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 parking lot was unusually empty, the setting sun casting long, distorted shadows. I walked towards my car, the key trembling in my hand. Coach Johnson's words, "We're going in a different direction," were still ringing in my ears. I felt like the ground beneath my feet was shifting. I’d given my all, every practice, every game. What had I done wrong? What did "different direction" even mean?

I saw my reflection in my car window and I didn't even recognize myself. I looked pale.

My head swam with questions, each one a sharp jab, a painful reminder of the dream that was now slipping away. The car keys slipped from my grasp, falling onto the pavement with a metallic clang. I bent, to pick them up, and for a moment, the world spun.

Emotion: bewildered

Cluster: Surprise / Confusion
PC1 (Valence): -1.87 Negative
PC2 (Disposition): -0.48

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
будто0.300
ness0.255
像是0.241
worse0.231
Worse0.219
Suppressed:
la-0.373
மகி-0.229
happy-0.226
ecstatic-0.226
K-0.219