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 library’s hushed atmosphere offered no solace. Amelia sat hunched over a book, but the words swam before her eyes. She couldn’t concentrate. The picnic. The shared picnic. It should have been a perfect day, basking in the sun. Instead, a cloud of confusion, a bitter taste, had settled between them.

She ran a finger along the worn cover, the texture a comfort. She glanced down at her own reflection in the pages of the open book, seeing the tired lines etched around her eyes. She remembered the tension, their clashing interpretations of a simple afternoon, the way his voice had sounded when he spoke of her perception.

She sighed, the sound echoing in the silent space. How could such a joyous occasion have soured so quickly? The question remained, unanswered, an unwelcome guest in her thoughts.

Emotion: brooding

Cluster: Sadness / Despair
PC1 (Valence): -2.46 Negative
PC2 (Disposition): -0.60

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.749
😞0.329
그의0.299
un0.273
但他0.270
Suppressed:
de-0.917
la-0.603
a-0.369
l-0.335
her-0.330