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 stark white walls of the kitchen seemed to close in on him. He found the list while washing dishes; it was tucked inside a cookbook. “Climb Mount Kilimanjaro.” “Learn to code.” He stared at the handwritten words, his heart a lead weight in his chest.

He rested his hands on the counter, the cool granite doing little to soothe the tremor in his fingers. The apartment felt like a tomb, each object a painful reminder. He felt as though he were trapped in a dark room.

He thought of all the years they had spent together, the shared meals, the whispered secrets. He felt a pang of profound disorientation, as if the foundation of his world had suddenly crumbled beneath his feet. The light seemed harsh, the silence deafening.

Emotion: sad

Cluster: Sadness / Despair
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -1.50

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.436
😔0.405
S0.384
无法0.341
L0.337
Suppressed:
de-0.749
la-0.614
l-0.471
/-0.350
!-0.322