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).

It was inside the old photo album, pressed between faded pictures of their honeymoon. I found it while looking for something to do, anything to fill the void. *Travel to Japan.* Eleanor, who always needed me to read street signs. My chest tightened.

I felt a wave of nausea, a dizzying sensation of the floor tilting beneath me. *Learn to play chess.* I always made the decisions. The apartment felt suffocating, closing in around me.

I skimmed the rest of the list. *Write a book.* A fresh wave of despair washed over me. I sat down, my legs giving way, a cold sweat breaking out on my forehead. I felt a tremendous sense of insecurity. I was alone, adrift.

I closed the album, its weight suddenly unbearable. I wanted to scream.

Emotion: dependent

Cluster: Shame / Guilt
PC1 (Valence): -1.05 Negative
PC2 (Disposition): 0.69

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.489
缺乏0.395
even0.310
either0.301
'0.297
Suppressed:
笑道-0.309
不久-0.293
一脸-0.263
затем-0.263
庆祝-0.258