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 world seemed to tilt on its axis. The sight of the sheared birch trunk, so brutally efficient, was a punch to the gut. It felt like a part of her was missing, ripped away without warning. A dull ache settled in her shoulders, forcing them inward, as if trying to protect herself.

She took a shaky step back, as if the ground itself was unstable. The silence was deafening, amplified by the absence of the rustling leaves and the gentle sway of the branches. The weight of the moment felt immense, like an invisible cage surrounding her.

"How could they?" she whispered, her voice barely audible. Her chest ached, a deep, persistent throb. The world suddenly felt close, as though the boundaries had shrunk. The sky seemed a little lower.

Emotion: trapped

Cluster: Sadness / Despair
PC1 (Valence): -3.02 Negative
PC2 (Disposition): -0.80

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.573
ness0.401
无法0.366
hopeless0.360
😞0.353
Suppressed:
la-0.740
de-0.376
l-0.339
several-0.323
excitedly-0.322