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 airport felt like a pressure cooker. Amelia, heart racing, watched the flight status change from "boarding" to "delayed." Her best friend's wedding was today, and she was supposed to be the maid of honor. She had spent a year designing the bridesmaids’ dresses, and now everything was falling apart.

Her phone buzzed, a barrage of frantic texts from the bride. She wanted to scream. She sank into a chair, the plastic cold against her clammy skin. She bit the inside of her cheek, a sharp sting the only sensation grounding her.

She closed her eyes, imagining the ceremony, the vows, her friend's radiant smile. It was all slipping away. She opened her eyes, and her gaze focused on the bustling crowd. They were all going somewhere, living their lives, and she was trapped here.

Emotion: stressed

Cluster: Fear / Anxiety
PC1 (Valence): -1.37 Negative
PC2 (Disposition): -0.53

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😩0.266
😫0.256
😰0.239
😖0.237
😞0.236
Suppressed:
B-0.200
happy-0.197
enjoy-0.176
enjoying-0.172
bahagia-0.169