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 grimy bell above the pawn shop door jingled as Amelia entered. A wave of stale cigarette smoke and the scent of old leather hit her immediately. She pulled her scarf tighter, a gesture more of defense than warmth. This place... it felt like a judgment.

The clerk, a man whose face seemed permanently etched with boredom, barely glanced up. “Looking for something?” he mumbled, his gaze fixed on the screen of a flickering TV.

Amelia swallowed. “I… I think I saw something in the window. A ring. A… diamond ring.” Her voice cracked slightly, the sound lost in the cacophony of the shop. She pointed towards the display case, a trembling finger betraying her carefully constructed composure. The thought of what she was doing twisted in her gut, a familiar ache.

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