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 sterile white walls of the hospital room felt suffocating. Liam clutched the drawing his nephew, Finn, had made. It was a perfect depiction of his childhood friend, a small, furry creature with a crooked grin named Sparkle.

Liam had stifled Sparkle's existence in order to be taken seriously. That act, that rejection, was a persistent echo in his life, and now here was Sparkle again, a testament to what he'd discarded. He'd traded his friend for success, for respect. But the emptiness he felt now was a constant reminder.

He looked at the drawing, a fresh wave of something akin to sadness washing over him. He felt hollow, as though he'd abandoned a piece of himself. He sat down heavily, the weight of the past settling upon him.

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