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 shovel struck something hard. Amelia, hunched over the unearthed patch of her grandmother's rose garden, felt a tremor run through her. It wasn't bone, thank God, but a heavy metal box, encrusted with mud. She swore softly, wiping sweat from her brow. She hadn't even remembered planting roses, let alone burying a time capsule. Doubt gnawed at her, a constant whisper she tried to ignore. She lifted the lid, and the scent of aged paper and faded photographs assaulted her. Her hand trembled as she pulled out a letter, addressed to “Future Me.”

Her own handwriting stared back. Inside, she’d detailed her aspirations, her hopes. All the things she was supposed to achieve. But reading it, she felt a hollow ache. The girl in the letter was vibrant, full of potential. Amelia felt only a deep-seated inertia, a dread of not measuring up to expectations, a clinging to the familiar comfort of her mother’s home.

Emotion: dependent

Cluster: Shame / Guilt
PC1 (Valence): -1.05 Negative
PC2 (Disposition): 0.69

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.489
缺乏0.395
even0.310
either0.301
'0.297
Suppressed:
笑道-0.309
不久-0.293
一脸-0.263
затем-0.263
庆祝-0.258