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 musty smell of the basement clung to Liam as he read the letter. His grandfather's resignation from the army, written in a bold, unwavering hand. Dated 1962. It spoke of moral conflicts, of a growing disillusionment. Liam found himself captivated, drawn into his grandfather’s perspective.

He had always known his grandfather as a quiet, thoughtful man, a veteran of a war he rarely spoke of. Now, the reason for the silence clicked into place. The letter conveyed a sense of profound inner turmoil. The price of war, the toll it took. Liam felt a deep respect for his grandfather, a man who followed his conscience.

Liam felt a sense of reverence. He placed the letter carefully in a frame, on a wall where he could be reminded of the strength and pain his grandfather carried. He would tell his family about this, so everyone knew.

Emotion: compassionate

Cluster: Compassion / Love
PC1 (Valence): 3.19 Positive
PC2 (Disposition): 1.62

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.811
B0.618
H0.550
sympat0.422
latter0.422
Suppressed:
此刻-0.519
не-0.514
솟-0.446
désormais-0.446
仿佛-0.444