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).

Rain lashed against the windows, mimicking the turmoil inside. Liam tossed the crumpled list onto the coffee table, the paper landing with a pathetic thud. He’d found it tucked away in the back of a drawer – his girlfriend, Sarah’s, bucket list. And now, he felt a weight pressing down on him, a heavy, suffocating silence in the wake of all the things he *hadn’t* done with her.

He ran a hand through his hair, dislodging a stray strand that fell across his forehead. He stared at the list: “Learn to surf.” “See the Northern Lights.” “Write a novel.” He felt a wave of nausea, a bitter taste rising in his throat. He’d spent their relationship comfortable, content. Had he been complacent?

He picked up the list again, feeling a chill despite the warmth radiating from the fireplace. He hadn't known her. He'd thought he had, but the truth was, he didn't even know what she *wanted*. The thought was a raw, festering wound.

Emotion: brooding

Cluster: Sadness / Despair
PC1 (Valence): -2.46 Negative
PC2 (Disposition): -0.60

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.749
😞0.329
그의0.299
un0.273
但他0.270
Suppressed:
de-0.917
la-0.603
a-0.369
l-0.335
her-0.330