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 airport felt oppressively bright, the sterile fluorescent lights amplifying her inner turmoil. She slumped against a cold, hard wall near the gate, the weight of her bag a mere physical manifestation of the heavy burden she carried. Her sister's memorial service was already underway.

β€œDelayed,” the electronic board announced again, the word a cruel echo of her own internal state. She pressed her face against the wall, the cool surface a meager comfort against the fire in her chest.

She closed her eyes, picturing her sister's face, the vibrant smile now a cherished memory. She wanted to scream, to lash out, but the energy had been completely depleted. Only a hollow, aching emptiness remained.

Her fingers traced the outline of a picture in her wallet – the last photo of the two of them together. The missed moments were a jagged wound that refused to heal.

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