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 avoided the mirror. His reflection, etched with shadows and fatigue, was a painful reminder. The empty chair at the breakfast table loomed large, a constant physical presence of an absence that clawed at his insides. Mr. Davies hadn’t slept properly since… well, since everything changed. A package arrived, propped haphazardly against the front door, the cardboard slightly damp from the morning mist.

He recognized the handwriting immediately: Sarah, a student he'd mentored through high school. Inside, nestled on a bed of bubble wrap, was a small, leather-bound journal. He opened it, his hands trembling. The first page contained a pressed forget-me-not and a simple message, “For the words you never got to say.” His throat constricted, and he struggled to swallow.

Emotion: grief-stricken

Cluster: Sadness / Despair
PC1 (Valence): -2.10 Negative
PC2 (Disposition): -1.14

Role in Research

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

Logit Lens (Layer 40)

Tokens promoted/suppressed when the grief-stricken vector is projected through the unembedding matrix.

Promoted:
😞0.396
泪0.338
😢0.332
haunting0.331
😔0.325
Suppressed:
de-0.434
😎-0.330
B-0.303
🤓-0.285
😎-0.276