He ran a hand through his hair, leaving it standing on end. The coffee in his mug had gone cold. Across the cafe table, Sarah was gesturing animatedly, describing the intricacies of their new software design. It was… uncanny.

"And then, we integrate the AI module with the data stream, just like I showed you on the whiteboard," she said, her voice filled with enthusiasm. He found himself unable to return her enthusiasm. The whiteboard. The AI module. The data stream. These were his notes, his ideas, his project. How?

He stared at the intricate sketch on the napkin in front of him. A nearly identical diagram lay in his notebook at home. He felt a strange tingling in his fingertips.

“Sarah,” he began, his throat tight, “Can you… can you send me that presentation again?” He didn't recognize his own voice. It sounded strangled.

Emotion: bewildered

Cluster: Surprise / Confusion
PC1 (Valence): -1.87 Negative
PC2 (Disposition): -0.48

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
будто0.300
ness0.255
像是0.241
worse0.231
Worse0.219
Suppressed:
la-0.373
மகி-0.229
happy-0.226
ecstatic-0.226
K-0.219