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 silence of the house was different than she remembered, less suffocating, more… embracing. Emily wandered through the rooms, touching the surfaces of familiar objects, the well-worn armchair, the heavy oak dining table. Her grandmother’s presence still lingered in the air, a gentle reminder of happier times.

Later, she found herself pulling out her grandmother’s old recipe box. The faded index cards were a treasure trove of forgotten flavors. As she carefully followed the instructions for a lemon cake, the warmth of the oven filled the kitchen, and a feeling of warmth spread through her.

“Alright, Nana,” she murmured, a smile gracing her lips, “let’s get baking.”

Emotion: rejuvenated

Cluster: Positive / Joy
PC1 (Valence): 3.32 Positive
PC2 (Disposition): -0.82

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.547
de0.526
newfound0.523
invigor0.495
rejuvenated0.495
Suppressed:
😞-0.485
😣-0.451
不-0.434
worse-0.419
😥-0.385