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 windows, mirroring the tempest inside Amelia. She stared at the email, the words blurring despite her best efforts to focus. A half-sister. Twenty-eight years old, living just a few towns over. A whole life, a whole person, that her father had conveniently omitted.

Amelia slammed the laptop shut, the metallic clang echoing in the suddenly silent apartment. She picked up a ceramic mug, the warmth of the tea she'd brewed hours ago now long gone. It felt heavy in her hand, a weight she suddenly felt inside too. She threw it in the sink, the shatter a sharp, brittle sound.

She paced the length of her small living room, her breath coming in ragged gasps. Every framed photograph, every knick-knack, every carefully chosen piece of furniture suddenly felt like a lie, a carefully constructed facade built on a foundation of deceit.

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