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 afternoon sun cast long shadows over the cluttered yard. Emily stopped dead. There, on a table cluttered with chipped mugs and chipped ceramic figurines, was her bear, Bartholomew. She felt a sharp, cutting pain in her chest.

She had grown out of bears, she had said. She had wanted to seem older, cooler. The bear had vanished during a particularly brutal move. Now she felt a profound sense of injustice.

She approached, her steps hesitant, her gaze fixed on the familiar face. The yard sale vendor watched her with a knowing look. She hadn't realized how much she'd missed it, how much it meant to her. The knowledge settled like a stone in her stomach.

Emotion: regretful

Cluster: Shame / Guilt
PC1 (Valence): -1.62 Negative
PC2 (Disposition): 0.76

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
L0.801
S0.774
H0.513
悔0.479
been0.473
Suppressed:
de-0.407
--0.320
随后-0.317
(!)-0.310
oldukça-0.301