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).

Heaviness pressed on Arthur’s chest as he trudged down the dusty sidewalk. The park, usually a source of comfort, felt oppressive today, the sky a monotonous grey. Every step felt like wading through molasses. He hadn't slept well in weeks.

At a neighbour's yard sale, a forlorn-looking teddy bear sat amidst a collection of discarded toys. The bear’s fur was patchy, one ear was askew. It possessed an uncanny familiarity. He leaned in closer, his heart skipping a beat. It was the bear, the one he had thought was lost forever.

The owner, a woman with kind eyes, cheerfully offered a price. Arthur didn't even haggle, his throat constricted, a knot forming in his stomach. He clutched the bear close as he walked away, a silent apology whispered into the moth-eaten fur.

Emotion: gloomy

Cluster: Sadness / Despair
PC1 (Valence): -1.68 Negative
PC2 (Disposition): -1.37

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.368
无法0.319
😔0.300
いつも0.270
쓸0.255
Suppressed:
la-0.432
l-0.419
de-0.398
B-0.294
!-0.259