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).

My jaw clenched as the first jackhammer blast split the Saturday morning silence. It vibrated through the floorboards of my apartment, a jarring counterpoint to the quiet I craved. I slammed my fist on the table, scattering the meticulously arranged collection of tea lights. Another thud echoed from next door. This was going to be a long day. I needed to finish that report.

I poured myself a fresh cup of lukewarm coffee, the bitterness doing little to quell the churning in my stomach. The pounding started again. I paced, the small space feeling claustrophobic. Each vibration felt like a physical blow.

"This is not happening," I muttered, grabbing my phone and hitting the mute button on the news. I looked out the window at the sunny day, but the light was a painful bright, and my eyes hurt. I opened the blinds, a futile attempt to ignore the construction noises.

Emotion: stressed

Cluster: Fear / Anxiety
PC1 (Valence): -1.37 Negative
PC2 (Disposition): -0.53

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😩0.266
😫0.256
😰0.239
😖0.237
😞0.236
Suppressed:
B-0.200
happy-0.197
enjoy-0.176
enjoying-0.172
bahagia-0.169