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 humid air of the pawn shop clung to Elias like a second skin. He swatted at a fly buzzing around a chipped porcelain doll, its vacant eyes mirroring the emptiness he felt. He hated this place, the smell of dust and regret, the way the cheap fluorescent lights cast everything in a sickly yellow glow.

The man behind the counter, a fleshy individual with a stained white t-shirt, barely acknowledged his presence. "Looking for something?" he mumbled, fingers stained with nicotine. Elias just glared.

He saw it then, tucked away in a velvet-lined display case, amidst a collection of tarnished watches and broken chains. A familiar glint. The ring. His grandmother’s engagement ring. He took a long, slow breath, the air thick with the smell of old metal. It was a damn tragedy, really.

Emotion: contemptuous

Cluster: Anger / Hostility
PC1 (Valence): -0.40 Negative
PC2 (Disposition): 2.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.502
l0.423
fact0.287
either0.265
👎0.262
Suppressed:
own-0.367
unexpectedly-0.303
serendip-0.265
Suddenly-0.253
previously-0.250