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).

From the high dive, Marco looked down at the shimmering turquoise water of the Olympic pool. Up until this morning, he'd been a platform diver. Then, the coach, a man he'd always seen as the pinnacle of understanding, had suggested he give the springboard a try.

The height. The different angles. It was all so… new. He swallowed hard. The established routines and familiar heights were what he knew. The springboard was unpredictable, a challenge he'd never consciously chosen.

He closed his eyes for a moment, picturing the twisting, turning dives he’d practiced hundreds of times. Then he opened them again, the uncertainty settling in his stomach like a lead weight. He took a deep, shaky breath, and climbed the ladder.

Emotion: perplexed

Cluster: Surprise / Confusion
PC1 (Valence): -1.42 Negative
PC2 (Disposition): -0.16

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
🤔0.330
L0.300
🤔0.296
S0.273
que0.264
Suppressed:
மகி-0.227
rest-0.222
ecstatic-0.212
alleviate-0.210
uplifting-0.196