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).

Rain lashed against the windows, mirroring the tempest within Amelia. She clutched the offending book, its glossy cover a taunt. Her life, laid bare for public consumption. She slumped onto the sofa, the plush cushions offering no comfort. A tremor ran through her, leaving her hands icy. She had always prized her privacy, meticulously guarding the small intimacies of her life. Now, they were merchandise.

The biography, “Amelia: A Life Unveiled,” felt heavy in her lap. Her throat constricted, making it difficult to breathe. She picked at a loose thread on the sofa. Each chapter heading – “Childhood’s Shadows,” “The Lonely Years,” “Unfulfilled Dreams” – felt like another punch. The air in the room felt thick, oppressive.

A single tear escaped and trailed down her cheek. She scrubbed at it, the motion frantic. This violation – it was unbearable.

Emotion: sad

Cluster: Sadness / Despair
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -1.50

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.436
😔0.405
S0.384
无法0.341
L0.337
Suppressed:
de-0.749
la-0.614
l-0.471
/-0.350
!-0.322