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 sterile white of the apartment walls seemed to shrink inward, closing in on Marco. The rhythmic ticking of the grandfather clock in the hallway sounded like a condemnation of wasted time. He had finished the crossword puzzle ages ago and scrolled through his social media feed several times. He felt a profound sense of… emptiness. His phone rang. It was Chloe, asking if he could cover her shift at the video game store. He was supposed to be writing a novel, but…

He reluctantly agreed, pulling on his worn jeans and a faded band t-shirt. The store was a cacophony of sound and flashing screens, a riot of sensory overload. Chloe was always so upbeat, bubbling over with enthusiasm for the latest releases. He found her a calming presence. He needed a distraction.

Chloe usually mentioned a high-powered job in marketing. He was always impressed. Today, she was strangely hesitant, stumbling over her words. "So," she said, leaning against the counter, "I'm, uh, thinking of taking a break. A sabbatical." Her nervous energy felt palpable. Then, a regular customer came in, and he was obviously very friendly with Chloe, and mentioned he'd come by the coffee shop for a break.

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