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 world felt too loud. Too much. Sarah sank onto the sofa, the cushions doing little to absorb the turmoil inside her. The noise from the party next door, a cacophony of laughter and music, was like a physical blow. She wrapped her arms around herself, as if trying to hold her scattered pieces together.

She needed calm, needed quiet. Just a moment of peace. A blank screen, a silent world. She pictured herself disappearing into a deep, silent cave, a place where no one could touch her.

Her phone began to buzz again – a text from Mr. Henderson, the eternally grumpy neighbor. Another complaint. Another battle. She took a shuddering breath and picked up the phone, her fingers trembling. "Yes, Mr. Henderson," she said, her voice strained. The battle began.

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