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 news electrified him. The transfer to the Graphic Design department was a dream come true for Mark. He'd spent his days in the IT department, which was fine, but his heart had always belonged to art, to the creative process. A tremor of pure happiness shot through him.

He cleared his throat and felt compelled to sit at his desk and re-examine the email. He re-read the words, the department name, the start date. He knew he wanted to start making a difference immediately.

He imagined the vibrant colors, the collaborative atmosphere, the freedom to express himself visually. His hands began itching to pick up a pencil and start sketching. He even considered asking to be placed in a more creative position in advance.

Emotion: joyful

Cluster: Positive / Joy
PC1 (Valence): 3.96 Positive
PC2 (Disposition): -0.67

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.615
de0.600
!0.572
excitedly0.552
enthusi0.533
Suppressed:
S-0.893
ness-0.513
😞-0.470
😣-0.337
L-0.334