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).

Rain lashed against the window as Olivia stared at the photograph album. The images were a portal to a different time, a simpler existence. A picture of her and Chloe, their faces smeared with ice cream, brought a soft smile to her lips.

They were at the seaside, a trip they made as children. The sun had shone, they had built sandcastles, and laughed until their sides ached. It was a perfect day, etched into Olivia's mind as a beacon of joy.

Chloe had called last week, a catch-up chat about their lives. The conversation shifted, as it often did, to the past. Chloe talked about how *she* had bravely rescued a lost baby crab from the tide. Olivia had no memory of any rescue. In her recollection, it was *she* who had bravely waded into the waves.

A sudden chill ran through her, despite the warmth of the room. The perfect day suddenly felt... complicated.

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