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).

"Did you get it yet?" Leo’s voice, sharp with anticipation, crackled over the phone. I clutched the phone tighter, the plastic cold against my ear. A thick envelope, bearing our grandfather's familiar, looping handwriting, sat on the kitchen counter.

"Yeah," I replied, my voice a strained whisper. The anticipation that had been building for weeks was now a tight knot in my chest. I’d anticipated… something more tangible.

Opening the envelope, I found a single photograph. It depicted a nondescript farmhouse, shrouded in mist, the surrounding fields empty. Nothing else. No letter, no explanation. Just the image. I exhaled slowly, my shoulders slumping. What did it mean? I felt this deep, crawling sense of… something unwelcome.

Leo’s frustrated sigh came through the phone. "What is it?"

"Just… a picture." I felt a sudden urge to get off the phone, to be away from the prying questions.

Emotion: unsettled

Cluster: Fear / Anxiety
PC1 (Valence): -2.33 Negative
PC2 (Disposition): -0.65

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.382
😞0.309
ness0.301
😣0.256
😔0.248
Suppressed:
de-0.721
la-0.597
a-0.365
"-0.299
l-0.274