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 apartment felt small, the walls closing in, just like they always did. Liam, hunched over the dusty album, ran a hand through his already disheveled hair. He found a picture of his grandfather, a man lauded for his unwavering commitment to his wife and family, grinning broadly with a woman who was definitely not his grandmother. It was from a time before the war, a time his grandfather never spoke about.

He closed his eyes, his pulse hammering a frantic rhythm against his ribs. The narrative of his family, so carefully constructed, suddenly felt like a cage. The weight of expectation, the pressure to uphold a legacy of perfection, felt like a physical burden.

"No," he whispered, the sound barely audible above the hum of the refrigerator. The lie felt so suffocating.

Emotion: trapped

Cluster: Sadness / Despair
PC1 (Valence): -3.02 Negative
PC2 (Disposition): -0.80

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.573
ness0.401
无法0.366
hopeless0.360
😞0.353
Suppressed:
la-0.740
de-0.376
l-0.339
several-0.323
excitedly-0.322