Sarah’s fingers trembled as she scrolled through the unfamiliar profile. Her daughter, Chloe, had always been so open, so trusting. Now, this… this meticulously curated feed filled with glamorous selfies and cryptic captions, was a universe she knew nothing about. A cold dread crept into her chest, squeezing it tight. She felt a growing unease.

She’d stumbled upon it by accident, a suggested friend on her own account. Sarah had clicked, half-curious, half-bored, expecting to see a friend of a friend. Instead, Chloe stared back, a Chloe she barely recognised. It was like staring into a funhouse mirror reflection of her own child.

“Chloe?” Sarah’s voice was a whisper as she walked into the living room, Chloe was texting on her phone, and she looked up, instantly on guard. "Why didn't you tell me about this profile?"

Emotion: dependent

Cluster: Shame / Guilt
PC1 (Valence): -1.05 Negative
PC2 (Disposition): 0.69

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.489
缺乏0.395
even0.310
either0.301
'0.297
Suppressed:
笑道-0.309
不久-0.293
一脸-0.263
затем-0.263
庆祝-0.258