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).

Rain lashed against the window of the coffee shop, mirroring the storm brewing inside Amelia. She'd been nursing a lukewarm latte, meticulously sketching in her notebook, when *he* walked in. She knew that smirk anywhere.

He strolled right over, tugging up his sleeve with a flourish. There it was: her vibrant, swirling "Cosmic Dust" inked onto his forearm. The man grinned, flashing teeth that seemed too white. Amelia's hand tightened around her pen. A pulse hammered in her temple.

"You wouldn't believe how many compliments I get," he said, his voice grating on Amelia's ears. "It's, like, a conversation starter, you know?" She wanted to scream. Instead, she stared at the droplets of rain tracing patterns down the glass, silently willing them to wash her away.

Emotion: annoyed

Cluster: Anger / Hostility
PC1 (Valence): -0.59 Negative
PC2 (Disposition): 1.27

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.376
😠0.251
骂0.249
🤬0.245
fucking0.239
Suppressed:
own-0.352
渗透-0.251
gradually-0.234
unfolding-0.226
超越-0.224