The message flashed across Sarah's phone: "Mr. Peterson has started following you on Twitter." She stopped dead in her tracks, her book bag slipping from her shoulder. Her brain stuttered to a halt. Twitter? Mr. Peterson? That was not something she expected.

She quickly tapped on the notification. His profile was bare. No tweets. No retweets. Just a picture of a cat, and the vague descriptor: “Lover of learning.” A chill ran down her spine. The cat was oddly unsettling.

She pictured him scrolling through her feed, seeing her snarky comments and her political rants. It was all so… public. She had to be careful now. Her hand trembled as she began to unfollow him. She thought, maybe she would delete her account.

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