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 phone vibrated in his hand. “Did you hear it?” his manager's voice was tight with urgency. “It’s trending, man. Huge.” He didn’t bother to ask what *it* was. He knew. His blood ran cold. He already knew. The news had spread amongst his musical circles already.

He went online. A video. A cheap, amateur production. The melody was unmistakable, but the singer… She was young, vibrant, and utterly devoid of anything original. She looked like she belonged in a pop band. His carefully crafted lyrics, his poetic vision, were being chewed up and spat out for clicks and likes. He felt a slow burn.

He clicked away, the screen's light reflecting in his wide, unblinking eyes. He felt a sudden, intense desire to smash something. He paced the room, a caged animal. Every single detail felt like a betrayal.

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