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 crackled with unspoken disappointment. The coach's pronouncements had been clear, decisive. Tim, a promising young player, had been one of the unlucky ones. He felt his face flush as the coach delivered the verdict.

He barely looked up, pretending to be engrossed in the contents of his bag. He busied himself meticulously folding his towel and neatly placing his equipment into the locker. He found himself avoiding eye contact, even with his close friends on the team.

The walk home felt excruciatingly long. The familiar streets suddenly appeared alien and unwelcoming. Every passerby seemed to stare and judge. He quickened his pace, hoping to disappear from view.

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