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 flyer arrived in the mail, promoting the community theater's gala. It featured a list of sponsors, and there, at the top, was "The Miller Foundation." It took Clara a moment. Miller. The Millers. Her neighbors.

She looked out her window, at their modest, yet well-kept house. The Millers, who always brought her a pie on Thanksgiving. Who always seemed so… ordinary.

She read the flyer again. The Miller Foundation was underwriting the entire event. A sudden chill ran down her spine. They were funding scholarships, art programs, and community outreach. Her gaze drifted back to her own, humble living room. A sense of profound disorientation settled over her. How had she missed it?

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