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 attic air hung thick and still, dust motes dancing in the single shaft of sunlight slicing through a crack in the boarded-up window. Sarah knelt, the old wooden floorboards creaking softly under her weight as she sifted through a trunk overflowing with forgotten memories. She wasn't rushed. She’d promised herself a leisurely day of sorting, and the slow unfolding of the past felt like a balm to her soul. She carefully pulled out a leather-bound journal, its pages brittle with age, and traced her fingers over the embossed initials on the cover. "A.C." Her heart skipped a beat. This belonged to Adam. It wasn't the kind of thing he ever mentioned, not someone who seemed to keep a diary.

She opened the journal cautiously. The first entry was a list, a bucket list scrawled in Adam’s familiar handwriting. "Learn to play the ukulele. See the Northern Lights. Run a marathon." Her gaze drifted over the neatly penned items. She felt a warmth bloom in her chest, a quiet joy. The thought of all the things he wanted to experience, all the secret hopes he harbored, filled her with a sense of connection she hadn't anticipated. She read on, the afternoon sun casting long shadows across the attic floor, each entry a small, intimate revelation.

Emotion: patient

Cluster: Passivity
PC1 (Valence): 2.71 Positive
PC2 (Disposition): 0.40

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
rest0.467
eventually0.415
leisurely0.369
慢慢0.357
a0.355
Suppressed:
无法-0.432
even-0.428
인해-0.421
無法-0.415
😣-0.391