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).

“Another level, Dad!” Liam’s voice rang out, filled with triumphant glee. Liam was playing a video game he’d created, a game that had been based on his father's hobby of making them, a niche thing he called "interactive stories". Liam’s dedication and talent was beginning to make him feel quite…queasy. He looked at the wall and felt a cold sweat.

The screen lit up with bright colors. Liam's eyes, fixated on the screen, glowed with enthusiasm. Liam was building a world, a place he could control, a place where he could spend all his time.

"Liam, maybe it's time for a walk?" he asked, attempting a casual tone.

Liam didn’t pause, didn't answer. He was lost in his world. He wanted to go home. He closed his eyes, and put his hand on his heart.

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