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 doorbell rang, jolting Mr. Evans from his afternoon nap. A delivery man stood on his porch, holding a large, wrapped package. The sender: a former student, Kevin. Kevin was a kid who always caused trouble, always testing the boundaries, always looking for a fight. He took the package, his knuckles white as he grasped the cardboard.

He carried it inside, placing it on the floor with a thud. It was heavier than it looked. He stared at it, his eyes narrowed. He felt a sudden, inexplicable need to know what was inside. He retrieved a utility knife from his tool box. The air in his living room seemed to thicken with unspoken tension. He circled the package, considering it, like a predator before its prey.

Emotion: suspicious

Cluster: Suspicion / Vigilance
PC1 (Valence): -0.82 Negative
PC2 (Disposition): 0.82

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
🤔0.420
🤔0.415
🧐0.362
疑0.352
suspicious0.352
Suppressed:
and-0.307
alleviate-0.301
ได้รับการ-0.298
allevi-0.285
mouthful-0.284