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).

“Well, this is a… predicament,” Miles heard the radiologist say, his voice thick with condescension. The scan hung on the screen, a bizarre swirl of color he didn’t understand. Miles felt a wave of heat wash over him, a burning in his chest. “A very… unique situation.” It was the way the man said “unique,” like it was a punchline.

Miles hadn't expected this meeting to go well, but the doctor's lack of tact still took him aback. He’d spent years feeling like an outcast, a walking anomaly, and here was someone confirming that, but with a palpable sense of disdain. The man tilted his head as if he were studying a painting he didn’t like. Miles clenched his fists.

“And?” Miles asked, his voice strained. The doctor just stared back with a placid expression. It felt like he was being mocked, reduced to a problem to be solved, not a person. Miles bit the inside of his cheek, a sharp, angry sting.

Emotion: insulted

Cluster: Anger / Hostility
PC1 (Valence): -0.98 Negative
PC2 (Disposition): 1.46

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.354
骂0.277
愤0.267
C0.237
🤬0.233
Suppressed:
own-0.325
давно-0.229
adventurous-0.223
soon-0.219
optimistic-0.217