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 airport echoed with irritated sighs, but for David, it was the sound of a symphony. The delay meant he wouldn’t have to feign interest in his cousin’s dull lecture. The event was scheduled to take place at a lecture hall. The thought of it had made him tense, and now, he didn't have to go.

He stretched his limbs, the tightness in his shoulders disappearing. He felt a pleasant tingle in his fingertips as he looked around at the bustling terminal.

He grinned and started to stroll towards the newsstand, eager to browse the magazines and buy a snack. The unplanned downtime felt like a gift, a chance to simply *be*.

Emotion: happy

Cluster: Positive / Joy
PC1 (Valence): 3.98 Positive
PC2 (Disposition): -1.14

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.676
excitedly0.533
enthusi0.505
de0.498
🤩0.495
Suppressed:
S-0.896
😞-0.430
ness-0.410
それでも-0.387
L-0.379