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 stadium lights blazed down on him, illuminating the coach's face as he approached. Marcus, usually a cornerback, braced himself for the usual criticisms about his tackling technique. Instead, the coach placed a hand on his shoulder, a surprisingly supportive gesture. “Marcus, we want you to try linebacker.”

A jolt of pure energy went through Marcus. It was an opportunity to be a primary defensive force, a chance to be the heart of the defense. A grin spread across his face, seemingly of its own accord. His shoulders straightened, and he took a deep breath, savoring the crisp night air.

The other players gave him a wide range of welcoming responses. He felt a newfound confidence, a sense that he could make a real impact. He could hardly wait for the next practice.

Emotion: pleased

Cluster: Positive / Joy
PC1 (Valence): 3.62 Positive
PC2 (Disposition): -0.28

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.728
la0.496
happy0.462
🤩0.433
🥳0.430
Suppressed:
S-0.486
😞-0.481
ness-0.454
😣-0.431
worse-0.401