This work attracted Spelke when she was still an undergraduate at Radcliffe College. From 1967 to 1971, she studied with Harvard child developmental psychologist Jerome Kagan and quickly found herself hooked on the excitement of investigating the essential workings of human cognition by analyzing children. She continued that research while pursuing her Ph.D. in psychology at Cornell University, where famed developmental psychologist Eleanor J. Gibson served as her graduate adviser and mentor. Gibson, one of only a handful of psychologists to win the National Medal of Science, had revealed much about infant cognition with some elegant experiments of her own. Her best known was the "visual cliff," a piece of heavy glass extending from a tabletop. Would early crawlers avoid the apparent drop-off? Most do, a discovery that revised theories of infants' spatial understanding.
Under such tutelage, Spelke hit on her own landmark experiment. "At dinner one night," she recalls as we talk in her office at Harvard, "I was musing with a fellow student over whether, when babies look at and listen to something, do they perceive [the sight and sound of an event] as two separate things, or do they recognize a link between the two? How would you find that out? Suddenly, I had this image of two visual events going on side by side, like movies, and between them a loudspeaker that you could switch from the sound of one event to the sound of the other event. Would a baby turn to look at the event matching the soundtrack the speaker was playing? That experiment became my Ph.D. thesis. It was the first time I was able to start with a general question about how we organize a unitary world from multiple modalities and turn the question into a ridiculously simple preferential-looking experiment--which actually ended up working."
Sure enough, Spelke found that babies recognized the link between sound and sight, switching their gaze back and forth as the soundtrack changed. Thus began Spelke's career of pondering big questions with straightforward experiments on tiny people. The mixed-modality approach addressed the same "binding problem" faced by blind people who suddenly can see: How does the brain mesh the signals from different senses into a single impression? Spelke did not answer how, but she did show persuasively that this ability seems innate.
Over the years Spelke has conjured up many other elegant and productive investigations on object and facial recognition, motion, spatial navigation, and numerosity (grasping of numerical relationships). She is able to envision simple but powerful tests, she says, "because I think like a three-year-old." By showing babies objects in motion and then interrupting their logical speed or course, she has found that even a four-month-old infers that a moving object is supposed to keep moving. Yet it takes an eight-month-old to grasp the principle of inertia and expect the object's path to be consistent and smooth. By showing babies different arrays of disks, she has found that six-month-olds can distinguish eight from 16 and 16 from 32--but not eight from 12 or 16 from 24. By having babies watch a person reach for one of two objects on a table, she has found that although 12-month-olds know from an adult's gaze which object he will grab, eight-month-olds do not.
As the data from such clever designs mounted, Spelke began to develop her theory of core knowledge, often inspired by or collaborating with colleagues such as noted Massachusetts Institute of Technology linguist Noam Chomsky, French mathematician turned cognitive neuropsychologist Stanislaus Dehaene and Harvard psychologist Susan Carey. Core knowledge systems, Spelke says, are neuronal "modules" that are in place at birth for building mental representations of objects, persons, spatial relationships and numerosity. Somewhat akin to the "deep grammar" that Chomsky believes underlies all human language, these core knowledge modules enable all infants to organize their perceptions.