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 hadn't told anyone about the essay. He’d poured his heart out, writing about his struggles with anxiety and depression. Now, it was published, and he felt exposed. He was surprised how much it affected him.

He was scheduled to give a presentation in class the following day, and he realized he couldn’t do it.

He spent the night pacing in his room, unable to sit still. He couldn't concentrate. He felt the weight of the world pressing down on him. He considered dropping out of the class. The thought of facing the judgmental stares of his classmates was unbearable. He felt cornered and panicked. He texted his professor he was too sick to come.

Emotion: vulnerable

Cluster: Fear / Anxiety
PC1 (Valence): -2.07 Negative
PC2 (Disposition): -0.69

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.327
ness0.319
😞0.299
😔0.277
😣0.258
Suppressed:
de-0.567
la-0.488
l-0.271
(!)-0.267
(!)-0.257