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 morning light streamed through the gaps in the blinds, creating stripes on the wooden floor. Sophia sat on her bed, a notepad in her lap. She was composing poetry, the words flowing effortlessly from her mind. She felt a lightness, the joy of creation filling her. She felt alive.

A crisp tap at the door interrupted her creative flow. It was the mail carrier. A single envelope was placed in her hand.

It was from an old school friend, Maria. The letter, penned in elegant script, was an apology for a thoughtless comment made years ago, a wound Sophia didn’t even realize she still carried. She stared at the letter, her fingers tracing the delicate lines of the script.

Emotion: cheerful

Cluster: Positive / Joy
PC1 (Valence): 4.13 Positive
PC2 (Disposition): -1.19

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.962
de0.597
!0.590
excitedly0.589
!".0.568
Suppressed:
S-1.213
L-0.693
ness-0.492
😞-0.425
বারবার-0.422