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 sunlight slicing through the dusty windows of the old Victorian house felt too bright, irritating. I ran a hand through my hair, pushing back the strands that kept falling into my face. This was supposed to be a bonding experience, a way to remember Grandma. Instead, I just felt... weighed down.

My brother, Daniel, was already inside, poring over documents, a strange, almost manic energy about him. His face was lit, excited. I stood in the doorway and watched him, feeling a pang of… something. Resentment? Jealousy? I didn't know.

The lawyer had explained the split: “Two equal portions,” he’d said. But the things Daniel found—a vintage car, a valuable stamp collection—felt a world away from my meager share: a handful of faded letters, a chipped teacup, and a stack of recipes. I let out a breath and went back outside.

Emotion: sullen

Cluster: Sadness / Despair
PC1 (Valence): -1.51 Negative
PC2 (Disposition): -0.22

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.404
😞0.282
’0.254
:(0.233
无法0.215
Suppressed:
de-0.539
!-0.276
la-0.263
🤩-0.225
que-0.216