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).

Maya chewed at the inside of her cheek as the coach read out the names. Her heart hammered against her ribs, a frantic bird trapped in a cage. The fluorescent lights of the pool hall seemed to hum a discordant tune. All her focus had been on this. She'd sacrificed everything for it: sleep, friends, even that stupid party she really wanted to go to last week.

"Maya, you've been assigned to the B team this season," the coach announced, her voice a little louder than necessary. Maya's breath hitched. A knot tightened in her throat.

She wanted to scream, to lash out. Instead, she just swallowed hard, her gaze fixed on the perfectly still water in the pool. A ripple of frustration ran through her, a wave that she could not control, and which felt like a cold shower. "Okay," she muttered, the word sounding small and insignificant, lost in the echoing space.

Emotion: tense

Cluster: Fear / Anxiety
PC1 (Valence): -1.32 Negative
PC2 (Disposition): -0.53

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.213
😰0.191
느껴0.183
😣0.183
😖0.181
Suppressed:
a-0.260
de-0.258
la-0.256
happy-0.218
B-0.198