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 sound of clinking dice and hushed conversations filled the basement. Robert had always cherished his weekly Dungeons & Dragons sessions, a haven from the pressures of work. Now, it was different.

He went downstairs to find his son, Timmy, running a game, surrounded by his friends, their faces illuminated by the flickering candlelight of the game. Timmy was animated, immersed in the game and in charge. Robert felt a hollow sensation.

"How long have you been doing this?" he asked, feeling a need for a question even though he already knew the answer.

"A few weeks, Dad." Timmy grinned. "It's awesome!"

Robert retreated upstairs, feeling a peculiar lack of control.

Emotion: puzzled

Cluster: Surprise / Confusion
PC1 (Valence): -1.17 Negative
PC2 (Disposition): -0.25

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
que0.453
🤔0.421
🤔0.402
🧐0.328
a0.294
Suppressed:
ecstatic-0.265
gladly-0.238
swiftly-0.237
alleviate-0.234
wholeheartedly-0.232