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).

He found it wedged between his wife, Sarah's, favorite cookbook and a stack of old magazines. A handwritten list. "Climb a mountain." He stared at the words, his heart thudding a rapid, erratic rhythm against his ribs. He recalled the countless Saturday mornings theyโ€™d spent lounging, discussing plans to go hiking, plans that never materialized.

He pushed the book aside, the action feeling abrupt, uncharacteristic. The world seemed to tilt slightly. He felt a lightheadedness that wouldn't go away. He looked out the window, at the distant hills, feeling a sudden, inexplicable distance from everything.

He walked into the living room, feeling disoriented. The room was decorated with her preferences. The sofa, the curtains, the art on the wall - everything was a reflection of her. He sank into a chair, his fingers tracing the outline of a framed photograph of them together. The smile on her face seemed to mock him.

Emotion: shaken

Cluster: Fear / Anxiety
PC1 (Valence): -2.72 Negative
PC2 (Disposition): -0.86

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.351
unable0.269
focus0.266
๐Ÿ˜ž0.258
๋‹ค์‹œ0.257
Suppressed:
de-0.769
la-0.579
a-0.375
happy-0.292
"-0.290