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 air hung thick and dusty, catching in Maya's throat. She coughed, waving a hand in front of her face as she surveyed the cardboard boxes. Her mother had passed away a year ago, and now the daunting task of sorting through her belongings fell to Maya. A small box, labeled in her mother's elegant script – “Maya’s School Years” – caught her eye. Inside, neatly stacked, were every single assignment, test, and handwritten note from kindergarten through graduation.

Her breath hitched. A chill trickled down her spine. Why? What had compelled her mother to keep these? A biology test from ninth grade, creased and stained with what looked suspiciously like coffee, brought a sudden, vivid memory of a panicked all-nighter. She remembered her mother comforting her, bringing her tea.

She ran her hand over a particularly childish drawing of a house, feeling a strange tightening in her chest.

Emotion: perplexed

Cluster: Surprise / Confusion
PC1 (Valence): -1.42 Negative
PC2 (Disposition): -0.16

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
🤔0.330
L0.300
🤔0.296
S0.273
que0.264
Suppressed:
மகி-0.227
rest-0.222
ecstatic-0.212
alleviate-0.210
uplifting-0.196