What does it mean to "permute" a predictor in the context of random forest?

Question

I am reading the vignette for the R package randomForestExplainer.

There the result accuracy_decrease (classification) is defined as

mean decrease of prediction accuracy after X_j is permuted

What does it mean "permuted" in this context? How is a predictor variable "permuted"?

Answer 1

Sample without replacement, i.e. we get a scrambled/"permuted" version of $x_j$.