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 window, mirroring the storm brewing inside Amelia. She watched her niece, Clara, meticulously color the figure in the drawing, a tall, spindly figure with glowing green eyes. It was Barnaby, the creature Amelia had invented to combat loneliness when she was a child.

She’d discarded Barnaby the moment she'd found popularity in high school. The betrayal of her inner world was a constant weight, a silent ache. The memory of leaving Barnaby behind burned within her, a reminder of the person she once was and the person she'd become.

Amelia’s hands trembled as she reached for a cup of tea, the ceramic cold against her skin. "Barnaby's a good friend, isn't he?" she said, her voice strained. Clara looked up, her face lit with the joy that Amelia used to know.

Emotion: regretful

Cluster: Shame / Guilt
PC1 (Valence): -1.62 Negative
PC2 (Disposition): 0.76

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
L0.801
S0.774
H0.513
悔0.479
been0.473
Suppressed:
de-0.407
--0.320
随后-0.317
(!)-0.310
oldukça-0.301