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 rhythmic thudding was the final straw. Sarah flung the remote control across the living room, missing the television and hitting the wall with a sickening thwack. Her tea had gone cold, and her carefully planned afternoon of relaxation was now a shambles.

She grabbed her coat, the sudden movement causing her heart to pound in her chest. She had to *do* something. The tremors had begun in her hands, her fingers twitching involuntarily. Her teeth were clenched so hard she thought she might crack a molar.

Downstairs, the air smelled of fresh-cut wood and something else, something acrid that made her nose wrinkle. She felt a tremor in her knees. There was Mr. Davis, whistling cheerfully, surrounded by a mountain of debris. He looked up, his smile wide and oblivious. "Just making some improvements!" he boomed.

Emotion: outraged

Cluster: Anger / Hostility
PC1 (Valence): -1.96 Negative
PC2 (Disposition): 1.32

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.467
恨0.334
🤬0.327
aggravated0.323
骂0.312
Suppressed:
H-0.315
☺️-0.286
optimistic-0.278
adventurous-0.262
मुस्कुरा-0.260