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 phone vibrated again, another notification. Amelia ignored it, focusing instead on the canvas before her, the still-life she had been working on for weeks. It featured a single, wilting sunflower. The vibration persisted, a relentless, irritating buzz. Finally, she grabbed the phone and saw the photo. It was a close-up of a forearm, inked with her exact rendition of the wilting sunflower.

Her stomach lurched. The canvas, her sanctuary, suddenly felt like a prison. She dropped her brush, the bristles splattering paint across the floor. A wave of nausea rolled over her, and she felt the blood drain from her face.

She slumped onto a stool, her shoulders slumping forward. The vibrant colors of her studio seemed to mock her, the beauty she had tried to capture now tainted by this… this blatant disregard.

Emotion: exasperated

Cluster: Anger / Hostility
PC1 (Valence): -0.99 Negative
PC2 (Disposition): 1.23

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.435
l0.378
P0.331
que0.320
la0.269
Suppressed:
own-0.433
gradually-0.254
soon-0.248
unfolding-0.216
joyful-0.216