Performance in the midsession reversal task according to the probabilities of reinforcement