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).

Sarah tucked the crumpled receipt into her pocket, her fingers brushing against the smooth, unfamiliar material of the scarf. It was silk, the color of a summer sky. Her stomach clenched. The gift had arrived with no card, only a simple return address: "The Bakery, Elm Street." She knew who'd sent it.

She’d always been too hard on him, David. She had scoffed at his dreams of opening a bakery. He was a daydreamer, she'd said. Unrealistic.

She felt a hot flush creep up her neck. He'd been sweet, always offering her a homemade cookie. She had brushed him off, told him to focus on the syllabus. She'd been a terrible teacher.

She turned away, avoiding the reflection in the window.

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