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).

He cleared his throat, the sound unusually loud in the quiet study. The letter was a pale blue, a color his father hadn't worn in years. It was addressed to the university, confirming his father’s departure from the prestigious history department. He’d always imagined his father as a pillar of academic success, a bastion of knowledge. Now this… this fragile piece of evidence.

James stared at the carefully crafted prose. The meticulously chosen words were a stark contrast to the casual way his father spoke. He felt a chill run through him, a coldness that had nothing to do with the room's temperature.

He ran a hand through his hair, a nervous gesture he’d picked up from his father. What had made him do it? What had he been running from? The questions swirled in his head, a confusing tempest. He was suddenly overwhelmed, and felt a strong urge to leave the room.

Emotion: mortified

Cluster: Embarrassment
PC1 (Valence): -1.36 Negative
PC2 (Disposition): 0.79

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.406
尴尬0.257
C0.252
worse0.245
逃0.242
Suppressed:
own-0.340
了他-0.218
他对-0.203
ል-0.194
楽し-0.193