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 afternoon sun cast long shadows across the lawn. Emily walked aimlessly, her hands shoved deep into her pockets. She felt like a figure adrift, floating through a landscape she didn’t belong in. The yard sale was just another distraction.

She wandered past a table piled high with childhood cast-offs. A chipped music box, a tarnished silver locket… and then, the bear. Its fur was faded, and its once bright smile was a mere suggestion. It was a symbol of a past that now felt impossibly distant.

“Five dollars,” the seller said, her eyes devoid of expression. Emily paid without a word, feeling a pang of… something. She looked at the bear. Not hope. Not sadness. Just… acceptance.

Emotion: resigned

Cluster: Passivity
PC1 (Valence): -0.75 Negative
PC2 (Disposition): 0.16

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.437
a0.351
que0.323
வதில்லை0.267
sighed0.265
Suppressed:
C-0.375
!-0.316
인해-0.310
Furthermore-0.298
🤩-0.296