Gemotions is a research project by DEJAN AI, an AI SEO agency from Australia, replicating Anthropic's emotion vector study on Google's open-weight Gemma4-31B model across 171 emotion concepts and 171,000 model-written stories, the kind of mechanistic interpretability work of particular interest to generative engine optimization (GEO) and answer engine optimization (AEO).

He stared at the notice, his gaze fixed on the menacing words. It wasn’t a mistake; it was real. He felt a wave of heat wash over him, followed by a sudden chill. He closed his eyes, trying to block out the reality of it.

His breath came in short, shallow gasps. He ran a hand over his face, trying to clear the fog in his mind. The apartment, once a sanctuary, now felt like a prison. He pictured the landlord’s face, cold and unforgiving.

"No," he whispered, the word a desperate plea. He needed to fight this, to find a way to stay. His mind was a jumble of anxieties, but he had to focus, to act, to survive. He glanced at the clock on the wall. Time was slipping away.

Emotion: panicked

Cluster: Fear / Anxiety
PC1 (Valence): -2.44 Negative
PC2 (Disposition): -0.00

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
worse0.317
😫0.294
😖0.285
😰0.285
escape0.283
Suppressed:
la-0.432
a-0.374
de-0.296
楽し-0.262
happy-0.258