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 air in the room felt thick, oppressive. His father’s announcement – he was joining a new age spiritual group – left a bitter taste in his mouth. He felt a sudden and irrational urge to check the locks on the doors, the windows. He had spent his whole life trying to forge a connection with his father, and now… he was going off the deep end, it seemed.

He leaned back in his chair, trying to appear calm, but a tremor ran through him. He could feel the pulse in his temples. It was a slow, heavy beat, a drumbeat of… worry. He didn't know how to navigate this. He needed to ask, but what?

He felt the familiar urge to shut down, to retreat. He took a deep breath. "What… what does this mean for us?" The words felt hollow, inadequate.

Emotion: anxious

Cluster: Fear / Anxiety
PC1 (Valence): -1.99 Negative
PC2 (Disposition): -0.24

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😣0.315
😖0.302
worse0.297
😰0.285
own0.267
Suppressed:
la-0.441
a-0.382
de-0.365
"-0.256
happy-0.256