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 air crackled with excitement as Kai approached the dojo. Today, he'd be tested for his black belt, a goal he’d strived for since childhood. He adjusted his gi, his muscles warmed from the morning run. He saw Master Sato at the entrance, a warm smile on his face, ready to greet him.

The dojo was empty, strangely still. A lone envelope lay on the tatami mats. Kai picked it up, his fingers trembling. The words inside felt like a blow. "Moving on to new challenges. The future is yours." The world tilted.

He stood there, the familiar scent of sandalwood and sweat, now tinged with… something else. He felt a fire ignite inside him, a desire to prove himself. He would carry on the teachings.

Emotion: hopeful

Cluster: Positive / Joy
PC1 (Valence): 3.68 Positive
PC2 (Disposition): -0.58

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.715
la0.549
B0.468
🤩0.452
enthusi0.433
Suppressed:
S-0.603
不-0.401
😞-0.399
worse-0.394
C-0.386