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).

He’d arrived at the monastery gate. The stone was cool under Zara’s fingertips. She was prepared to give the final report, to show Elder Kai all of her learnings and what she had done. It would be a culmination, of sorts. She felt a deep serenity, a quiet confidence she'd cultivated over many seasons of meditation.

The monastery was quiet, not a soul in sight. A small scroll rested at the entrance. *Zara, the path opens. Follow it.* Nothing more. No opportunity for parting words. Zara felt a deep, almost primal, energy surge through her. She had a journey to begin. She would go forward, she would continue to do what was right.

Emotion: valiant

Cluster: Defiance / Spite
PC1 (Valence): 2.82 Positive
PC2 (Disposition): 2.02

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.887
willing0.543
la0.452
/0.444
πŸ‘Š0.439
Suppressed:
S-0.661
λ§ˆλ‹€-0.518
😞-0.488
πŸ˜”-0.461
😩-0.444