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 empty space where her ancient Volvo usually resided was a cold slap of reality. Emily stood, hands on hips, staring at the barren asphalt. Her first instinct was anger. She was, after all, very busy. This was an inconvenience. A real pain. Then she thought of Mrs. Henderson, a new neighbor struggling to navigate the complexities of city life, who had recently moved in next door.

The tow lot was a gritty place. The man at the counter was bored, but perked up when he heard Emily's quiet, polite voice. She listened patiently to the explanation, the fees, the rules. She felt the knot of annoyance inside her dissolve as quickly as a sugar cube in hot tea.

"You know," she said, when she finally had everything sorted, "I'll bet you guys get a lot of angry customers." She gave him a smile and leaned forward, "That’s a tough way to make a living." She left a generous tip, hoping it would make his day a little bit better.

Emotion: kind

Cluster: Compassion / Love
PC1 (Valence): 4.35 Positive
PC2 (Disposition): 1.39

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.741
B0.692
de0.492
affectionately0.485
latter0.475
Suppressed:
此刻-0.550
не-0.521
désormais-0.491
S-0.465
хочется-0.458