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, mirroring the tempest inside Amelia. She hunched over her laptop, the glow reflecting in her tired eyes. Another all-nighter, fueled by lukewarm coffee and despair. The deadline loomed, a monstrous shadow over her already fragile state. She slammed her fist on the desk. This presentation, this *project*, was everything. She just couldn't bear to fail.

A shrill ping from her email shattered the silence. A new message. Subject: "Important Project Update." Her heart hammered. With clumsy fingers, she clicked. It was from the project lead. The attached file… it was her presentation. Except, it wasn't. It was *his*. Identical. Word for word, slide for slide.

Amelia’s breath hitched. She felt a cold dread creep through her. Hours of frantic work, down the drain. All her hopes, dashed. All that remained was a profound, aching emptiness. She wanted to scream.

Emotion: miserable

Cluster: Sadness / Despair
PC1 (Valence): -2.19 Negative
PC2 (Disposition): -0.87

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.369
无法0.321
無法0.270
💔0.265
😔0.262
Suppressed:
de-0.490
la-0.329
B-0.300
l-0.292
/-0.224