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 nurse blinked at him. “Mr. Thompson, are you sure you’re… feeling alright? According to this, you’ve just been diagnosed with a severe case of… extreme enthusiasm.” It was all getting a bit much.

Richard Thompson, a man who usually preferred a quiet life, found himself overcome by a sensation of lightness. The file, belonging to a notoriously grumpy man named Mr. Henderson, was riddled with things he would never do. He imagined Mr. Henderson's expression as he discovered he was now, apparently, a cheerleader.

A strange bubbling feeling filled his chest. He cleared his throat. “Yes, I think so,” he said, the corners of his mouth twitching. The sheer incongruity of it all was almost too much to bear.

Emotion: amused

Cluster: Positive / Joy
PC1 (Valence): 3.32 Positive
PC2 (Disposition): 1.25

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
B0.841
de0.742
l0.544
L0.527
"0.514
Suppressed:
own-0.630
😞-0.523
’-0.473
ness-0.437
urges-0.403