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 late afternoon sun warmed Amelia's back as she sat on the porch swing, the rhythmic creak a comforting soundtrack to her quiet grief. Her father's old house, filled with the scent of pipe tobacco and lemon polish, held her captive in a world of memories. He'd only passed a month ago. Her fingers, stained with ink from poring over legal documents all day, felt strangely light as she reached for the small, leather-bound notebook. She’d found it tucked away in the attic, amongst his war medals. Its cover was cracked and faded, but the handwriting inside, a familiar scrawl, was unmistakably his.

She chuckled as she scanned the first entry: “Learn to play the ukulele.” The next one made her smile widen: “Finally try that habanero pepper.” Amelia knew her father's aversion to spicy food. She flipped through the pages, a gentle curiosity blooming in her chest.

Emotion: safe

Cluster: Calm / Serenity
PC1 (Valence): 1.86 Positive
PC2 (Disposition): -2.55

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
soon0.402
affectionately0.357
disfrut0.329
необходи0.325
🥰0.317
Suppressed:
S-0.474
C-0.416
L-0.408
worse-0.379
even-0.370