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).

Anya stirred her tea, the clinking of the spoon against the ceramic mug a rhythmic sound that soothed her. The aroma of Earl Grey filled her small kitchen, a familiar comfort. Sunlight streamed in through the window, illuminating dust motes dancing in the air. She felt utterly at home.

Her inbox, however, presented a stark contrast. The email from her professor contained a single sentence: "Your programming project earned a D-." Her hand trembled, sending a small splash of tea onto the counter. A strange tightening seized her stomach.

She stared at the screen, the cheerful colours of the website now seeming garish and mocking. A wave of cold washed over her, and she felt a tremor pass through her.

Emotion: peaceful

Cluster: Calm / Serenity
PC1 (Valence): 2.40 Positive
PC2 (Disposition): -2.79

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
прият0.408
disfrut0.407
gently0.390
affectionately0.379
thoughtful0.376
Suppressed:
de-0.541
S-0.493
L-0.396
'-0.388
worse-0.367