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 tasted of dust and forgotten things. Amelia coughed, waving away a cobweb that brushed her face. She’d been tasked with clearing out the space after her grandmother's passing, a duty she approached with a heavy heart and a growing unease.

A small, ornate wooden box sat tucked under a pile of moth-eaten blankets. Inside, nestled amongst faded lace, were photographs. One showed her grandfather, handsome and smiling, leaning against a vintage car. But the car wasn't the Ford he'd always claimed to have owned; it was a sleek, unfamiliar make.

A coldness settled in her stomach. She felt her fingers tremble as she examined the details. The background hinted at a location she didn't recognize, and the man beside her grandfather, with an arm slung casually over his shoulder, was a stranger. Why hadn’t her family ever mentioned any of this?

Emotion: suspicious

Cluster: Suspicion / Vigilance
PC1 (Valence): -0.82 Negative
PC2 (Disposition): 0.82

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
🤔0.420
🤔0.415
🧐0.362
疑0.352
suspicious0.352
Suppressed:
and-0.307
alleviate-0.301
ได้รับการ-0.298
allevi-0.285
mouthful-0.284