Making perceptual categories, or why Nora calls me "Daddy."

(source of original images: Wikipedia. See license)

When my daughter Nora was 15 months old, she had started to pick up a few words. She could say "Mommy," and "Daddy," and "JimmyNO," which is what she called her brother. However, she had only two words for animals: all animals were called either "duck" or "dog." Animals with two legs were ducks, and animals with four legs were dogs. This worked fine, even when her grandparents took her to the zoo. The warthog was a "dog," and so was the giraffe and the zebra. Peacocks, pigeons, and kangaroos were all "ducks" to Nora. The only difficulty came when we arrived at the gorilla habitat. She paused for a second or two, but finally there was a smile of understanding: "DADDY!"

Nora had added another category to her structure of the animal world. Some two-legged animals weren't ducks. Big, hairy two-legged animals with grasping hands were "daddy," a category which included both gorillas and the man who changed her diapers and read her bedtime stories.

So how do we learn to make categories? As Nora's example shows, it's actually quite a difficult task. After all, "Daddy" looks different when he's just gotten out of the shower, compared to, say, when he's ready to head off to work. There are times when "Daddy" doesn't look much different from a gorilla. Certainly there are birds that look more different from each other than humans and gorillas do. How's a toddler supposed to know that an emu and a hummingbird are the same kind of thing? For that matter, how can she learn that, in another way, a blue jay is different from a bluebird? As she grows older and sees more and more birds, will she be able to naturally learn the differences, or does she need more specialized training?

This latter question is what a team of researchers addressed in a recent study (James Tanaka, Tim Curran, and David Sheinberg, of University of Victoria, University of Colorado, and Brown University, "The Training and Transfer of Real-World Perceptual Expertise," Psychological Science, 2005). They wanted to know how bird experts can easily distinguish between species, and quickly commit additional species to memory, while novices have trouble telling a duck from goose.

Tanaka et al. trained two groups of volunteers to recognize pictures of owls and wading birds, but they trained each group in a different way. One group learned the species names of a set of 10 different owls. They were also shown a similar set of wading birds, but these they only identified as "water bird." There were six different pictures of each bird, so the participants could learn to recognize them from different angles and in different body positions. The second set of volunteers was trained in the same way, but taught species names for the wading birds and only learned the general category "owl."

After a week-long training period (this was a dedicated set of volunteers!), participants were given a simple test of knowledge. They were shown two pictures of birds in succession, and asked to indicate whether they were the same or a different species. For example, take a look at this picture:

These are both great gray owls, even though in some ways they look fairly different.

Now look at this picture:

The owl on the left is a great gray owl, but the owl on the right is a northern hawk owl. This is harder than you thought, isn't it?

When the pictures in the test were chosen from the set which had been used for training, all participants did better than they had done on a pre-test. However, if they had been trained with owl species names, they did much better on the test for owls than the test for wading birds, even though they had been trained on just as many wading bird pictures as owl pictures. When shown pictures of new species of owls that had never been used in training, they still performed better than they did before the training. However, they were no better with new wading birds than they had been before. The results were similar for the group that had been trained on wading bird species.

So it appears that "experts" have learned not only how to classify items they have seen before, but they have also learned a process that helps them classify new items within their field of expertise. Other researchers have done brain scans of people while they categorize items, and found that the part of the brain called the fusiform gyrus is activated shortly after they see an object in their area of expertise, but this region is not activated in novices. After novices are trained to categorize objects, their fusiform gyrus is also activated. This suggests that experts are using a different mental process to categorize objects than novices are.

Nora, now 11 years old, has since become an expert in many fields, including knitting, skiing, geometry, and broadway show tunes. She can even tell the difference between her dad and a gorilla. Now if only we could apply those skills to putting away her clean laundry....