Gary Lupyan Discusses Labels’ Impact on Thought

Editor’s note: This article was initially published in The Daily Gazette, Swarthmore’s online, daily newspaper founded in Fall 1996. As of Fall 2018, the DG has merged with The Phoenix. See the about page to read more about the DG.

Gary Lupyan, a postdoctoral fellow at the University of Pennsylvania, came to Swarthmore last Thursday to give a talk about the impact of language on our thought processes. The lecture, titled “What Do Words Do?”, posited that giving labels to objects can have a powerful effect on the way we think about, perceive and remember those objects.

The key point of Lupyan’s talk was that category labels could both improve and impair people’s perception and memory: words are a powerful tool for sorting sights and memories, but also hide the details of the things they describe.

Lupyan’s main theory about labeling says that our mind works with top-down feedback loops, wherein our perception of the world affects the higher-level way we conceive of things, which in turn determines the way we label them. Then, the labels we’ve created feed back into our conceptions of these objects, shaping the way we categorize what we see; this conception even feeds back into our perception, deciding what information we absorb from the world and how we interpret that information.

In a study conducted at Carnegie Mellon, Lupyan asked students to look at strange blue sculptures of aliens. Some of these, he told them, could be approached, while the others should be avoided. The difference between the aliens was a minor one—the shape of their heads varied slightly—and students learned through trial and error how to tell them apart. Half of these students, the experimental condition, were told that the friendly aliens were called “leebish” and the malevolent ones were “grecious,” while the control group didn’t learn names for the aliens. Lupyan found that the students who had those nonsense labels for the aliens not only learned more quickly how to differentiate the creatures, but also retained this knowledge after they stopped getting feedback on their performance. The performance of the control group gradually declined, while the students who could name the aliens continued to perform almost perfectly.

Lupyan’s conclusion from this study was that labeling the aliens helped the students to mentally solidify the two categories: having a word for a group “set up a hypothesis for classification.” This hypothesis allowed the subjects to distinguish more quickly between members of the groups and to remember which were which when they were no longer given feedback on their responses.

Another study that Lupyan ran focused on the recollection of items that participants first saw and were later quizzed on. Subjects in the control group in this experiment responded to each image they saw (of a long-legged chair, for instance) based on whether they liked it or not; those in the experimental group were asked to name the categories that the images belonged to. The results of this study, Lupyan said, showed that labels could actually impair people’s memories of what they had seen: those who labeled the items were worse at recognizing ones they had already seen. These data suggested that once we’ve categorized an item, we don’t feel obliged to remember as many details about it — or, as Lupyan put it, “labeling [an object] produces a distortion of the original item.”

The third type of study that Lupyan ran focused on the effect of a redundant label stimulus on people’s perception. Subjects in one of the studies were either prompted with the letter they might be about to see, “emm,” or heard silence; then, sometimes, they very briefly saw an M and then an obscuring image, or sometimes just the final image. Subjects who were prompted with the M were much better at identifying it when it was there than those who heard only silence, even though the timing was the same for both groups. Lupyan and his colleague, Michael Spivey of Cornell University, found that valid cues helped subjects to identify what they had seen, while invalid cues did not: thus, somehow being reminded of the label immediately before seeing a letter influenced subjects’ perception of the world in front of them.

Lupyan (who, his introducer said, was rejected from admission to Swarthmore as an undergraduate) concluded his lecture by saying, “Language evokes mental states. That is its purpose.” Words can harm or help our performance on certain tasks, having a powerful influence on the way we see and make sense of our environments.

After the talk, audience members asked Lupyan about further applications of his work; Assistant Professor Jodie Baird of the Psychology department asked whether labels like the ones Lupyan studied might influence people’s ability to differentiate between members of social categories like those divided by race and gender. Lupyan acknowledged having little solid data on the subject, but said that “just a few minutes of categorization [tasks] had a noticeable effect on ability to observe and remember within-category differences.” He hypothesized that the same effect would hold true for sorting people into groups; thus, using labels for groups of people causes us to think of the members as more similar to each other, and more distinct from other groups.