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 sun was setting, casting long shadows across the living room as Clara examined the photos. She'd been told her great-aunt, Evelyn, was a prim and proper woman, never one for adventure. The stories were always the same: Evelyn, a librarian, always dressed in beige, lived a long and orderly life.

But here, in these faded images, was a different Evelyn. Evelyn on a motorcycle, wind whipping through her hair, a wide grin splitting her face. Evelyn in a brightly colored dress, flirting with a handsome man at a bustling street market. The gap between expectation and reality left a cold emptiness inside Clara.

She felt a hollowness in her chest, a profound sense ofโ€ฆ being let down. The air in the room felt thick, heavy with unspoken truths and concealed lives. She rose slowly, the photographs suddenly seeming to have lost their appeal.

Emotion: melancholy

Cluster: Sadness / Despair
PC1 (Valence): -1.78 Negative
PC2 (Disposition): -1.58

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.445
๐Ÿ˜ž0.376
๐Ÿ˜”0.337
ใ„ใคใ‚‚0.327
sadness0.312
Suppressed:
de-0.739
la-0.548
l-0.497
/-0.314
!-0.290