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).

She had been avoiding the mailbox for days. The emptiness she found there was a mirror reflecting back her current state. When she finally forced herself to check, the usual deluge of advertisements and overdue notices awaited. And then, a card. Unsigned.

Her hand trembled as she opened it. Inside, a simple message, “Happy Birthday.” The words seemed to mock her. A date she didn’t even acknowledge anymore, a milestone passed with barely a ripple in her stagnant existence. She wondered who sent it, a forgotten acquaintance perhaps, or a well-meaning relative, now distant.

The apartment felt cavernous, the silence amplified by the dripping faucet. She let the card fall to the floor, the cheerful image of a party clown a cruel contrast to her reality. She sank onto the couch, pulling the threadbare blanket tighter, a shield against a world that seemed to forget her.

Emotion: worthless

Cluster: Sadness / Despair
PC1 (Valence): -2.10 Negative
PC2 (Disposition): -0.15

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
L0.427
S0.405
😞0.386
ness0.339
缺乏0.326
Suppressed:
de-0.320
(!)-0.286
随后-0.268
😍-0.262
tc-0.255