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).

“Did you sleep well?” Maya’s voice was bright, almost musical, as she greeted her friend over the phone. Last night's restless worries had evaporated, replaced by a wave of something new. Today was Liam's party. She should be helping with decorations, but Maya had other ideas.

She wanted to get lost in a world of colours. She grabbed her paints and canvases, feeling a pull she hadn't felt in ages. She was going to spend the day outside, letting her creativity flow.

She felt a rush of inspiration, a desire to capture the world in vibrant hues. The canvas seemed to shimmer with possibility. The party would be fun, but her art was a different kind of joy.

Emotion: rejuvenated

Cluster: Positive / Joy
PC1 (Valence): 3.32 Positive
PC2 (Disposition): -0.82

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.547
de0.526
newfound0.523
invigor0.495
rejuvenated0.495
Suppressed:
😞-0.485
😣-0.451
不-0.434
worse-0.419
😥-0.385