Joseph Kable's research challenges assumptions about the "marshmallow" test.