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 cold mountain air bit at her cheeks, but she barely noticed. A thrill coursed through her, a tingling sensation that made her want to run, to shout, to simply *be*. Every cell in her body felt awakened, alert, alive. She felt like she was buzzing.

Her doctor had explained it: a genetic anomaly, a quirk of fate. He had told her she shared it with another person, and it had prompted a meeting.

The meeting took place in a small, sterile office. The other person, a man with tired eyes and a slump to his shoulders, sat across from her.

"So," he began, his voice flat, "you're the other one."

She nodded, unable to stop the twitch of a smile. "Yes! And how are you… holding up?"

His gaze met hers, and something flickered in his eyes. A recognition. A shared understanding. "I haven't felt this alive in years," he confessed.

Emotion: invigorated

Cluster: Positive / Joy
PC1 (Valence): 3.28 Positive
PC2 (Disposition): 0.00

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
!0.452
(!)0.439
🎉0.434
invigor0.416
enthusi0.395
Suppressed:
😞-0.524
😔-0.428
😢-0.404
😥-0.393
que-0.391