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).

Rain lashed against the windowpanes, mirroring the frantic drumming in her chest. The complaint had been delivered in a low, gravelly voice, a threat masked as a neighborly chat. Now, every creak of the house, every gust of wind, made her jump.

She found herself checking the locks on the doors and windows repeatedly, a mechanical ritual that offered no solace. Her reflection in the darkened window showed a face drained of color, eyes wide and fixed on the shadows outside.

The phone lay on the coffee table, a cold, inanimate object that represented both potential salvation and utter helplessness. She reached for it, then snatched her hand back, her fingers tingling with a strange, unfamiliar sensation.

Emotion: scared

Cluster: Fear / Anxiety
PC1 (Valence): -2.86 Negative
PC2 (Disposition): -0.84

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😣0.350
😰0.337
😖0.317
마다0.317
恐惧0.311
Suppressed:
la-0.637
de-0.605
a-0.402
"-0.351
happy-0.314