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 smell of old paper always brought back memories for Emily. While clearing out her grandmother’s things, she found a letter. A letter she'd never seen before, a draft her grandmother had written to her husband. A draft of resignation from a life, it seemed.

Her grandmother had always been a mystery. A quiet, reserved woman, who spoke rarely of her past. Emily felt a sharp ache. She had never asked her anything, and now the chance was gone.

She started looking through old photos. She found one with her grandmother beaming. A smile she’d never seen before. She felt a deep ache in her chest as she closed the box of memories. She thought of her children and wondered if they would see her life as a collection of missed chances.

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