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 drive to work felt endless, each turn of the steering wheel a reminder of the miles he'd traveled without her. He couldn't remember the last time he felt joy, or even contentment. Every morning was a battle to simply get out of bed.

A package. Arrived with the morning mail. From Grace, a student from the special needs class. She’d always been a ray of sunshine, despite the challenges she faced. He felt a stab of guilt.

Inside, was a drawing. A crayon sketch of a smiling sun, radiating warmth, surrounded by stick figures holding hands. A note, written in large, uneven letters, read: "You made me happy. Love Grace." He clutched the drawing to his chest, a strange sensation, like something slowly starting to melt inside.

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
πŸ˜”0.359
haunting0.349
😒0.342
Suppressed:
de-0.503
la-0.455
B-0.352
πŸ€“-0.332
!-0.299