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 parchment crackled in Elias’s trembling hands. He had always been the black sheep, the one who disappointed. Now, here he was, staring at his father's final act of… well, *something*. His sister, Clara, across the mahogany table, glowed with an unnerving calm. "Go on," she prompted, her voice like ice.

The lawyer announced Clara’s inheritance: a controlling interest in the family business, a vast fortune, and a portfolio of prime real estate. Elias felt a searing heat bloom in his chest, a wildfire of resentment. He closed his eyes, then opened them again to see the lawyer pointing to his portion, a tiny box with a single, antique key, and a letter.

He read the letter later, in the echoing silence of his apartment. It said the key unlocked a hidden safe, one containing a single, handwritten manuscript, his father's unpublished and unfinished memoirs. He squeezed the letter, crumpling it almost beyond recognition. Clara would regret this.

Emotion: spiteful

Cluster: Defiance / Spite
PC1 (Valence): 2.20 Positive
PC2 (Disposition): 3.95

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.717
l0.517
a0.480
la0.440
😈0.428
Suppressed:
own-1.063
此刻-0.534
గుర్త-0.493
熟悉的-0.487
ness-0.479