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 constant noise of the city, the incessant demands of the job, the ever-present feeling of being behind – all of it swirled inside her. The sun beat down, turning the afternoon into a hazy, oppressive heat. A knock on the door, persistent and unwelcome. It was her neighbor, Ms. Chen.

"Hello," Ms. Chen said, a bright smile gracing her face, "about the sycamore...the roots are causing problems with my fence. I'm wondering if we might..."

The words faded away. It was all a blur. Her chest felt tight.

β€œLater,” she mumbled, her voice flat. She quickly closed the door, leaning back against it, her body trembling.

Emotion: overwhelmed

Cluster: Fear / Anxiety
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -0.94

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.385
😫0.332
πŸ˜–0.319
😣0.319
😩0.314
Suppressed:
la-0.576
de-0.496
a-0.297
happy-0.290
l-0.264