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 rejection letter, crisp and official, lay on the kitchen table. Sarah barely glanced at it. She was already halfway through a spontaneous kitchen dance, humming a tuneless melody while she scrambled eggs. The sunlight, streaming through the window, seemed to have amplified the colours of everything – the vibrant green of the basil, the fiery red of the tomatoes she was about to slice. A laugh bubbled up from her chest as she twirled, nearly knocking over a stack of cookbooks. This was going to be a good day, rejection or not.

Emotion: invigorated

Cluster: Positive / Joy
PC1 (Valence): 3.28 Positive
PC2 (Disposition): 0.00

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
!0.452
(!)0.439
🎉0.434
invigor0.416
enthusi0.395
Suppressed:
😞-0.524
😔-0.428
😢-0.404
😥-0.393
que-0.391