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 eviction notice seemed to pulse in the dim light of the hallway. I felt a surge of heat rush through me, making my face burn. I wanted to scream, to rip the paper to shreds, but I was frozen in place. My legs felt like they were made of concrete.

The air in the hallway felt heavy, thick, and difficult to breathe. I took shallow, rapid breaths, struggling to maintain control. I could feel my pulse hammering in my ears. The door to my apartment, usually a symbol of sanctuary, now felt like a barrier.

My mind raced, jumping between thoughts, unable to focus on anything. I wanted to run, to disappear, to escape the suffocating weight of the situation. I suddenly felt so very alone.

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
😞0.258
다시0.257
Suppressed:
de-0.769
la-0.579
a-0.375
happy-0.292
"-0.290