This is actually a really complicated question and a lot of research is being done to understand it. Stuttering in young children naturally has a high recovery rate-- as many as 80-90% of kids who stutter will outgrow it with no therapy whatsoever. So, we try to only put the kids in therapy who we think are likely to NOT grow out of it. Many (not all) of those kids do seem to "recover" from stuttering as a result of the therapy. Was the recovery *because* of the therapy? Would they have phased out of it anyway, without therapy, despite the risk factors for continuing? If in fact the recovery was due to therapy, why is this? What happened on a neural level? It probably has something to do with neuroplasticity, but you also have to account for the kids who continued to stutter even though they received therapy. You're right that for early childhood, we do have a hope/goal that the outcome of therapy will be that the child doesn't have to deal with a stutter for the rest of their lives, which is very different from adolescent/adult therapy where the goal is more lifelong management. But, the processes and understanding of why this is "sort of" possible in young kids still carries a lot of mystery.