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 scent of his aftershave, which he’d supposedly stopped wearing months ago, clung to the air in the hallway. Sarah found the bottle, half-empty, hidden beneath a pile of old socks in the back of the linen closet. A knot tightened in her stomach, a familiar ache she'd come to associate with his absences. She slammed the closet door, the sound echoing a silent accusation in the otherwise tranquil house.

Later, at dinner, she watched him expertly fillet the salmon, the casual grace of his movements grating on her nerves. “That was a good call,” he’d said earlier, referring to the phone conversation she'd accidentally overheard. His voice was bright, too bright, like a poorly-lit stage.

She pushed her food around her plate with a fork, the silence between them thick with unspoken resentments. "Who was it?" she finally asked, her voice flat, the question hanging in the air like a poorly concealed secret. He wouldn’t meet her eyes.

Emotion: contemptuous

Cluster: Anger / Hostility
PC1 (Valence): -0.40 Negative
PC2 (Disposition): 2.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.502
l0.423
fact0.287
either0.265
👎0.262
Suppressed:
own-0.367
unexpectedly-0.303
serendip-0.265
Suddenly-0.253
previously-0.250