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 air around me seemed to shimmer. Everything else faded away. It was Rupert. My stomach did a flip. I felt a pressure in my chest. He was there, perched atop a table laden with dusty trinkets. The years had not been kind; his fur was patchy, his one ear flopped, but there was no mistaking him.

I went rigid. The seller, a young girl with bright pink hair, looked up at me expectantly. "He's five dollars," she said with a shrug. I couldn't tear my eyes away from him.

"I'll take him," I managed, my voice raspy. I fumbled for my wallet, my hands trembling. As I walked away, holding Rupert close, I felt tears prick my eyes. It was like a piece of my heart had been returned.

Emotion: awestruck

Cluster: Surprise / Confusion
PC1 (Valence): -0.17 Negative
PC2 (Disposition): -1.20

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
L0.271
suddenly0.260
stranger0.252
plötzlich0.246
wonder0.242
Suppressed:
la-0.450
싶-0.350
K-0.319
rest-0.248
싶은-0.238