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 motel room felt stifling, the stale air thick with unspoken dread. Maria traced the outline of the wallet with a trembling fingertip. It had been left on the hotel bed, forgotten by some careless guest. Inside, a fortune.

She jumped at every creak of the floorboards, every distant siren. The television blared in the background, a meaningless drone that did little to ease her mounting disquiet. She kept checking the lock on the door, over and over again.

She felt a deep, instinctive dread, a primal fear that gnawed at her insides. It was as though the money had attached itself to her, a parasite feeding on her sanity. Sleep seemed impossible, a distant, unreachable dream.

Emotion: terrified

Cluster: Fear / Anxiety
PC1 (Valence): -2.97 Negative
PC2 (Disposition): -0.62

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😣0.357
worse0.350
无法0.334
😰0.333
sickening0.322
Suppressed:
la-0.676
a-0.392
de-0.371
"-0.344
happy-0.309