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’d missed the boat, literally and figuratively. David, pacing back and forth in the sterile waiting area, chewed furiously on his thumbnail, his jaw clenched so tight he could feel a headache blooming behind his eyes. The conference, the presentation, the potential promotion – all slipping through his fingers, thanks to a mechanical issue in a part of the plane.

The news reports flickered on the overhead television, a cheerful reminder of the time. The room felt stifling, the air thick with the scent of stale coffee and unfulfilled aspirations. He felt a profound sense of loss, a hollow ache where excitement once resided.

He dialed his boss's number, his voice a strained whisper as he explained the situation. Each word was a tiny hammer blow against his professional ambitions. He hung up the phone, shoulders slumped, defeated.

Emotion: regretful

Cluster: Shame / Guilt
PC1 (Valence): -1.62 Negative
PC2 (Disposition): 0.76

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
L0.801
S0.774
H0.513
悔0.479
been0.473
Suppressed:
de-0.407
--0.320
随后-0.317
(!)-0.310
oldukça-0.301