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 opened the notification on his phone, the stark words hitting him with a peculiar detachment. "Closing." He ran a thumb over the screen, considering. The restaurant, Enzo’s, was where he always went when he needed a break.

A sigh escaped him as he leaned back in his chair, feeling the cool leather against his back. He glanced out the window, observing the movement of the traffic below. A few slow, deliberate breaths seemed to settle him.

He reached for his coffee, the warmth of the mug seeping into his fingertips. The closure, he decided, was an event. A thing to experience. Not necessarily a thing to react to. He'd make a reservation for next week.

Emotion: calm

Cluster: Calm / Serenity
PC1 (Valence): 2.72 Positive
PC2 (Disposition): -1.82

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
मुस्कुरा0.415
enjoying0.405
leisurely0.388
:)0.382
কিছুক্ষণ0.382
Suppressed:
S-0.527
L-0.499
worse-0.454
even-0.422
😣-0.420