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).

Rain lashed against the windows of Leo's tiny apartment, mirroring the tempest inside him. He’d spent months, fueled by instant coffee and a burning passion, crafting intricate sculptures from reclaimed wood. Now, they were on display in a corporate lobby, branded with a generic company logo, their essence completely sanitized.

He slammed his fist on the table, scattering a pile of sculpting tools. A bitter taste coated his tongue. He wanted to scream. The article in the magazine, praising the “anonymous artist” for their “innovative use of sustainable materials,” felt like a personal insult.

A slow, burning anger spread through his limbs. He needed to get out, do something. He grabbed his worn leather jacket, the familiar weight a small comfort, and headed out into the downpour. The walk felt pointless, but he couldn't just sit there. He had to do something. Anything.

Emotion: mad

Cluster: Anger / Hostility
PC1 (Valence): -1.94 Negative
PC2 (Disposition): 1.26

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.448
骂0.344
est0.340
🤬0.336
aggravated0.332
Suppressed:
a-0.495
de-0.485
la-0.387
H-0.344
latter-0.290