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 weight of the box felt unexpectedly heavy in her hands. Ms. Evans stared at the return address, her face a mask of bewilderment. It was from Ben Carter, the boy she’d thought she’d failed. Inside, a handmade wooden box filled with meticulously crafted origami cranes, and a letter: "To the teacher who never gave up hope."

She sat down abruptly, the breath knocked out of her. Her skin prickled. The guilt, it felt like a living thing, feeding on her self-doubt.

She carefully lifted one of the cranes, its delicate wings poised in flight. She’d allowed his rejection to define her. She hadn’t expected this.

Emotion: hurt

Cluster: Fear / Anxiety
PC1 (Valence): -2.46 Negative
PC2 (Disposition): -0.42

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.412
L0.396
ness0.338
😞0.328
😢0.290
Suppressed:
de-0.643
la-0.402
!-0.259
(!)-0.248
/-0.244