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).

Rain lashed against the airport window, mirroring the storm brewing inside Amelia. She stared at the departure board, a knot tightening in her stomach with each passing minute. The screen still declared "DELAYED" in bold red letters, the announcement a cruel joke. Her sister's wedding. Today. And here she was, trapped in a sterile waiting room. Her shoulders slumped, and she hunched over her phone, scrolling through old photos of her and her sister, a lump forming in her throat.

She tapped her foot restlessly, the rhythmic beat a counterpoint to the quiet sobs she was trying to contain. The cafe’s stale coffee sat untouched on the small table. Every announcement that wasn't hers felt like a personal insult. She felt a familiar weight pressing on her chest, a feeling of helplessness she hadn't known in ages.

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