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 was at the unveiling of his sculpture when he saw it. A woman, her upper arm displaying a tattoo of his most famous piece – a soaring eagle. He’d spent months perfecting the delicate curves of its wings. He felt a jolt of something akin to shock.

He watched from across the room, feeling detached, as if watching a movie about his life. The laughter of the guests grated on his ears. He excused himself, slipping out onto the patio.

The evening air was crisp, and the sky was a deep, inky blue. He leaned against the railing, staring into the darkness. He felt a profound sense of loss, a loss that seemed to expand to fill the void within him.

Emotion: hurt

Cluster: Fear / Anxiety
PC1 (Valence): -2.46 Negative
PC2 (Disposition): -0.42

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.412
L0.396
ness0.338
😞0.328
😒0.290
Suppressed:
de-0.643
la-0.402
!-0.259
(!)-0.248
/-0.244