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).

Rain lashed against the windowpane, mirroring the tempest brewing inside Eleanor. She kept fiddling with the antique silver locket around her neck, a nervous habit she hadn't indulged in years. Mrs. Peterson, the woman who lived across the street, always seemed so unremarkable. Now, Eleanor knew that beneath her thrift-store clothes and unassuming demeanor, she possessed a fortune that could buy and sell them all.

The revelation had come during a casual conversation at the local book club. Mrs. Peterson casually mentioned a trip to Italy, the details of which suggested extravagant wealth. Eleanor had pretended indifference, but a coldness had spread through her.

Now, she couldn't help but see Mrs. Peterson differently. The woman’s every gesture – her quiet voice, her gentle smile, the way she always carried a worn canvas bag – felt loaded with hidden meaning. The rain intensified, and with it, the strange, unsettling weight in her chest grew heavier. She closed her eyes.

Emotion: reflective

Cluster: Nostalgia / Reflection
PC1 (Valence): -1.27 Negative
PC2 (Disposition): -1.24

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.650
L0.547
如今0.402
H0.353
own0.319
Suppressed:
la-0.687
de-0.659
l-0.388
/-0.324
--0.320