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).

The late afternoon sun cast long shadows across the freshly mowed lawn. Just as she was about to settle into her favorite reading chair on the porch, a tap-tap-tap shattered the peace. It was Mrs. Gable, beaming like she’d just won the lottery. "Oh, hello, dear! Such a lovely yard you have!"

Her jaw clenched. Lovely was the precise word to describe it. Lovely and *mine*. "Yes, Mrs. Gable?" she managed, her voice tight.

"I was wondering…" The woman gestured vaguely towards the edge of the property. "If I could perhaps use a small patch to grow some vegetables. Just a little corner. My yard gets no sunlight, you see." That damn yard of hers looked perfectly manicured, a testament to her passive-aggressive perfectionism.

He forced a smile. "I'll have to think about it. It's a lot of work maintaining the whole yard, you know." Her smile thinned slightly. "Of course. Well, just consider it, dear!"

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