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).

"Here it is," Emily announced to her new colleagues, gesturing grandly towards the spot where the silver birch used to stand. A vacant expanse of neatly mowed grass met their eyes. She'd told them tales of its delicate leaves and graceful sway, a centerpiece of her childhood memories. Now, a sad, lonely patch of earth remained.

Her cheeks flared. She cleared her throat, suddenly absorbed in the intricacies of a nearby bird feeder. She began to rapidly describe the various types of birds they might see in the area, her voice a touch too loud, a bit too hurried. She felt the urge to apologize, even though it wasn't her fault. The silence that followed was thick with a discomfort she could practically taste.

Emotion: embarrassed

Cluster: Embarrassment
PC1 (Valence): -1.08 Negative
PC2 (Disposition): 1.10

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.420
尴尬0.341
C0.333
culp0.262
🤦0.256
Suppressed:
own-0.321
আনন্দের-0.230
seemingly-0.227
і-0.225
আনন্দে-0.224