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 hum of the espresso machine filled the small cafe, a stark contrast to the booming music and flashing lights echoing from the club across the street. "Thanks, Alice," I mumbled, accepting the coffee. Today, on my birthday, my best friend, Jessica, was hosting a dance party. I just wanted a quiet morning.

I took a sip of the bitter liquid, letting the heat soothe the chill that had settled in my bones. I watched the people pass by, their faces alight with a kind of joy I couldn't seem to access. I felt disconnected, as if I were observing the world through a thick pane of glass.

I clutched the cup tightly, the porcelain a small comfort. The thought of the loud music, the crowded dance floor, the forced conversations filled me with a deep sense of dread. I felt the familiar tightening in my chest, the familiar urge to retreat, to disappear. I put down the cup. I didn't want to go.

Emotion: brooding

Cluster: Sadness / Despair
PC1 (Valence): -2.46 Negative
PC2 (Disposition): -0.60

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.749
😞0.329
그의0.299
un0.273
但他0.270
Suppressed:
de-0.917
la-0.603
a-0.369
l-0.335
her-0.330