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 attic stairs groaned under Amelia's weight, each creak amplifying the silence of the house. Sunlight sliced through a grimy window, illuminating dust motes dancing in the air. She’d always hated attics, the claustrophobia tightening her chest. She needed to find the Christmas decorations, but the task felt daunting.

A cardboard box, half-collapsed, sat in the corner. Curiosity, a dangerous thing at the moment, tugged at her. Inside, nestled amongst yellowed tissue paper, were handwritten letters tied with a faded ribbon. She reached for them, her fingers trembling slightly.

The letters spoke of love, loss, and a life lived in the house long before hers. She imagined the woman who wrote them, her spirit somehow still lingering in the dusty space. Amelia felt a pang of something akin to being exposed.

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
πŸ˜”0.277
😣0.258
Suppressed:
de-0.567
la-0.488
l-0.271
(!)-0.267
(!)-0.257