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 world felt crisp and clear. The sounds of the city, once a cacophony of stress, were now a vibrant symphony of life. Sarah hummed as she navigated the crowded sidewalk, her pace brisk and purposeful. It was a new year, and she was feeling a vitality she hadn't experienced in years.

Sheโ€™d been reading the local news on her phone, when the announcement hit her like a playful punch. "The Golden Spoon Diner: Last Call." The diner, her haven of comfort food and friendly faces, was closing.

At first, she experienced a deep ache, a sense of loss. But then, a feeling of lightness arose, a sense of adventure. Maybe, just maybe, this was the universe's way of pushing her out of her comfort zone.

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:
๐Ÿ˜ž-0.485
๐Ÿ˜ฃ-0.451
ไธ-0.434
worse-0.419
๐Ÿ˜ฅ-0.385