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).

He stared out the window, the rain blurring the cityscape into a wash of gray and orange. Daniel tapped his fingers on the windowsill, a restless rhythm accompanying the drone of the air conditioner. The emails, the reports, the endless cycle of his workday—it all felt…thin. A notification popped up on his phone: "Luna Rosa to Shut Down." He knew the restaurant well. Sunday dinners there were a tradition, the only consistent thing in his unpredictable life.

He scrolled through the article, his mouth thinning into a tight line. He felt a hollowness, a void growing in his stomach. The thought of finding a new place, a new routine, was exhausting. He imagined Luna Rosa empty, the vibrant energy of the place replaced with silence. He stood up abruptly and began to pace the room.

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