irritated - Story 78245 | Emotion Vectors in Gemma4-31B

The phone felt heavy in Elena's hand, like a brick. She squeezed it, the plastic groaning under the pressure. The nurse had just informed her, in a blithe voice, that her mammogram results indicated a suspicious lump. The problem? Elena had already had a mammogram last month and her results were clear.

She took a deep breath, the air barely filling her lungs. “Excuse me, but I… I just had a mammogram last month, and everything was fine.” The words felt strangled, barely audible.

The nurse responded in the same saccharine tone: “Well, let me pull up your file, honey.”

"Honey?" That was the last straw. The woman's patronizing tone, combined with the utterly ludicrous situation, sparked an internal fire.

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:

`l`	0.370
`S`	0.282
`P`	0.264
`😠`	0.248
`aggravated`	0.233

Suppressed:

`own`	-0.300
`joyful`	-0.211
`joyous`	-0.207
`खोले`	-0.206
`newfound`	-0.205