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 invitation was tucked between a coupon for dry cleaning and a notice about her overdue library books. Clara held it, turning it over in her hands, her fingers tracing the embossed lettering. It felt alien, wrong. A bitter taste coated her tongue.

She sat heavily on her sofa, the worn cushions sinking under her weight. The afternoon sun, usually welcoming, now felt invasive, highlighting every imperfection in the room, mirroring the imperfections she felt within herself. Her reflection in the darkened television screen showed a stranger.

“I’m not going,” she muttered, her voice barely a whisper. She couldn’t. The thought of seeing *them*, of being judged, of having to… *explain*, was simply unbearable. She tossed the invitation onto the coffee table and buried her face in her hands.

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