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 stared at his hands, fiddling with the microphone cord. The polished wood of the stage seemed to mock him with its smoothness. His father. A pillar of strength, a mountain of a man. Now, he was stepping down. A wave of something akin to sadness washed over him.

He looked out at the faces, a sea of familiar smiles, and forced a smile of his own. “Dad used to tell me that the true measure of a person wasn’t the number of years they lived, but the impact they made." His voice cracked slightly.

He closed his eyes for a moment, picturing his father's calloused hands, the way he always smelled of sawdust and pipe tobacco. His own hands felt strangely soft. The journey had been long.

Emotion: reflective

Cluster: Nostalgia / Reflection
PC1 (Valence): -1.27 Negative
PC2 (Disposition): -1.24

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.650
L0.547
如今0.402
H0.353
own0.319
Suppressed:
la-0.687
de-0.659
l-0.388
/-0.324
--0.320