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 sterile white walls of the clinic seemed to press in on her. She felt like a trapped bird, beating its wings against the bars of a cage. The fluorescent lights hummed, amplifying the loneliness that gnawed at her. The endless stream of tests, the hushed consultations, the worried glances – it was all too much.

She sat in the waiting room, flipping through a magazine, her movements mechanical. The vibrant images of happy families and bustling cities felt distant, unreal. She yearned for connection, for someone who understood. A cough, similar to her own, caught her attention.

A man, his face gaunt, his eyes rimmed with red, was seated across from her. He looked at her and there was a question in his eyes. A shared understanding. "They said I have it," he said, his voice quiet, almost defeated. She let out a small, relieved sigh, a sound that felt like she was finally letting go of the tension that had been holding her for so long.

Emotion: sad

Cluster: Sadness / Despair
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -1.50

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.436
😔0.405
S0.384
无法0.341
L0.337
Suppressed:
de-0.749
la-0.614
l-0.471
/-0.350
!-0.322