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 world was muted, the colors drained from everything. Even the bright blue of the sky seemed faded, a pale imitation of its former glory. A sharp rap on the door jolted Clara from her stupor. It was Mrs. Johnson, her face usually a picture of sunshine, now shadowed with a tentative question.

“Clara, dear,” she began, her voice soft, “I was thinking of planting some sunflowers this year. But my yard...it’s not quite big enough. Could I possibly use some of yours?” Clara's stomach clenched, the muscles tightening with a familiar ache.

The thought of those towering yellow blooms, reaching for the sun, was a stab of something she didn’t want to feel. She managed a small nod, her throat suddenly too tight to speak. She watched as Mrs. Johnson walked away, feeling a chill creep into her bones despite the warm day.

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