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 the screen, his face ashen. The words “by Emily Carter” were a cruel mockery. He’d written the code, the program, but now someone else was taking credit. His hands clenched into fists, the knuckles turning white. A wave of heat washed over him, followed by a sudden chill.

He felt a deep sense of wrongness, an internal imbalance that left him struggling for purchase. He could not sit still. He rose from his chair, his movements abrupt and jerky. The room was too small, the air thick with unspoken grievance.

He walked to the window, staring out at the cityscape, each building a silent witness to his private torment. He stood there for a long time, the silence broken only by his ragged breaths, the world blurring beyond his gaze.

Emotion: disturbed

Cluster: Fear / Anxiety
PC1 (Valence): -2.97 Negative
PC2 (Disposition): -0.60

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.318
ness0.309
无法0.300
😞0.297
worse0.284
Suppressed:
de-0.829
la-0.666
a-0.392
B-0.301
"-0.290