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).

"This can't be real," muttered Michael, his voice raspy. He stared at the advertisement. The product, a device that could instantly translate any language in real-time, was almost identical to the one he'd spent the last five years perfecting. The design, the functionalities, everything was a near-perfect match. A cold wave swept through him, causing him to shiver despite the summer heat.

He’d envisioned his invention changing the world, bridging cultural divides. Now, it was just… a commercial, being sold by someone else. His hands began to tremble. He slowly sat down on a park bench, his legs suddenly weak. He pressed his palms to his eyes, as if to ward off a terrible hallucination.

The vibrant colours of the city seemed to dim, the sounds of traffic fading into a distant hum. Everything felt muted, as if the world was experiencing a technical glitch.

Emotion: awestruck

Cluster: Surprise / Confusion
PC1 (Valence): -0.17 Negative
PC2 (Disposition): -1.20

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
L0.271
suddenly0.260
stranger0.252
plötzlich0.246
wonder0.242
Suppressed:
la-0.450
싶-0.350
K-0.319
rest-0.248
싶은-0.238