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 waiting room smelled of stale coffee and desperation. The other candidates fidgeted, each feigning nonchalance while surreptitiously glancing at one another. I sat ramrod straight, my gaze fixed on a blurry abstract painting that offered no distraction. They were all grinning too much.

I picked at a loose thread on my skirt, pulling it until it snapped. The sound was surprisingly loud. I felt a surge of… something. I looked at the receptionist’s smiling face. She was probably getting her fill of the whole charade.

I ran my hand across my forehead; a slight dampness had appeared. Across the room, I caught sight of David, my supposed friend, flipping through a magazine. His posture was too relaxed, his smile too wide. I wanted to scream.

Emotion: hostile

Cluster: Anger / Hostility
PC1 (Valence): -1.67 Negative
PC2 (Disposition): 1.56

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.592
S0.387
骂0.324
est0.301
aggravated0.295
Suppressed:
a-0.457
de-0.375
H-0.289
🤩-0.280
la-0.273