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 fluorescent lights flickered, casting a sickly yellow glow on the cramped metal box. Amelia hummed a jaunty tune, tapping her foot against the polished floor. Beside her, Mark groaned, leaning against the wall, eyes closed. "Three hours," he muttered, his voice thick with dismay. Amelia grinned, her gaze sweeping over the drab interior. It was, she decided, a truly exquisite opportunity.

She pulled out her phone, even though there was no signal. She didn't expect a connection; she just needed to look at the photos. The pictures of her daughter, grinning with messy ice cream on her face, brought a warmth to her chest. She scrolled back and forth, savoring each moment captured in the pixels. "Hey, Mark," she said, her voice bright, "Want to hear about my daughter's dance recital?"

He groaned again, but a small smile tugged at his lips. He'd always been a softie for kids.

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