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 air in the terminal felt thick and stagnant, the stale air doing nothing to help the situation. Robert sighed, a sound filled with the accumulated weariness of a long and fruitless wait. He had already tried everything to keep his mind occupied. He’d read a book, done a crossword puzzle, and even attempted to write a short story. None of it had worked. The anticipation for the concert, the one he had been waiting months for, was now turning to ashes.

He tapped his foot impatiently, a nervous rhythm that echoed his internal frustration. He glanced at the clock again, knowing the time hadn't changed, and felt a fresh wave of disappointment wash over him.

He ran a hand through his hair, a gesture of resignation. He closed his eyes, conjuring up the energy to take a walk, and get some fresh air.

Emotion: bored

Cluster: Fatigue / Lethargy
PC1 (Valence): 0.04 Positive
PC2 (Disposition): -1.11

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.387
’0.350
unremarkable0.325
दिलचस्प0.323
seemingly0.319
Suppressed:
B-0.592
P-0.304
because-0.292
因为它-0.287
ancar-0.278