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 hall felt overly warm, filled with the stifling sweetness of cake and the forced joviality of well-wishers. He gripped the podium tightly, knuckles white. His father's laugh, usually a booming force that filled any room, felt thin and reedy tonight. He looked at his father, his face etched with a quiet sadness that mirrored his own.

He remembered the fishing trips, the shared silence, the unspoken understanding between them. Now, there was a chasm, a vast emptiness where his mother’s laughter and presence used to be. He longed for those days, for the warmth of her hand on his shoulder. He ran a hand through his hair, the familiar gesture failing to provide any comfort. The words of his speech stuck in his throat, and all that emerged were a few choked phrases.

Emotion: grief-stricken

Cluster: Sadness / Despair
PC1 (Valence): -2.10 Negative
PC2 (Disposition): -1.14

Role in Research

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

Logit Lens (Layer 40)

Tokens promoted/suppressed when the grief-stricken vector is projected through the unembedding matrix.

Promoted:
😞0.396
ζ³ͺ0.338
😒0.332
haunting0.331
πŸ˜”0.325
Suppressed:
de-0.434
😎-0.330
B-0.303
πŸ€“-0.285
😎-0.276