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 scent of lavender and mothballs clung to everything in the house, a constant reminder of Evelyn’s presence and her expectations. I, Daniel, felt like a trespasser, moving through the rooms that were once filled with such vibrant life. The space felt too big, too empty.

I'd always been so busy, so self-absorbed. The calls to my grandmother were few and far between, the visits even rarer. I’d offered empty promises, excuses. Now, surrounded by her belongings, I felt a knot in my stomach.

My sister, Sarah, was methodically going through the kitchen drawers, pulling out old utensils and chipped plates. "Think we should keep this?" she asked, holding up a tarnished silver spoon.

I couldn't meet her eyes. My voice was a choked whisper, "Whatever you think."

Emotion: ashamed

Cluster: Shame / Guilt
PC1 (Valence): -1.97 Negative
PC2 (Disposition): 1.06

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.376
ashamed0.344
ness0.327
逃0.297
😞0.295
Suppressed:
de-0.539
que-0.300
😍-0.288
(!)-0.269
🤩-0.242