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 house felt too big, the silence echoing in the empty rooms. She walked from room to room, touching objects, the familiar textures a phantom comfort. She’d been so thrilled for Olivia, that new life in Colorado, that adventure, but now…

The scent of Olivia’s perfume lingered in the air, a cruel reminder. She picked up a photograph of them, a silly picture from their college days, and stared at it. A single tear traced a path down her cheek, and she quickly wiped it away.

She sat on the sofa, clutching a cushion. She felt a profound sense of isolation, a gnawing hollowness that threatened to consume her. She closed her eyes, trying to block out the future, the empty space Olivia’s departure would leave.

Emotion: disturbed

Cluster: Fear / Anxiety
PC1 (Valence): -2.97 Negative
PC2 (Disposition): -0.60

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.318
ness0.309
无法0.300
😞0.297
worse0.284
Suppressed:
de-0.829
la-0.666
a-0.392
B-0.301
"-0.290