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 sterile smell of the hospital room seemed amplified. Sarah leaned back against the stiff pillows, the papers in her hands feeling heavy as lead. They detailed a history of allergies, procedures, and conditions she knew nothing about. Someone else’s life, filled with a foreign language of medical jargon, now rested within her grasp. A life that belonged to a man named Arthur Bellweather.

Her gaze drifted to the window, where the city skyline seemed to blur and soften. A feeling of… wonder, perhaps? Certainly, not horror. The unfamiliar information revealed a narrative of resilience, a struggle she hadn’t known existed. The very act of reading it felt like an intrusion, a violation, yet also… a privilege.

She traced the contours of the unfamiliar documents, her fingers brushing against the cool paper. A tremor ran through her, making her heart pound against her ribs. Arthur's details, a portrait of someone else's pain, yet also his strength, was now hers.

A knock came at the door. "Ms. Miller?" a nurse asked. "Dr. Chen is ready to see you."

Emotion: awestruck

Cluster: Surprise / Confusion
PC1 (Valence): -0.17 Negative
PC2 (Disposition): -1.20

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
L0.271
suddenly0.260
stranger0.252
plötzlich0.246
wonder0.242
Suppressed:
la-0.450
싶-0.350
K-0.319
rest-0.248
싶은-0.238