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 coffee tasted like ash. Ethan slammed the mug down on the counter, the ceramic clattering. Heโ€™d barely slept in days, the image of his wife, Sarah, disappearing into the garage with her phone, repeating itself in his mind. Heโ€™d seen her glance at him, a flicker of something unreadable in her eyes, before sheโ€™d quickly ended the call. His hands trembled as he reread the call logs on his phone, the numbers like accusatory fingers pointing at her. He felt a gnawing unease spread through his core.

Emotion: stressed

Cluster: Fear / Anxiety
PC1 (Valence): -1.37 Negative
PC2 (Disposition): -0.53

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
๐Ÿ˜ฉ0.266
๐Ÿ˜ซ0.256
๐Ÿ˜ฐ0.239
๐Ÿ˜–0.237
๐Ÿ˜ž0.236
Suppressed:
B-0.200
happy-0.197
enjoy-0.176
enjoying-0.172
bahagia-0.169