FORTUNE -- Early on, in the 1950s, Robert Sternberg flubbed IQ tests, and his elementary school branded him a loser. "As a result of my low scores, my teachers thought I was stupid, and I did too," he writes in his passionate new book, College Admissions for the 21st Century. "They never came out and told us our IQ scores, but one could tell from the way the teachers acted I was a mediocre student, which made my teachers happy because they got what they expected." In a "self-fulfilling prophecy," Sternberg performed a little bit worse each year. But he lucked out in fourth grade when a teacher "had high expectations for me." He got A's and altered his "entire future trajectory."
Sternberg went on to become a leading cognitive psychologist at Yale and Tufts. He has long challenged traditional measures of human intelligence based on such skills as memory and vocabulary. In particular, he's argued that college admissions rely too heavily on standardized tests like the SAT and ACT, which he says overvalue analytical abilities at the expense of leadership qualities like creativity and wisdom. As a dean at Tufts, Sternberg instituted Project Kaleidoscope, which aimed to test high school applicants on leadership attributes: They could invent a product, design a costume, or answer a question like "Was Kermit the Frog right that it's not easy being green?" While some admissions officers encourage such imaginative submissions, Sternberg maintains that answers can be scientifically evaluated to predict success. It takes time -- and Sternberg had to add staff at Tufts -- but he says it produces a better class, chosen more fairly.
Now the provost of Oklahoma State University, Sternberg wants his methods to become mainstream. His book convincingly indicts the SAT and ACT exams. A single test lasting a few hours, he writes, "ends up having a weight equal" to the product of "years of effort and dedication" in high school. We know the justification for the tests: Admissions officers believe that only an exam administered to thousands of applicants offers a legitimate basis for comparison; there's no way to compare GPAs from Podunk High and Exeter. Extracurricular activities are even fuzzier reference points. So it's easy to conclude quantitatively that Susie's 1600 makes her a better candidate than Billy's 1500. And Andy's 1200 is so low we can immediately put him in the reject pile.
You have to be skeptical that essays about Kermit can be judged in any way that's not deeply subjective, though Sternberg's attempt to assess data scientifically is laudable. Still, he wants it both ways: He wants creativity yardsticks to augment tests. So why not be bolder and unilaterally disarm, at least for a trial period? Let Princeton or Stanford declare that it will no longer accept SAT or ACT scores, instead adding Sternberg's measures. In four years we can see whether the class of 2015 turns out to be less successful than the class of 2014 or 2016. If it doesn't, that provides a compelling case for scuttling the tests. We won't know if someone doesn't try.