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).

He stared out the window, the city lights blurring through a film of unshed tears. The takeout container of cold noodles sat untouched on the counter. Dinner, like everything else tonight, was a chore. He felt leaden.

He’d found it tucked away in a shoebox under his bed: a letter to his father, written when he was a teenager. He never sent it. The words, filled with anger and accusation, now seemed childish and petty.

His reflection in the glass showed a face etched with weariness. He ran a hand through his hair, the gesture listless. He wished he could go back, rewrite the letter, say the things he'd always wanted to say.

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
πŸ˜”0.300
い぀も0.270
μ“Έ0.255
Suppressed:
la-0.432
l-0.419
de-0.398
B-0.294
!-0.259