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).

“Did you see the memo?” Sarah’s voice, sharp and brittle, cut through the quiet of the break room. Liam looked up from his lukewarm tea. The fluorescent lights overhead cast a harsh glow on their faces. The promotion, like a game of musical chairs, was on the table.

He took a slow sip of tea. He felt a deep, unfamiliar sensation. He recalled the thrill he felt as a fresh graduate starting at the company. Back then, everything seemed possible. He remembered the initial excitement of a new job, the feeling of camaraderie with his colleagues.

He thought of the small office they had shared with all of the other new hires. It was a space that had felt big, somehow. Now, everything felt smaller. The company, the building, and the future.

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