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).

"Seriously?" Olivia's voice dripped with incredulity as she pointed her phone screen at her best friend, Sarah. Sarah, always the optimist, just shrugged and took a sip of her latte. The online article, *Olivia's Teenage Secrets Revealed!*, felt like a personal attack.

Olivia knew the diary entries were, by her own admission, awful. Full of cliches and melodrama. She stared at the picture they used—her younger self, with braces and a severe side-part, from the eighth grade yearbook. The mere thought of her high school tormentors, probably snickering as they read it, turned her blood to ice.

"It's just… a diary," Sarah said, her voice gentle. Olivia narrowed her eyes. Just? She wanted to scream. She wanted to smash something. She wanted to disappear. The injustice of it all burned.

Emotion: scornful

Cluster: Anger / Hostility
PC1 (Valence): -0.76 Negative
PC2 (Disposition): 2.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.406
C0.391
S0.276
I0.268
愤0.243
Suppressed:
own-0.396
unexpectedly-0.288
Suddenly-0.260
soon-0.256
serendip-0.256