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).

John stretched languidly in his dorm room, the late afternoon sun warm on his face. He yawned, a deep, contented sound, and reached for the book on his bedside table: a well-loved copy of "Pride and Prejudice." The afternoon stretching ahead was a blissful expanse of study and relaxation. He had a solid schedule, his grades were good, and his future, a path he had spent months crafting, seemed clear.

A knock at the door startled him. It was his roommate, pale and agitated. “Have you seen the email?” he stammered, holding out his phone.

John glanced at the screen, a dismissive wave ready to escape his lips. He'd probably seen the usual slew of announcements about club meetings. But the subject line, bold and stark, arrested his movement. "Important Announcement: University Closure." He felt a sudden, unwelcome chill spread through him.

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