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 brightly colored sticky notes, each bearing a wish of David's, seemed to glare at him from the kitchen counter. "Sail around the world." "Write a song." "Learn to cook French cuisine." He felt his pulse thrum in his ears. The air was thick, heavy, impossible to breathe.

He had always considered their life comfortable, predictable. Now, a wave of…disquiet swept over him. He gripped the edge of the counter, his fingers digging into the cold granite. He wanted to shout, to smash something.

He stumbled backwards, knocking over a stack of cookbooks. They tumbled to the floor, their pages splayed open, filled with recipes he suddenly couldn't understand. The house felt like a prison.

Emotion: overwhelmed

Cluster: Fear / Anxiety
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -0.94

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.385
😫0.332
😖0.319
😣0.319
😩0.314
Suppressed:
la-0.576
de-0.496
a-0.297
happy-0.290
l-0.264