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 floral arrangements seemed to mock Eleanor with their vibrant colors. She’d meticulously chosen this shade of lavender for her own wedding, a wedding that never happened. Now, here she was, at her friend’s celebration, enduring the well-meaning questions about why she was single, and the sympathetic glances. The atmosphere of joy was like a physical pressure, squeezing the air from her lungs.

She saw him across the dance floor. Michael, holding hands with a girl she’d never seen before, her back arched as she danced. Eleanor felt a hot flush rush over her skin. The music, which had sounded pleasant moments ago, now grated on her nerves. She abruptly excused herself, using a bathroom break as an excuse.

Emotion: troubled

Cluster: Fear / Anxiety
PC1 (Valence): -2.20 Negative
PC2 (Disposition): -0.81

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.343
S0.317
ness0.302
😔0.294
😢0.272
Suppressed:
de-0.553
la-0.511
l-0.307
/-0.260
(!)-0.236