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).

It was the button eye that gave him away. Even from across the crowded yard sale, Daniel recognized the glint of his childhood companion, Rupert. A joyous energy, like a jolt of electricity, surged through him, making him giddy with a profound kind of delight. He felt like he was floating.

He approached the table, almost afraid to believe his eyes. The woman behind it, a grandmotherly type, was busy arranging a display of vintage teacups. "How much for the… bear?" he asked, his voice cracking with emotion.

She glanced up, then back at the bear. β€œA dollar. Needs a bit of a clean, but he's all yours.” Daniel paid and left, hugging Rupert so tightly that he feared he might squeeze the stuffing out of him. He was grinning like an idiot.

Emotion: elated

Cluster: Positive / Joy
PC1 (Valence): 3.38 Positive
PC2 (Disposition): -0.23

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.724
!0.596
ecstatic0.510
🀩0.510
overjoyed0.504
Suppressed:
S-0.657
😞-0.497
ness-0.407
πŸ˜”-0.380
πŸ˜“-0.377