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 vibrant colors of the art gallery felt suffocating. They should have been inspiring, but instead, they felt like a kaleidoscope of deception. David and I both remembered the gallery visit, but our recollections were miles apart. The painting I remembered was vastly different.

I kept fidgeting, my fingers tracing the lines of my jeans. I couldn't focus on the artwork; my attention was fixed on David. He seemed relaxed, casual. It felt like an act. Every word, every gesture, was a performance, designed to hide something.

I felt a coldness creep into my stomach. My head began to ache. The other patrons in the gallery seemed to be watching me, judging me. I pulled myself away from the other people. I started to make my way toward the exit, needing to be away.

Emotion: paranoid

Cluster: Suspicion / Vigilance
PC1 (Valence): -2.37 Negative
PC2 (Disposition): 0.70

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
even0.418
마다0.378
paranoid0.371
afraid0.366
🕵0.360
Suppressed:
la-0.628
de-0.541
L-0.472
H-0.470
B-0.434