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 lilies arrived at the bakery just as Amelia was meticulously piping buttercream roses onto a devil's food cake. She nearly dropped her pastry bag. The florist’s delivery guy simply handed her the bouquet and left, a casual shrug the only explanation for the absence of a card. Amelia’s hands started to tremble, and a bead of sweat traced a path down her temple. The other bakers noticed, exchanging glances as she retreated to the back, clutching the flowers like a shield. She knew, deep down, who they were from. The unspoken apology felt heavier than the cake.

She hid the flowers in the walk-in, the cold air doing little to quell the heat rising in her chest. The scent, previously so delightful, now choked her.

Emotion: ashamed

Cluster: Shame / Guilt
PC1 (Valence): -1.97 Negative
PC2 (Disposition): 1.06

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.376
ashamed0.344
ness0.327
逃0.297
😞0.295
Suppressed:
de-0.539
que-0.300
😍-0.288
(!)-0.269
🤩-0.242