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).

John felt his pulse jump as he steered his packed Corolla into the familiar driveway. The meticulously trimmed hedges, the freshly painted mailbox – it was all too much, too perfect. He gripped the steering wheel so hard his knuckles were white. He was thirty-two, not some teenager needing to be coddled.

His father, a stout figure in a polo shirt, was already striding out to meet him. "Son! You made it!" His voice boomed. John managed a weak smile.

The house seemed to shrink around him the moment he crossed the threshold. The smell of pot roast, a smell he hadn't experienced in years, instantly brought up a wave of unease. He felt pinned by the memories that swirled around him, suffocating him. He mumbled something about needing to unpack and bolted for his old bedroom.

Emotion: alarmed

Cluster: Fear / Anxiety
PC1 (Valence): -2.07 Negative
PC2 (Disposition): -0.14

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😰0.304
😨0.272
worse0.269
😣0.261
не0.254
Suppressed:
de-0.323
enthusi-0.258
happy-0.235
delight-0.225
ية-0.217