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 late afternoon sun cast long shadows across the dusty pawn shop, painting the chipped Formica countertops in stripes. Amelia hummed under her breath, a faint melody that accompanied the gentle sway of her shoulders as she browsed. She had just finished a particularly satisfying gardening session, the scent of damp earth still clinging to her fingertips. Finding a vintage cookbook at a steal was the perfect finishing touch to her day.

Her eyes drifted over the glass display cases, filled with forgotten treasures. A glint of something familiar caught her attention. It was a ring, nestled amongst a collection of tarnished watches and broken chains. The setting was unmistakable: a delicate filigree of silver, holding a single, shimmering diamond.

"Oh my," she breathed, her voice a soft whisper. She pointed at the ring, her heart beginning to drum a quick little rhythm. The shop owner, a man with a tired smile, brought it out, placing it carefully in her outstretched hand. A warmth bloomed in her chest as she examined it. This was the ring her grandmother had worn, the ring she'd believed was lost forever.

Emotion: content

Cluster: Calm / Serenity
PC1 (Valence): 2.98 Positive
PC2 (Disposition): -2.19

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.446
affectionately0.435
😊0.392
😊0.386
thoughtful0.386
Suppressed:
S-0.682
L-0.537
C-0.485
😣-0.362
est-0.350