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 at the canvas awning, mirroring the tempest inside Leo. He shoved his hands deeper into his pockets, the damp seeping through his jeans. This ridiculous sale was a monument to boredom. He didn't *need* anything. He *wanted* to be anywhere else.

Then he saw it. A familiar, threadbare form slumped against a plastic bin filled with broken toys. Its one remaining button eye stared blankly. He froze, a cold jolt spreading through him.

“How much?” he asked, the words forced from his clenched jaw.

The seller, a jovial man in a stained apron, chuckled. "Five bucks. He's seen better days, that fella."

Five dollars. Five measly dollars for something that dredged up a whole childhood's worth of… *stuff*. He paid, barely glancing at the man. His mind raced. He had to get home. Now.

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