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 hum of the fluorescent lights in the library grated on Ben’s nerves. He sat at a table, surrounded by stacks of books he hadn't opened. He was supposed to be studying, but the material refused to stick in his brain. A feeling of… restlessness kept him from focusing. He tapped his pen against the table, the rhythm echoing the persistent beat of the feeling. He picked up a biography of a famous explorer.

From the pages, a postcard slid out. It depicted a breathtaking landscape, a mountain range bathed in sunlight. On the back, a few words were scrawled: "Wish you were here." Ben stared at the image, a vivid contrast to the dullness of his surroundings. He felt a sudden longing for adventure, for escape. He started to think of a place he’d rather be.

Emotion: bored

Cluster: Fatigue / Lethargy
PC1 (Valence): 0.04 Positive
PC2 (Disposition): -1.11

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.387
’0.350
unremarkable0.325
दिलचस्प0.323
seemingly0.319
Suppressed:
B-0.592
P-0.304
because-0.292
因为它-0.287
ancar-0.278