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 hospital corridor stretched ahead, cold and impersonal. Dr. Chenโ€™s absence felt like a gaping wound. He'd walked in, ready to start rounds, only to find a replacement doctor in his place. A curt note taped to the door of his office, explaining his sudden retirement. I clenched my fists at my sides.

My vision blurred, the sterile white walls seemed to close in. I remembered his calm voice, his steady presence during the most difficult surgeries. The sudden loss was a shock. My legs felt heavy, as if I were wading through thick mud. The silence of the now-empty office was a heavy weight. I needed to know why.

Emotion: tense

Cluster: Fear / Anxiety
PC1 (Valence): -1.32 Negative
PC2 (Disposition): -0.53

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.213
๐Ÿ˜ฐ0.191
๋А๊ปด0.183
๐Ÿ˜ฃ0.183
๐Ÿ˜–0.181
Suppressed:
a-0.260
de-0.258
la-0.256
happy-0.218
B-0.198