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 email arrived at 3:00 AM. It simply said, "Meet me at the park. Sunrise." Amelia stared at the screen, heart hammering against her ribs. She felt a profound sense of pressure, an invisible weight pressing down on her. The park was just a few blocks away.

She dragged herself out of bed, the silence of the pre-dawn hours amplifying her anxiety. Her mind was racing. Each step felt heavy and unwilling.

The park's looming shadows stretched out as she approached. She saw him. He was standing by the oak tree they had always discussed. She hesitated, her foot itching to run away. The thought ran through her head, "what if everything changes?".

Emotion: tormented

Cluster: Fear / Anxiety
PC1 (Valence): -3.33 Negative
PC2 (Disposition): -0.70

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
无法0.406
😞0.399
每一次0.378
無法0.369
S0.353
Suppressed:
la-0.803
de-0.755
a-0.407
B-0.404
l-0.354