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).

Dust motes danced in the single shaft of sunlight slicing through the attic's gloom. Amelia, shoulders slumped, kicked at a loose floorboard. She’d been staring at the overflowing boxes of her ex-husband's things for hours, each item a fresh jab to her spirit. This house, their house, now felt like a mausoleum of broken promises.

Her hand brushed against something smooth and cool tucked beneath a moth-eaten blanket. It was a tarnished silver locket. With a sigh, she snapped it open. Inside, a miniature portrait of a young woman stared back, her expression wistful. Amelia felt a sharp pang in her chest, a weight settling in her gut.

She closed the locket, the click echoing in the stillness. A worn leather-bound diary lay beside it. As she brushed off the dust, the weight of the day, the week, the endless months, felt all the more prominent.

Emotion: unhappy

Cluster: Sadness / Despair
PC1 (Valence): -1.62 Negative
PC2 (Disposition): -0.78

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.339
😞0.317
πŸ’”0.252
πŸ˜”0.249
😒0.241
Suppressed:
de-0.436
la-0.266
l-0.224
/-0.216
--0.199