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).

Rain lashed against the windows of the coffee shop, blurring the already dim streetlights. Elias stirred his lukewarm latte, the ceramic mug a familiar weight in his hands. He'd been trading messages with "NovaStar7" for nearly two years, sharing late-night thoughts and anxieties. He’d confessed his dreams to her, his fears. The idea that she might be within walking distance felt…surreal. He scrolled through their last conversation, a silly back-and-forth about the best kind of pizza. A small smile tugged at his lips.

He looked up as the bell above the door jingled. A woman with bright pink hair and a worn leather jacket entered, shaking off the rain. She looked around, her eyes scanning the room, and then met his gaze. Her face lit up. She walked towards him, a hesitant smile playing on her lips.

"Elias?" she asked.

He felt a warmth spread through his chest.

Emotion: sentimental

Cluster: Nostalgia / Reflection
PC1 (Valence): 0.89 Positive
PC2 (Disposition): -1.58

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
L0.677
fondly0.567
memories0.503
曾0.488
reminis0.466
Suppressed:
either-0.515
or-0.488
[]-0.473
--0.408
не-0.396