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 damp smell of the basement was usually unpleasant, but not today. Lisa, holding her breath, began the arduous task of sorting through her father's old keepsakes. It was a job she had been putting off for far too long.

Amongst the forgotten tools and forgotten tools, she found a small, battered metal box. Inside, amongst faded photographs and forgotten trinkets, was a neatly folded piece of paper. She unfolded it, her heart thumping in her chest. It was a draft letter of resignation, dated decades ago.

A peal of laughter burst from Lisaโ€™s lips. She felt a lightness, a bubbling sensation within her. She started clapping her hands, re-reading the letter with wide, amazed eyes. She practically skipped over to the window, throwing it open to let the fresh air flood in.

Emotion: exuberant

Cluster: Positive / Joy
PC1 (Valence): 3.42 Positive
PC2 (Disposition): -0.58

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
!0.676
de0.605
la0.596
!"0.551
๏ผ0.548
Suppressed:
S-0.892
๐Ÿ˜ž-0.474
ness-0.407
L-0.389
๐Ÿ˜”-0.376