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 scent of burgers wafted from the grill, the sounds of summer laughter filled the air. She wanted to join in, to be happy. But the bear, lying amongst a pile of unwanted items, stole all her joy. The familiar shape, the threadbare fur, each detail a stab of regret.

Sheโ€™d left the bear behind, along with everything else. The broken promises, the shattered dreams. Here, in the harsh light of a yard sale, it felt like an undeniable truth.

She reached for the bear, her fingers brushing against the worn fabric. A wave of nausea washed over her. She bought it, then walked away, the weight in her chest crushing.

Emotion: guilty

Cluster: Shame / Guilt
PC1 (Valence): -1.62 Negative
PC2 (Disposition): 1.23

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.492
L0.398
ํƒ“0.354
been0.323
ashamed0.319
Suppressed:
de-0.400
(!)-0.279
la-0.276
๐Ÿ˜-0.274
"!-0.264