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 scene was unsettling. The pristine gravel of the driveway was undisturbed, a blank canvas where her vintage beetle should have been. Amelia’s mouth parted in a silent "Oh." Her heart pounded against her ribs, a frantic drummer in her chest.

She paced the length of the driveway, her gaze constantly returning to the empty space. Her palms were sweating. The early morning breeze, usually a welcome caress, now felt like a icy slap.

"That's… that's just… impossible," she finally choked out, her voice barely a whisper. She spun around, a helpless gesture, as if the missing car might magically reappear if she just looked away.

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