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).

From the moment she walked into the cafe, everything had shifted. The smell of fresh coffee had become a dull background note. Before her, displayed in the glass counter, sat the "Dapper Dan's Delight." It was her cake. The one she had carefully entered into the local bake-off. The one she had lost with.

Her breath hitched in her chest. A wave of dizziness washed over her. The room began to spin. She felt a strange tingling in her fingertips. Her fingers twitched.

She pointed, her voice a fragile whisper. "Excuse me..." She barely recognized her own voice. The barista, with a beaming smile, turned to her. "Yes, miss?"

"That… that cake…" she managed, the words catching in her throat.

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