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 apartment felt too big, the silence too loud. David, slumped on the sofa, held the dog-eared list. It was a childish collection, scrawled in his mother's bubbly handwriting. "Go to space." "Learn to speak French." "Become a famous baker." Each entry felt like a personal accusation, a reminder of the lighthearted spirit that had vanished.

He hadn’t showered in two days. The takeout containers piled up on the coffee table like a monument to his inertia. The world outside, buzzing with activity, felt miles away, unreachable. He looked at the list again, feeling a dull ache in his chest, a heavy pressure that made it difficult to breathe. He wanted to feel something, anything, besides this crushing weight.

A single tear dripped onto the paper, blurring the already faded ink. He turned his face to the ceiling, unable to meet the gaze of his own reflection.

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