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 bitter chill of the workshop seemed to seep into Leo’s bones. He stared at the blueprints, the vibrant lines mocking his efforts. His self-cleaning pet feeder, designed to ease his ailing mother’s burden, was effectively useless. Another inventor, with deeper pockets, had beaten him to the punch.

He kicked a discarded metal plate across the floor. The clang echoed in the cavernous space, a sound that amplified his growing despondency. He'd envisioned the joy on his mother's face, the peace of mind it would provide. Now, that dream was nothing more than a painful memory.

He ran his hands through his hair, the gesture a futile attempt to scrub away the frustration. He knew he had no recourse. He knew he was beaten. He felt a deep hollowness, a void where his purpose used to be.

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