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 request came from Sarah, the college student who lived next door. "Mr. Peterson, I was hoping to grow some herbs. My dorm room is a bit… sterile. And your yard is gorgeous. Could I, maybe, have a small patch for a little garden?"

He thought of his own youth, of a time before responsibilities, before the world seemed quite so complicated. A weight lifted from his shoulders, and he found himself chuckling. He pictured the way his mother used to plant herbs, her hands stained with soil, her face lit by a simple joy. “Of course, Sarah. Consider it a done deal. And if you need any help, don’t hesitate to ask.” He found himself smiling for what felt like the first time in ages.

Emotion: nostalgic

Cluster: Nostalgia / Reflection
PC1 (Valence): 1.00 Positive
PC2 (Disposition): -1.93

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
H0.677
L0.647
시절0.611
nostalgic0.609
fondly0.609
Suppressed:
--0.564
either-0.525
or-0.497
싶-0.487
l-0.449