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).

"Welcome to the legal team!" The head of the department, a stern woman with a sharp gaze, had greeted David upon his arrival. Now, surrounded by stacks of files and legal jargon, he felt a crushing sense of disinterest.

He looked at the papers. He looked at the computer. He looked at the window. He looked at the floor.

The legal work felt heavy, like a lead weight. He propped his chin on his hands, lost in thought. A long moment passed, and then another.

He closed his eyes. He wished he was somewhere else. He wished he wasn't here.

He stared at a spot on the carpet, trying to focus. He just couldn't seem to get started.

Emotion: lazy

Cluster: Fatigue / Lethargy
PC1 (Valence): -0.37 Negative
PC2 (Disposition): -0.02

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😴0.458
snooze0.425
惰0.401
Lazy0.392
躺0.386
Suppressed:
able-0.371
感激-0.289
because-0.274
å› -0.273
কারণ-0.272