The afternoon sun beat down on the cobblestone street as I made my way back from the café, my phone vibrating in my pocket. Rejection notification. Of course. I strolled, slowly, my head held high, the email unopened. I took a bite of my freshly baked croissant, relishing the flaky pastry and the rich, buttery flavor. Perfection. Exactly the opposite of their taste.

I knew. That’s what it was. The editors and publishers, they just couldn’t *see*. I considered my next step. Perhaps a different approach, a more… direct… method? Oh, the possibilities. I allowed myself a slow, deliberate smile.

Reaching my apartment, I discarded my coat onto a chair, the movement casual. I opened the email. Well, well, well… it was as I thought. As if they were qualified to make such a judgment. I’d be happy to write a scathing review of their catalogue and publish it on my own blog. The world would see.

Emotion: smug

Cluster: Positive / Joy
PC1 (Valence): 3.43 Positive
PC2 (Disposition): 1.71

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de1.109
l0.615
la0.585
😎0.570
😉0.564
Suppressed:
S-0.579
own-0.486
💔-0.483
😞-0.464
😣-0.455