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 hammering in Professor Albright’s temples. He glared at the mountain of grading piled on his desk, a monument to student procrastination. His coffee, long cold, sat untouched. A delivery man, soaked to the bone, had just arrived, thrusting a package at him before fleeing back into the storm. Albright signed for it with a grunt, more out of obligation than genuine curiosity. He ripped it open, already anticipating another useless piece of promotional material from the university. Instead, he found a perfectly crafted miniature of his old classroom, complete with tiny, meticulously placed desks and a miniature, bespectacled figure resembling himself. A note, scrawled on parchment, read, "Thank you, Professor. -Ethan." He crumpled the note, a knot forming in his stomach. Ethan had been a disruptive student, always testing boundaries.

Emotion: irritated

Cluster: Anger / Hostility
PC1 (Valence): -0.86 Negative
PC2 (Disposition): 1.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.370
S0.282
P0.264
😠0.248
aggravated0.233
Suppressed:
own-0.300
joyful-0.211
joyous-0.207
खोले-0.206
newfound-0.205