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).

Rain lashed against the window of the old cottage, mirroring the tempest brewing inside Ben. He’d been expecting a simple clean-out, a quick disposal of his father’s belongings. Instead, he was wading through a sea of sentimental junk. A fishing rod, a collection of dusty vinyl records, and now, this… a brightly coloured list titled “My Last Adventures.”

He gritted his teeth, scanning the handwritten goals. “Climb Mount Everest.” “Write a novel.” “Learn to dance the tango.” He tossed the list aside, the paper hitting the cluttered desk with a resentful thud.

His jaw clenched. The audacity. This was what his father *really* wanted to do? Why couldn’t he have just left behind some helpful documents or at least a will that didn’t involve an endless legal battle? He grabbed a box, shoving the contents in haphazardly, the old grievances resurfacing with every item.

Emotion: irritated

Cluster: Anger / Hostility
PC1 (Valence): -0.86 Negative
PC2 (Disposition): 1.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.370
S0.282
P0.264
😠0.248
aggravated0.233
Suppressed:
own-0.300
joyful-0.211
joyous-0.207
खोले-0.206
newfound-0.205