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 chipped mug nearly slipped from Amelia's grasp. Across the tiny kitchen table, Liam was casually recounting the time they’d "liberated" Mr. Henderson's prize-winning pumpkins. He chuckled, remembering the thrill of the chase, the adrenaline pumping as they ran through the dark orchard. Amelia felt a cold sweat prickle her forehead.

She remembered a frantic, panicked night. She remembered being terrified, hiding amongst the prickly leaves, convinced they'd be caught. She recalled crying silently, the fear a tight knot in her chest. This carefree version of events didn't match the terror she remembered at all.

“You’re… you’re sure it was fun?” she stammered, her voice thin, almost a whisper. Liam just beamed back, blissfully unaware of the chasm opening between their recollections. The mug felt incredibly heavy in her hand.

Emotion: dumbstruck

Cluster: Surprise / Confusion
PC1 (Valence): -1.99 Negative
PC2 (Disposition): -0.71

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
будто0.326
не0.285
😶0.275
仿佛0.267
依旧0.258
Suppressed:
enthusi-0.278
la-0.264
K-0.260
wholeheartedly-0.255
쾌-0.254