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).

A bead of sweat trickled down his temple, and he resisted the urge to swipe it away. The room seemed to tilt, the pastel decorations blurring at the edges. *Him?* Give a speech? About his father, the stoic, reserved man who'd always been more comfortable with numbers than words? It felt like an elaborate prank. The realization hit him, a jolt that went through his whole body. He gripped the podium, knuckles white, the wood surprisingly cold against his trembling fingers.

He saw the smile on his father's face, a rare and precious sight, and a lump formed in his throat. He opened his mouth, but only a strangled sound emerged. He tried again.

Emotion: astonished

Cluster: Surprise / Confusion
PC1 (Valence): -0.84 Negative
PC2 (Disposition): -0.24

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
que0.448
de0.350
la0.323
l0.307
incred0.266
Suppressed:
own-0.289
cheerfully-0.231
enjoyed-0.224
开心-0.221
ี่ยว-0.219