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 faintly of lavender and old paper. Sarah fidgeted, smoothing down the non-existent wrinkles on her skirt. Her palms were damp, and she surreptitiously wiped them on her thighs. Each tick of the clock felt like a hammer blow against her skull. She'd been putting this off for weeks, but the panic attacks were getting worse.

The door opened. "Sarah Miller?" The voice, smooth and professional, sent a jolt of recognition through her. It was Mark, the bane of her existence in elementary school. The same cruel eyes, though they held a different expression now, stared back at her.

"Come in," he said, gesturing towards a large, comfortable armchair. She stumbled forward, her knees suddenly weak, and sank into the cushions. The room seemed to shrink around her.

Emotion: vulnerable

Cluster: Fear / Anxiety
PC1 (Valence): -2.07 Negative
PC2 (Disposition): -0.69

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.327
ness0.319
😞0.299
😔0.277
😣0.258
Suppressed:
de-0.567
la-0.488
l-0.271
(!)-0.267
(!)-0.257