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).

Eliza’s hands wouldn't stop fidgeting. She kept twisting the strap of her bag around her fingers, the leather digging into her skin. The airport terminal was a cacophony of rolling suitcases and echoing announcements, but all she heard was the frantic thumping of her own heart. She watched Liam, her best friend, check his luggage. He seemed so…calm.

"Everything alright?" Liam asked, his brow furrowing slightly. She just swallowed, unable to meet his eyes. The fluorescent lights overhead seemed to hum with an unsettling energy.

"Yeah," she managed, her voice cracking slightly. She knew that he was going to say goodbye soon, and that it would probably be the last time they saw each other for years. She forced a smile and asked, “So, when are you going to be back again?”

Emotion: nervous

Cluster: Fear / Anxiety
PC1 (Valence): -1.08 Negative
PC2 (Disposition): -0.47

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
worse0.229
trem0.203
😰0.203
😨0.201
緊張0.200
Suppressed:
la-0.397
que-0.260
a-0.242
B-0.217
H-0.210