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).

Sarah sank onto the sofa, the crisp white paper of the bucket list crinkling in her grasp. Each item was a challenge, a vibrant splash of color against the drab canvas of her own life. “Run a marathon.” “Learn to speak Italian fluently.” “Volunteer at an animal shelter.” Her partner, Emily, had never mentioned any of this.

Her breath hitched in her throat. The apartment, once a haven of shared meals and quiet evenings, now felt alien, the furniture mocking her with its stillness. She felt a deep, heavy feeling in her chest. She reached for the remote, finding herself unable to focus on anything.

Sarah flicked through the channels, the images blurring into a meaningless stream. Her eyes burned, but no tears fell. Emily’s list felt like a mountain she could never hope to climb. She needed air.

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
😖0.319
😣0.319
😩0.314
Suppressed:
la-0.576
de-0.496
a-0.297
happy-0.290
l-0.264