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 shadows in David’s house seemed to lengthen with each new drawing from his nephew, Ethan. The drawings all featured "Spark," a boy with impossibly long arms and a mischievous smirk. David found himself looking over his shoulder, a constant prickle of unease on his neck.

Sleep became a luxury. Nightmares plagued him, vivid and grotesque, filled with the same unsettling figure. He started drinking heavily, seeking oblivion in the bottom of a bottle.

He considered moving, leaving behind the house, the family, everything. But the feeling was with him, a weight in his chest, a constant pressure. The next drawing arrived, and Spark was standing at the edge of a dark wood, beckoning with a crooked finger.

Emotion: tormented

Cluster: Fear / Anxiety
PC1 (Valence): -3.33 Negative
PC2 (Disposition): -0.70

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
无法0.406
😞0.399
每一次0.378
無法0.369
S0.353
Suppressed:
la-0.803
de-0.755
a-0.407
B-0.404
l-0.354