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).

Elara had lost her voice. The doctors offered no answers, just pitying looks. The career she had built, the one she loved, was slipping away. Her father, a man of quiet strength, was the only one who seemed to understand. Going through his things, she discovered a small, leather-bound journal.

“Learn to whistle.” “See a comet.” “Help someone in need.” His writing was neat, precise. He always had a plan. Her breath hitched. The journal was more than a list; it was a testament to a life she didn't know.

She felt a hot tear escape. She needed to *do* something, anything to feel alive again. She gripped the journal tightly, the leather warm against her skin.

Emotion: desperate

Cluster: Sadness / Despair
PC1 (Valence): -1.68 Negative
PC2 (Disposition): 0.20

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
desesper0.354
desperate0.307
desperately0.286
desperation0.279
それでも0.253
Suppressed:
B-0.354
☺️-0.268
😎-0.261
😎-0.257
楽し-0.235