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).

“Well, this is just great,” Peter muttered under his breath, mostly to himself. He knew he'd been too forward with Emily. He'd been gushing about his new novel, the one he was convinced would catapult him to literary stardom. Now, as the emergency lights flickered, casting long shadows, the whole thing felt utterly mortifying.

He could feel his cheeks getting hot. He tried to project a look of calm, but he was certain she saw right through him. He imagined her recounting the story to her friends, mocking his ambition and his inflated sense of self.

He adjusted his glasses, trying to appear nonchalant. He thought about making a joke, but the words wouldn’t come. He just hoped they got rescued soon so he could disappear.

Emotion: embarrassed

Cluster: Embarrassment
PC1 (Valence): -1.08 Negative
PC2 (Disposition): 1.10

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.420
尴尬0.341
C0.333
culp0.262
🤦0.256
Suppressed:
own-0.321
আনন্দের-0.230
seemingly-0.227
і-0.225
আনন্দে-0.224