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).

He hummed as he unpacked, a little melody that hadn't escaped his lips in months. The spare room, once his childhood haven, felt strangely familiar, yet different. The posters, faded but still present, seemed to grin back at him. He grinned back.

His fingers traced the edge of the desk, worn smooth from years of homework and late-night gaming. He pictured himself here again, older, wiser, but also… relieved. This was a restart button, a chance to get things right.

A wave of warmth flushed through him as he heard his mother’s voice calling him for dinner. He practically skipped down the stairs. The aroma of roasted chicken and mashed potatoes filled the house, a sensory symphony that made his stomach rumble.

Emotion: joyful

Cluster: Positive / Joy
PC1 (Valence): 3.96 Positive
PC2 (Disposition): -0.67

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.615
de0.600
!0.572
excitedly0.552
enthusi0.533
Suppressed:
S-0.893
ness-0.513
😞-0.470
😣-0.337
L-0.334