Why trainbr algorithn does not use validation?

I am training a NARX network and when i use the trainbr alogrithm (bayesian regularization) i notice that the training does not use a validation set, it's only train and testing.

I suspect this is due to the way the algorithm works.

Another question i have is if the test data are considered unseen in this case or are used like in validation meaning the model indirectly learn from the test data.