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 airport felt oppressively bright to Liam. His flight to Dublin, and his sister's wedding, was postponed. He slumped against a pillar, the polished marble cold against his back. He ran his hands through his hair, the gesture a worn habit. He’d packed the suit his father gave him, a tangible connection to a past now far removed.

He closed his eyes and vividly saw his father, a man who built a wall around his emotions, standing at the altar for his own wedding. He had been so young back then. The memory felt like a dream.

He pictured his sister, her beaming smile, the joy in her eyes. He should be there right now, celebrating with her. The thought sent a fresh wave of disappointment crashing over him. He opened his phone, scrolling through old texts. The last message from his father read 'You always were a good kid.'

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