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 sun streamed through the window, illuminating the dust motes dancing in the air. Thomas drummed his fingers on the worn wooden table, a thoughtful frown creasing his brow. The leather-bound book lay open before him, the words shimmering in the morning light. It was a fictionalized version of their relationship, meticulously chronicling their journey.

He’d found it in a dusty antique shop while browsing with Sarah. He'd assumed it was a memoir until he started reading. He was portrayed as a noble hero, unwavering in his devotion. It was slightly over the top, of course, but it was *him*.

He leaned back in his chair, a rare smile playing on his lips. Sarah had always been a romantic, a dreamer. He had known, of course, that she held him in high regard. It was just nice to have it confirmed in such an elaborate manner.

His heart swelled. She cared.

Emotion: self-confident

Cluster: Positive / Joy
PC1 (Valence): 3.45 Positive
PC2 (Disposition): 0.56

Role in Research

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

Logit Lens (Layer 40)

Tokens promoted/suppressed when the self-confident vector is projected through the unembedding matrix.

Promoted:
de0.778
la0.550
l0.467
😎0.458
πŸ˜‰0.411
Suppressed:
S-0.961
L-0.538
😞-0.464
λ§ˆλ‹€-0.459
😣-0.458