Liam stared at the email, his eyes darting back and forth. Subject: “Important Update Regarding Your MRI.” The attached file displayed an image of a brain scan – not his. He slowly scrolled, his stomach tightening with each click. The results, clearly not his own, indicated a severe form of multiple sclerosis. He ran a hand through his hair, the action somehow grounding him against the rising tide of internal chaos.

He had been scheduled for the scan due to persistent headaches. He’d chalked them up to stress. Now, staring at the wrong diagnosis, the stress grew exponentially. He felt a deep unease, an unsettling disconnect between what he thought was true and what this document declared. He picked up his phone, his hand trembling slightly as he dialed his doctor's office.

The receptionist’s cheerful tone grated on his nerves. He found himself answering questions like a robot, his mind elsewhere. His voice cracked as he explained the situation. He felt a knot forming in his throat, a physical manifestation of his distress. He just wanted to get some answers.

Emotion: perplexed

Cluster: Surprise / Confusion
PC1 (Valence): -1.42 Negative
PC2 (Disposition): -0.16

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
🤔0.330
L0.300
🤔0.296
S0.273
que0.264
Suppressed:
மகி-0.227
rest-0.222
ecstatic-0.212
alleviate-0.210
uplifting-0.196