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).

"Are you serious?" Beatrice demanded, her voice dripping with disbelief. The seller, a young man with a bored expression, merely shrugged. Her bear. Bartholomew. His button eyes, his threadbare fur, everything was like a slap in the face.

She felt a shudder run through her body. He’d been everything to her. He had been her confedante, her comfort, her only safe place. He was now just an object, a relic.

"What's the story here?" she asked, her voice cracking. "Was he just not good enough to keep?" The boy yawned. "Lady, I don't know, I just sell the stuff." Beatrice stared at him, at the careless indifference in his gaze. She pulled out a ten dollar bill and thrust it at him, taking Bartholomew as if she were rescuing a hostage.

Emotion: scornful

Cluster: Anger / Hostility
PC1 (Valence): -0.76 Negative
PC2 (Disposition): 2.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.406
C0.391
S0.276
I0.268
愤0.243
Suppressed:
own-0.396
unexpectedly-0.288
Suddenly-0.260
soon-0.256
serendip-0.256