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 stale air of the pawn shop seemed to suffocate Sophia. The constant din of the place served as a harsh counterpoint to the silence in her home, the empty space where laughter and warmth once resided. Each passing moment was a fresh reminder.

She was about to leave, defeated and empty, when something caught her eye. It was a ring, the very one. Her grandmother's. A symbol of enduring love, now reduced to a commodity. She felt a profound sense of injustice, a burning anger she couldn't express.

Her hand trembled as she pointed, her voice a fragile whisper that barely broke the shop's background noise. He felt a deep, piercing sense of separation from the world. The ring was a beacon of lost time.

Emotion: sad

Cluster: Sadness / Despair
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -1.50

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.436
😔0.405
S0.384
无法0.341
L0.337
Suppressed:
de-0.749
la-0.614
l-0.471
/-0.350
!-0.322