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).

Sarah ran her finger along the smooth surface of the grand piano in the living room, a relic of her childhood. The instrument’s presence in the otherwise sparsely furnished room felt out of place. It was here, yet seemed so disconnected from her current life. The air hung thick with the scent of old wood and something else, something she couldn’t quite name, but that spoke of history.

Her father entered, his gait slower than she remembered. He seemed smaller, somehow. He smiled at her and said, "Remember that time you tried to teach the cat to play?" Sarah returned his smile. It was forced.

A knot tightened in her stomach. The house, once a sanctuary, now felt like a stage set for a play she didn't know the lines to. The silence wasn’t peaceful now, but strained, a constant reminder of the unfinished business between them. Sarah felt a peculiar ache behind her eyes.

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