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 warm sunlight dappled through the leaves, painting shifting patterns on Amelia's face. She sat on the park bench, a book open on her lap, but her gaze drifted upwards, following the flight of a robin. A gentle breeze rustled the pages, carrying the scent of freshly cut grass. She inhaled deeply, a slow, deliberate breath that filled her lungs and calmed her. A small smile played on her lips as she watched the bird disappear into the branches.

A cough startled her. A man, maybe a little older than her, with kind eyes and a hesitant smile, was standing nearby. He held a crumpled paper in his hand. “Excuse me,” he began, his voice soft, “but do you, by any chance, have a…a rash, of sorts?”

Amelia frowned, touching her wrist. A faint, almost imperceptible discoloration was there. She nodded, intrigued, and asked him what he'd seen.

Emotion: serene

Cluster: Calm / Serenity
PC1 (Valence): 2.59 Positive
PC2 (Disposition): -2.73

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
disfrut0.469
enjoyed0.430
enjoy0.419
serene0.411
楽し0.410
Suppressed:
S-0.611
L-0.502
de-0.478
H-0.455
worse-0.424