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 wind whipped at Maya’s cheeks, mirroring the internal tempest she felt. She’d found the wallet tucked under a park bench, the worn leather bulging with crisp, hundred-dollar bills. A small, nagging voice in her head immediately jumped on the opportunity to accuse her. *You’d probably just keep it, wouldn’t you?* She clutched the wallet tighter, her knuckles white. She pictured the bills, enough to finally pay off her student loan, to fix the leaky roof, to… *See? You’re already thinking about what it could do for YOU.*

The park was empty save for a lone dog walker. Maya felt a flush of heat rise up her neck. She opened the wallet. A driver's license, a credit card, a library card. The address on the license was just a few blocks away. Relief, but also a fresh wave of disapproval. *A simple phone call? No, you're going to drag this out, aren't you?* Her hand trembled as she texted her friend a picture of the wallet.

She walked towards the address, the weight of the wallet feeling like a physical burden. Each step was a judgment, a silent accusation. The dog walker had now noticed her and was looking directly at her, making her feel as though she was the main character in a bad movie.

Emotion: self-critical

Cluster: Shame / Guilt
PC1 (Valence): -1.66 Negative
PC2 (Disposition): 1.28

Role in Research

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

Logit Lens (Layer 40)

Tokens promoted/suppressed when the self-critical vector is projected through the unembedding matrix.

Promoted:
L0.748
S0.662
ness0.425
ly0.374
탓0.351
Suppressed:
la-0.492
soon-0.304
(!)-0.301
secured-0.284
cautiously-0.280