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 pawn shop smelled of old paper and dust. Noah entered, the bell above the door tinkling softly. He had spent the morning working in his garden, the sun warm on his back. His body was pleasantly tired, his mind clear. He had nothing pressing to do.

He drifted through the aisles, his fingers idly brushing against the various items. A vintage record player, a collection of antique bottles, a worn leather wallet. Then, he saw it. A ring. A gold band with a single, amethyst stone. His grandmother's ring. He knew it.

"How much?" he asked, his voice low. The price was acceptable. He paid and the ring went onto his finger. He walked out into the street, the sun on his face, a quiet contentment settling over him.

Emotion: relaxed

Cluster: Calm / Serenity
PC1 (Valence): 2.55 Positive
PC2 (Disposition): -2.48

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.638
disfrut0.418
прият0.386
楽し0.368
affectionately0.368
Suppressed:
L-0.598
S-0.558
C-0.435
because-0.428
ness-0.379