When I See a Monkey ReadPrint
By Jessica Love
May 31, 2012
About 15 years ago, my mother took an admissions test for a graduate program. Applicants needed to answer at least 35 percent of the questions correctly, and each question had four possible responses. Once the test had been collected, my mother attempted to assuage the feelings of a woman fretting over her performance. “I’m sure you did fine,” my mother said. “A monkey would get 25 percent.” The woman looked puzzled. “Well, a monkey circling B after every question would get the answer right a quarter of the time,” my mother continued. There was a sizable pause before the woman responded. “But what if the right answer was always C?”
Perhaps, then, this woman’s fretting was not unreasonable. But nor was my mother giving monkeys enough credit. Forget circling B. These days, monkeys can do lexical decision.
In a lexical decision task, letter strings (e.g., wind, jinp) appear in isolation and the responder must decide, often quickly, whether they form real words or not. The task is one of the simplest and most ubiquitous in all of cognitive psychology. Want to know if the age at which a word was learned affects how easily it can be accessed from memory? Match two sets of words—one learned early in toddlerhood and a second learned much later—on as many factors as you can imagine (such as word length, frequency of occurrence, concreteness of the concept it describes) and then throw the words into a lexical decision task. If age of acquisition matters, you’d expect to see faster (and possibly more accurate) responses to the words learned earlier. Interested in whether reading nurse makes it easier to access the word doctor? Put doctor immediately after nurse in a lexical decision task and see if responses to doctor speed up. The list of hypotheses this task has been used to test goes on and on.
Now they’ve got monkeys doing it. In a study published earlier this year in Science, French psychologist Jonathan Grainger and his colleagues taught six baboons to identify English words. The monkeys saw stimuli like wind or jinp on a computer and classified each as a word or nonword using a touch screen. Correct classification was rewarded with dry wheat. Incorrect classification condemned them to a full three-second wait—interminable, with dry wheat on the line—before they were allowed to resume. Over the course of six weeks, the baboons learned to classify dozens of words—hundreds, in the case of a wunderkind named Dan—at above 75 percent accuracy. Even more impressive, the monkeys learned to discriminate novel letter strings as well.
Now, a lexical decision task is only as difficult as its nonwords are wordlike. That is, everyone finds it much easier to classify zxjl as a nonword than jinp. In this case, the researchers selected nonwords that, though like English in that they didn’t violate any phonological or spelling rules, nonetheless had letter combinations that were statistically rarer than those in real words. It was statistical patterns then—nothing “meaningful”—that the baboons were learning. Had a set of nonwords been selected that did not differ statistically from real words, the baboons would have been unable to generalize beyond the letter sequences that they’d memorized.
The authors argue that this study highlights the role of visual and orthographic knowledge in reading. And the baboons’ accomplishment is impressive, especially considering that they first had to learn to distinguish “b” from “p” and “d” before they could start tracking letter combinations. But mastering orthography can only take a wannabe reader so far. In practice, it is extraordinarily difficult for humans to learn to read without already knowing a whole lot about the sounds that letters roughly correspond to. Literacy among deaf adults, for instance, is notoriously variable: while some deaf adults read quite proficiently, a 2000 study in the Journal of Deaf Studies and Deaf Education puts the average deaf 18-year-old’s reading vocabulary at a fourth-grade level.
Needless to say, reading sans meaning is hardly reading. Meaning is so essential to the reading experience that we can’t prevent ourselves from understanding what a word denotes even when we want to. Baboons, unlike us, will never be involuntarily drawn to billboards or ironic T-shirts. Lucky them.
Jessica Love is a contributing editor of the SCHOLAR. She holds a doctorate in cognitive psychology and edits Kellogg Insight at Northwestern University.