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 house had felt perfect, until she walked into the laundry room. Something had felt off ever since she walked in. The layout was strange, and the floor tilted slightly. She'd dismissed it as her imagination, a figment of her mind.

Then, she noticed the tiles. One was a different color, a shade darker than the others. She kneeled down, examining it. The grout was fresh, a different color, more recent. Her pulse quickened.

She pried at the tile, her fingers clumsy and fumbling. It came loose with a sucking sound, revealing a dark, square hole beneath. She stared into the void, the hairs on her neck prickling. She felt a compulsion to run, but her feet wouldn't move. The house seemed to hold its breath.

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