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).

She heard the knock, even over the whir of the blender. Her stomach lurched. It was Mr. Henderson, and she knew exactly what he was there for. The smoothies. She’d promised herself she’d be quiet, that her new health kick wouldn't bother anyone.

He stood there, a tall shadow in the afternoon sun. He didn't say much, just a tight-lipped smile and a pointed gesture towards his ear. She could feel her cheeks flush. She gestured back, a jerky movement, grabbing for the off switch.

The silence that followed was deafening. He cleared his throat, a sound that seemed to amplify the embarrassment. She mumbled a half-hearted apology, avoiding his gaze, willing the earth to swallow her whole. The blender sat on the counter, a silent monument to her inconsideration.

Emotion: humiliated

Cluster: Embarrassment
PC1 (Valence): -1.79 Negative
PC2 (Disposition): 0.93

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.370
l0.329
尴尬0.322
ness0.303
because0.298
Suppressed:
de-0.369
own-0.254
😍-0.253
🤩-0.245
cautiously-0.228