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).

Claire sat on the porch swing, the fading light painting the sky in hues of orange and purple. She had been exploring the contents of a box of old photographs, a task she had been putting off for weeks. She knew the contents would be emotionally challenging. She was right.

One photograph revealed her grandmother, a woman she remembered as the matriarch, a stern, unyielding figure, holding a baby. She looked… terrified. It was a look she had never seen before. Beside her, a man with a cruel smile.

A strange, unfamiliar feeling coursed through Claire, a ripple of unease. The woman in the photo wasn't the grandmother she had known. The story, as she knew it, was not the full truth. A new awareness, a quiet melancholy, settled over her.

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