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 phone slipped from Elias's sweaty palm, clattering against the tiled floor of the bathroom. He snatched it up, his breath catching in his throat. He'd been meticulously crafting a text to his boss, explaining why he was late. Now, a rogue tap had sent it to… Sarah? Sarah from accounting, who he barely knew. A wave of heat washed over him.

He quickly grabbed his phone, but it was too late. The dreaded double-blue ticks were already there, mocking him. He pictured Sarah’s reaction, the slight furrow of her brow, the way she tilted her head when confused. He should have used a draft. This was his third tardiness in a month.

He ran a hand through his hair, dislodging strands. He could feel his heartbeat thudding in his ears. Should he call? Text back immediately and apologise? Maybe feign ignorance, claiming a phantom finger? The possibilities tangled in his brain like a knotted ball of yarn.

Emotion: anxious

Cluster: Fear / Anxiety
PC1 (Valence): -1.99 Negative
PC2 (Disposition): -0.24

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😣0.315
😖0.302
worse0.297
😰0.285
own0.267
Suppressed:
la-0.441
a-0.382
de-0.365
"-0.256
happy-0.256