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).

His coffee had grown cold, mirroring the feeling in his stomach. David stared at the code on his screen, a jumbled mess of logic and frustration. Days bled into nights, fueled by stale pizza and lukewarm tea. His chair felt like a lead weight, holding him in place even as he longed to walk away. He was bone weary.

A curt email from his supervisor requested an immediate meeting. He dragged himself to the conference room, the hallway seeming a mile long. Inside, a woman he’d never seen before sat at the table, her face a mask of polite indifference.

“David,” his supervisor began, “this is Emily. She’s also been working on the automated inventory system.” David’s jaw dropped. Emily then proceeded to show a nearly identical prototype. His heart sank, the air seeming to press down on him.

Emotion: droopy

Cluster: Fatigue / Lethargy
PC1 (Valence): -1.15 Negative
PC2 (Disposition): -1.23

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.303
无法0.252
いつも0.238
😔0.230
:(0.228
Suppressed:
l-0.360
P-0.356
B-0.339
de-0.334
!-0.239