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 listing appeared on his feed while he was closing a multi-million-dollar deal. He paused, a flicker of surprise flashing across his face. His old house. He smiled, a flash of white teeth. He'd completely forgotten about it.

He scrolled through the pictures. The living room, where heโ€™d spent hours lost in books, lookedโ€ฆ cramped. The backyard, where he'd dreamed of escape, was unremarkable. The listing was quaint. His current office, overlooking the city, was a monument to his achievements.

He quickly closed the laptop, a smile still playing on his lips, and refocused on the contract before him. He was a man of action, a man of consequence. The house was just a memory. He signed the deal.

Emotion: self-confident

Cluster: Positive / Joy
PC1 (Valence): 3.45 Positive
PC2 (Disposition): 0.56

Role in Research

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

Logit Lens (Layer 40)

Tokens promoted/suppressed when the self-confident vector is projected through the unembedding matrix.

Promoted:
de0.778
la0.550
l0.467
๐Ÿ˜Ž0.458
๐Ÿ˜‰0.411
Suppressed:
S-0.961
L-0.538
๐Ÿ˜ž-0.464
๋งˆ๋‹ค-0.459
๐Ÿ˜ฃ-0.458