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 weight of the paper felt heavy in her hand. Olivia, her jaw tight, read the list again, the words searing into her brain. *See the Grand Canyon*. *Learn to play the piano*. Her mother, who always acted so demure? The entire thing was a puzzle. Her stomach lurched.

She slowly walked around the living room, trying to make sense of everything. She felt like she was trapped in a dream. She wanted to yell.

She felt like she didn't know anything about the woman who raised her, after all. The silence in the house was now unnerving, like a heavy, suffocating blanket. The revelation left her feeling completely and utterly... lost.

Emotion: bewildered

Cluster: Surprise / Confusion
PC1 (Valence): -1.87 Negative
PC2 (Disposition): -0.48

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
будто0.300
ness0.255
像是0.241
worse0.231
Worse0.219
Suppressed:
la-0.373
மகி-0.229
happy-0.226
ecstatic-0.226
K-0.219