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 photograph, faded and creased, showed us laughing in the sun, holding fishing poles. The lake shimmered behind us, a tranquil expanse of blue. "The boat..." David began, his voice raspy. "It capsized, didn't it?"

My chest tightened, a cold weight settling within. Of course, the boat had capsized. I had forgotten. He looked at me; I looked at the photograph. The water had been ice cold.

I focused on the chipped paint on the table as I spoke. The calm water had not remained. I remembered the panic, the struggle. The photograph was a trick. A lie. It was a memory that had been buried. And now, I was back in the water.

Emotion: terrified

Cluster: Fear / Anxiety
PC1 (Valence): -2.97 Negative
PC2 (Disposition): -0.62

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😣0.357
worse0.350
无法0.334
😰0.333
sickening0.322
Suppressed:
la-0.676
a-0.392
de-0.371
"-0.344
happy-0.309