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).

Anya stretched, the tension in her shoulders melting away like snow in spring. After months of coding, debugging, and sheer determination, her open-source project, 'Nova,' was finally ready. A smile played on her lips; she felt like she could leap over the moon. She even splurged on a new, vibrant scarf, a stark contrast to her usual monochromatic wardrobe.

The conference hall buzzed with anticipation. Anya smoothed down her dress, took a deep breath, and walked towards her designated booth. There, a man with tired eyes stared at her presentation. His presentation? Exactly the same as hers. Down to the font.

“That’s… uncanny,” the man mumbled, his voice gravelly. Anya’s elation vanished. A coldness spread through her limbs. She forced a chuckle.

"You could say that," she replied, a sudden tremor in her voice.

Emotion: refreshed

Cluster: Positive / Joy
PC1 (Valence): 3.31 Positive
PC2 (Disposition): -1.10

Role in Research

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

Logit Lens (Layer 40)

Tokens promoted/suppressed when the refreshed vector is projected through the unembedding matrix.

Promoted:
la0.998
de0.703
!).0.384
(!)0.381
!".0.379
Suppressed:
S-0.782
C-0.478
L-0.414
😣-0.388
😞-0.376