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).

Mr. Henderson sat in his worn leather armchair, the fire crackling merrily in the hearth. He’d just finished his evening tea, the steam curling up like a ghostly wisp. He picked up the old photo album, the leather cover cracked and faded. He flipped through the brittle pages, each photograph a portal to another time. Then, a knock at the door.

He opened it to find a delivery man holding a large, flat box. “For you, sir,” he grunted, handing over the package. Inside was a framed painting of the very same oak tree that stood in the school’s front yard, the one he used to point to from his classroom window. A note, penned in elegant cursive, read, "Remembering the best view from class, Mr. Henderson. - Sarah." He felt a tightness in his chest.

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