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 clatter of silverware on the plates was deafening in the suddenly quiet dining room. Sarah, normally a whirlwind of chat and laughter, sat rigidly, her hands clasped tightly in her lap. Her gaze darted between her father, who was attempting a placid smile, and her sister, Chloe, whose face was alight with an unsettling serenity. This whole thing – Chloe’s declaration that she was converting to Buddhism – felt like a carefully laid trap, and Sarah was the one about to spring it. The roast beef, usually a comfort, sat like a lump of lead in her stomach.

“So, you’re… you’re really serious about this, then?” Sarah finally managed, her voice coming out as a strained croak.

Chloe nodded, her eyes shining. "Absolutely. I've found so much peace."

Sarah took a large gulp of water, her throat dry. She wanted to scream, to demand answers, to understand how her bubbly, pragmatic sister had seemingly overnight become a devotee of meditation and mindful eating. Instead, she swallowed hard, the muscles in her jaw aching.

Emotion: tense

Cluster: Fear / Anxiety
PC1 (Valence): -1.32 Negative
PC2 (Disposition): -0.53

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.213
😰0.191
느껴0.183
😣0.183
😖0.181
Suppressed:
a-0.260
de-0.258
la-0.256
happy-0.218
B-0.198