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 house felt too quiet, the silence amplified by the discovery in the locked drawer. Maya ran her fingers over the crisp paper, the lines of her grandmother's delicate script a revelation. The letter, a resignation, a declaration of war against convention and expectation, had always been hidden.

She read the words, written with a fierce determination, detailing her grandmother's decision to leave her marriage and her job. This powerful woman, the family matriarch, had been hiding an entire other life, a secret self.

A tightness in her chest constricted her breathing. Maya sat motionless, the letter clutched tight. Her foundation had been shaken, and she felt as if a vital truth had been stolen from her. The world suddenly felt unsteady, and the echoes of her grandmother's bold decision resonated in her own life.

Emotion: heartbroken

Cluster: Sadness / Despair
PC1 (Valence): -2.60 Negative
PC2 (Disposition): -1.10

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.433
S0.385
😔0.359
haunting0.349
😢0.342
Suppressed:
de-0.503
la-0.455
B-0.352
🤓-0.332
!-0.299