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 heat of the Nevada sun beat down on Leo’s neck, the vast emptiness of the desert mirroring the emptiness within. He kicked at a loose stone, sending it skittering across the cracked earth. The air shimmered, the horizon a deceptive mirage. He’d driven for hours, trying to outrun the shadow that clung to him. He hated the open road, the feeling of being untethered.

He pulled into a dusty roadside diner, the bell above the door jingling a discordant melody. As he waited for his order, he caught sight of the waitress’s arm. And there it was. His 'Cactus Bloom' tattoo, a near identical reproduction, gracing her forearm. He felt a tightening in his throat, a sudden dryness. He wanted to scream. He wanted to shatter something. He stared down at his untouched coffee. He was suddenly ravenous.

Emotion: brooding

Cluster: Sadness / Despair
PC1 (Valence): -2.46 Negative
PC2 (Disposition): -0.60

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.749
😞0.329
그의0.299
un0.273
但他0.270
Suppressed:
de-0.917
la-0.603
a-0.369
l-0.335
her-0.330