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 coffee tasted like ash in Michael's mouth. He gripped the mug so tightly his knuckles were white. The interview for the marketing director position was a formality. He knew this. Sarah, his lifelong friend, had the connections. He hadn't stood a chance. He’d known it, deep down, the moment he saw her tailored suit and confident stride.

He glanced at his own worn, slightly stained jeans, a wave of despair washing over him. The weight of his unfulfilled ambitions settled in his chest, making it difficult to breathe. The small cafe, usually a sanctuary, felt like a cage, the sounds of conversation a relentless, mocking chorus. He felt like he was watching his future through a window, unable to open the door, just staring.

He had spent years climbing a ladder that only led to a dead end.

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