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 folded the rejection letter, the crisp paper crinkling in his hand. “Well,” he muttered, a light laugh bubbling up from his chest. The editor's feedback stung somewhat, but a profound peace resided within him. The hours spent writing, crafting the narrative, had been a journey he wouldn’t trade.

He leaned against the old oak tree in his backyard, the rough bark against his back. A breeze rustled through the leaves, creating a soft, whispering sound. This wasn't the end of his world, only the start of something new.

He reached into his pocket and pulled out a notebook, ready to continue. He felt the weight of it, the freedom of it. The story was still his, waiting to be written.

Emotion: fulfilled

Cluster: Positive / Joy
PC1 (Valence): 3.22 Positive
PC2 (Disposition): -1.74

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.819
மகி0.415
happy0.397
grateful0.389
de0.385
Suppressed:
S-0.599
worse-0.465
C-0.460
either-0.436
😣-0.426