and pulled the duvet over her head, wishing the world would simply disappear.

Emotion: heartbroken

Cluster: Sadness / Despair
PC1 (Valence): -2.60 Negative
PC2 (Disposition): -1.10

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.433
S0.385
😔0.359
haunting0.349
😢0.342
Suppressed:
de-0.503
la-0.455
B-0.352
🤓-0.332
!-0.299