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 old man’s office was locked, the brass plate on the door glinting ominously in the weak sunlight. "Retired," the new sign announced. It was a cold, hard fact, written in sterile, impersonal font. He felt a sudden, sharp pinch in his chest.

He wandered the halls, the familiar sounds of the institution now strangely muted. He replayed their conversations in his mind, searching for clues, for warnings, for anything that would explain the abrupt departure. He couldn't find any.

The absence of his mentor was a palpable thing, a vacuum that sucked the air from his lungs. He wanted to scream. He wanted to cry. He was alone.

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