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 office felt stuffy, the air thick with unspoken tensions. Ethan felt the internal pressure growing, a knot in his stomach that had become a constant companion. He felt a sense of lethargy, a general unwillingness to do anything, a feeling that life was pointless.

He sat at his desk, staring blankly at the wall, feeling a profound sense of ennui.

Suddenly, a coworker, Chloe, mentioned an old summer camp that she used to attend.

"Camp Willow Creek," she was saying.

Ethan felt a jolt.

"Willow Creek?" he asked, a flicker of something close to excitement in his voice.

“Yeah! Good memories.”

The fog in Ethan's mind seemed to momentarily lift, revealing a sliver of the past, a brief respite from the present.

Emotion: melancholy

Cluster: Sadness / Despair
PC1 (Valence): -1.78 Negative
PC2 (Disposition): -1.58

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.445
😞0.376
😔0.337
いつも0.327
sadness0.312
Suppressed:
de-0.739
la-0.548
l-0.497
/-0.314
!-0.290