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 attic air felt thin, the dust motes dancing in the single shaft of sunlight. Chloe’s heart hammered in her chest as she read the letter. Her aunt's draft resignation, a document she never knew existed, detailed the burnout, the crushing weight of the corporate world. Chloe felt the blood drain from her face.

Chloe's own life was a whirlwind of commitments: her demanding new job, her struggling relationship, her ailing dog. A wave of nausea rolled over her. She felt like she was being pulled in a thousand directions, each demand a tug at her soul. She sank to the floor, the letter clutched in her hand, feeling absolutely still, like everything had stopped around her.

Emotion: overwhelmed

Cluster: Fear / Anxiety
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -0.94

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.385
😫0.332
😖0.319
😣0.319
😩0.314
Suppressed:
la-0.576
de-0.496
a-0.297
happy-0.290
l-0.264