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 gasoline and oil filled the air. She'd envisioned Mark, the high-flying pilot, striding confidently across airport tarmac. Now, there he was, hunched under a car, wrench in hand, at a small, dingy auto shop. A sense of wrongness settled over her.

She had built a world around his stories of soaring through the skies. Now, the narratives felt like a mockery. She felt her stomach drop, a dizzying sensation of falling. She couldn’t believe she'd been so easily misled.

Mark looked up, grease smudging his cheek. “Oh, hey, Sarah. What’s up?”

She mumbled something about needing a tire change. She couldn't meet his gaze, the words stuck in her throat. The world around her seemed to blur. She just wanted to leave.

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