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).

Michael slammed his laptop shut, the metallic clang echoing in the otherwise silent apartment. The kitchen counter, usually a comforting space, was now a battleground littered with empty coffee cups and discarded energy bar wrappers. He felt like his chest was being compressed, making each breath a conscious effort. He needed to get this done, but every attempt to begin just resulted in a blank screen.

He’d poured weeks of his life into this prototype. He'd lost track of nights he'd stayed up. Now, a simple email from his supervisor, subject: "Synergy," had thrown him off. It had mentioned a meeting with a colleague, a “shared project initiative.” He dreaded this meeting.

The moment he walked into the conference room, he knew. There, on the projector screen, was his design. He stared at it, frozen. The room spun.

Emotion: overwhelmed

Cluster: Fear / Anxiety
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -0.94

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.385
😫0.332
😖0.319
😣0.319
😩0.314
Suppressed:
la-0.576
de-0.496
a-0.297
happy-0.290
l-0.264