The clatter of the teacup against the saucer was sharper than usual, echoing in the otherwise quiet kitchen. Aunt Carol had just finished her pronouncements about finding true spiritual fulfillment, and the silence afterwards felt thick, like stale air. My jaw clenched as I met her gaze, a forced smile plastered on my face. It was the forced nature of her new-found zeal that got to me. She was so… buoyant.

“So, you’re saying you’re…Buddhist now?” The words were carefully chosen, like navigating a minefield. Each syllable felt heavy in my mouth. I wanted to just go back to my book.

Her smile widened, and she started in on a story about the peace she had found, which was apparently more profound than the rest of us could ever dream of. I picked at a loose thread on the tablecloth. I could practically *feel* the tension building in my temples.

Emotion: irritated

Cluster: Anger / Hostility
PC1 (Valence): -0.86 Negative
PC2 (Disposition): 1.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.370
S0.282
P0.264
😠0.248
aggravated0.233
Suppressed:
own-0.300
joyful-0.211
joyous-0.207
खोले-0.206
newfound-0.205