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 new guy, Ben, was clearly eager. He practically vibrated with ambition, his questions rapid-fire. "And the accounts receivable process?" Ben asked. "Can you walk me through that?"

I just shrugged. I wanted to just disappear. The fluorescent lights overhead seemed to blur, and I had to blink a few times to get my vision back. The office chair felt soft, too soft, and I sank into it, my spine melting against the backrest.

"It's pretty straightforward," I finally said, the words barely audible. I pulled up the spreadsheet, the numbers swam before my eyes. I felt a sense of… letting go, of release. I could show him the steps, explain the formulas. He could take over. I didn't care.

Emotion: docile

Cluster: Passivity
PC1 (Valence): 0.25 Positive
PC2 (Disposition): -1.34

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
ค่อย0.311
淡淡0.302
enjoy0.273
enjoyed0.273
สบาย0.272
Suppressed:
la-0.589
de-0.413
l-0.373
!!-0.344
!!!-0.316