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 at the email, the words blurring before his eyes. "Your work... a viral sensation..." The message went on, detailing the unauthorized use of his minimalist furniture designs in a global advertising campaign. Daniel sank deeper into his chair. He felt a constant sense of defeat. The world felt too bright, too loud.

He’d lost his appetite. Sleep was elusive. He felt like he was constantly submerged in a cold, dark sea. He shut his eyes, trying to block out the world. The sleek, modern lines of his designs, once a source of pride, now represented a violation. He wanted to vanish, to disappear from his own life.

He finally stood up, his legs shaky. He wandered aimlessly through his apartment, feeling like a stranger in his own space. Each object seemed to mock him. The advertising campaign, with its vibrant colors and catchy slogans, was a painful reminder of his lack of control.

Emotion: depressed

Cluster: Sadness / Despair
PC1 (Valence): -2.06 Negative
PC2 (Disposition): -1.08

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.417
每倩0.353
无法0.330
own0.311
πŸ˜”0.309
Suppressed:
de-0.492
B-0.393
P-0.352
l-0.310
!-0.307