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 aroma of cinnamon wafted from the kitchen, where Martha hummed along to the radio, stirring a pot of applesauce. Sunlight streamed through the large window, illuminating dust motes dancing in the air. This time of day, this task – it always felt…right. No hurry, no worries. Just the low thrum of the refrigerator and the gentle bubbling of the fruit. She glanced at the worn, floral wallpaper, a familiar comfort. Tomorrow was the monthly book club, and she had a new recipe to try.

Her phone buzzed on the counter. A notification from the landlord. She opened it, a little frown creasing her brow. A meeting… regarding the building. She sighed. These things always meant some kind of upheaval. But the familiar kitchen, the familiar routine, it felt like a sturdy anchor in a shifting world.

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