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 silence amplified the anxiety that had been building inside him since the move. Michael found himself jumpy, easily startled by every creak and groan of the house. He felt a disquiet he couldn't name, a constant, low-level hum of apprehension. The house's history felt heavy on his shoulders.

The feeling intensified when he was inspecting the floorboards in the dining room, noticing a slight unevenness. A loose board. He pried it up, revealing a dark hole beneath. His heart pounded in his chest, a frantic rhythm against his ribs.

He peered into the darkness, his breath catching in his throat. The smell of dust and decay filled his nostrils, and a cold draft brushed his face. He quickly slammed the board back into place, but the image of what he'd seen remained, a dark stain on his mind.

Emotion: worried

Cluster: Fear / Anxiety
PC1 (Valence): -2.02 Negative
PC2 (Disposition): -0.43

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
worse0.285
S0.279
😰0.277
😣0.269
Worse0.250
Suppressed:
la-0.476
de-0.432
🎉-0.233
happy-0.231
"-0.228