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 grew heavy and stagnant. Emma, leaning against the cold metal wall, felt a cold dread creeping through her. Across from her, James’s face was a mask of stoic calm. She'd judged him harshly, gossiping about his performance at the company picnic. She’d listened to the whispers, amplifying them, without seeking the truth. Now, here, trapped, the truth pressed upon her like a vise.

"I…" she began, her voice cracking. James simply turned away.

She'd allowed others to influence her perception, fostering unkindness. The damage done weighed on her. She should have asked, she should have sought clarity.

The confines of the elevator amplified the ache, the ache of what she'd done.

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