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 was deafening. John’s jaw clenched, his teeth grinding. The abrupt absence of the tree was a physical wound. He gripped his hands into fists, feeling the nails biting into his palms. He ran a hand across his forehead, and his palms were damp. His legs felt shaky, and the world seemed to tilt slightly.

He imagined the satisfying strength of the wood, the way he used to climb it as a boy. A sudden, unfamiliar heat flared within him, a burning ache in his groin. He wanted to feel something, anything, other than this raw emptiness. He began to pace, his steps quick and agitated, a caged animal desperate for release. His gaze was fixed on nothing in particular, lost in a haze of sudden, urgent need.

Emotion: aroused

Cluster: Alertness
PC1 (Valence): -0.45 Negative
PC2 (Disposition): -1.13

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.386
S0.383
P0.379
own0.376
悸0.344
Suppressed:
a-0.877
l-0.501
de-0.474
que-0.438
been-0.384