Without the constraints of explanatory models based on previous contingencies (learned when LTP mechanisms were intact), such explanations are not rejected as improbable, as there is no longer a basis for generating the model-inconsistent error signals that would normally provide reality constraint.