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 slam of the door vibrated through the floorboards. Agnes sighed, the sound a low, breathy expulsion of air. She turned off the music, the silence immediately amplifying the gentle hum of the refrigerator. A shadow fell across the sunlit kitchen, and she knew without looking who was standing there. Mr. Henderson, his face a perpetual frown etched deeper by the years.

She set down her spatula. The weight of it, suddenly, felt immense. She’d been trying to make something – a cake, maybe? – but the ingredients seemed to mock her from the countertop, a chaotic jumble of flour and sugar. The scent of vanilla, once a comfort, now felt cloying.

"Sorry, Mr. Henderson," she mumbled, her voice surprisingly faint. She stared at the floor, picking at a loose thread on her apron. She couldn’t meet his gaze. It was easier to look down.

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