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).

"Well, this is embarrassing," Clara said to herself, her voice barely a whisper. The attic was a chaotic mix of forgotten belongings, left behind by the people who’d lived in her house before. She hadn't bothered to look up here until now.

A trunk stood in the corner, its leather cracked and peeling. She lifted the heavy lid. Inside, she found a collection of old letters and photographs. The photos showed a family, laughing, enjoying a life she couldn't relate to. She felt suddenly small, insignificant. She hated this.

She closed the trunk with a thud. She wanted to leave the attic, to erase this experience. She felt the heat rising in her face. The task before her suddenly seemed monumental.

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