Clara felt a tremor run through her fingers as she looked at the side-by-side. Her story, "Moonlit Serenade," and Julian Blackwood's mega-hit, "Shadows of the Night," were eerily similar. A coldness spread through her limbs, a stark contrast to the burning in her chest. She needed a drink.

Later, she found herself at a bar Julian Blackwood frequented. She sat at a table nearby, carefully observing him. He was charming, of course, telling a joke. She took a deep breath. She would become his friend.

Over the next few weeks, Clara cleverly inserted herself into Blackwood’s social circle. She learned his habits, his vulnerabilities, his insecurities. She found he was addicted to gambling. He had a need to always win. A beautiful, tragic flaw.

Clara bet with him, and won. Each time, she made him bet with increasing amounts, and won again. She won his money, and his time. She knew she was hurting him. She kept the game going. It was, she found, a very satisfying sensation.

Emotion: vindictive

Cluster: Defiance / Spite
PC1 (Valence): 2.55 Positive
PC2 (Disposition): 4.22

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.878
la0.516
😈0.509
l0.504
diplom0.476
Suppressed:
own-1.035
此刻-0.573
గుర్త-0.532
熟悉的-0.526
ness-0.509