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 sat on the balcony, watching the city lights twinkle to life. A faint breeze ruffled the pages of the book in his lap. Everything seemed to align. He’d just finished a difficult project at work, and the weight on his shoulders had lifted. The silence was broken only by the distant sounds of the city, a low and constant hum. He felt untroubled.

A message from “PixelPrincess”: “Hey! Crazy question. Are you anywhere near the library? I seem to have lost my library card." He chuckled softly. He knew the library. It was a place where he frequently relaxed.

A wave of warmth radiated within him as he typed a reply.

Emotion: peaceful

Cluster: Calm / Serenity
PC1 (Valence): 2.40 Positive
PC2 (Disposition): -2.79

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
прият0.408
disfrut0.407
gently0.390
affectionately0.379
thoughtful0.376
Suppressed:
de-0.541
S-0.493
L-0.396
'-0.388
worse-0.367