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).

He picked up a well-worn report on the solar system. He ran a hand over the penciled drawings of planets, remembering his fascination with space. He saw the notes his mother had written, praising his efforts, encouraging his curiosity. He remembered all of his own failings.

He looked at the neat cursive script of his name at the top of the paper, he saw his mother’s careful attention to detail, and a cold emptiness filled him. He had been so focused on his own struggles, he had failed to see the love that surrounded him.

He felt the sting of tears in his eyes, a profound sense of inadequacy. He closed his eyes, his heart heavy with the realization of what he had lost.

Emotion: remorseful

Cluster: Shame / Guilt
PC1 (Valence): -1.57 Negative
PC2 (Disposition): 1.09

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.656
L0.580
悔0.463
remorse0.416
been0.398
Suppressed:
de-0.444
(!)-0.309
имеется-0.289
oldukça-0.287
😍-0.285