From Science Daily: 

Two vital parts of mentally organizing the world are classification, or the understanding that similar things belong in the same category; and induction, an educated guess about a thing’s properties if it’s in a certain category. There are reasons to believe that language greatly assists adults in both kinds of tasks. But how do young children use language to make sense of the things around them? It’s a longstanding debate among psychologists.

A new study in Psychological Science, a journal of the Association for Psychological Science, challenges the predominant answer. “For the last 30 to 40 years it has been believed that even for very young children, labels are category markets, as they are for adults,” explains psychologist Vladimir M. Sloutsky, who authored the paper with Ohio State University colleague Wei Deng. According to this theory, if you show anyone an oblong, scaled, limbless swimming thing and say it’s a dog (its label), both adults and children will believe it’s a dog (in that category of four-legged domesticated mammals) and should behave like a dog — bark or wag its tail.

The study confirms that many adults do use labels this way. But children do not. “Our research suggests that very early in development labels are no different from other features,” says Sloutsky. “And the more salient features may completely overrule the label.” You insist the swimming thing is a dog. The child weighs all the evidence — and “dog” is no more important than scales or swimming — and concludes it’s a fish.

To test their hypothesis, the psychologists showed pictures of two imaginary creatures to preschoolers and college undergraduates. Both animals had a body, hands, feet, antennae, and a head. The “flurp” was distinguished by a pink head that moved up and down; the “jalet” had a blue sideways-moving head. The heads were salient — the only moving part. During training, the subjects learned what a flurp or a jalet looked like.

Then the experimenters changed some of the features, keeping the head consistent with most of them, and asked participants to supply the missing label. They also showed creatures with characteristics and a name, and the subjects had to predict — induce — the missing part. Both adults and children did best when the head was consistent with the name.

The difference arose when the head was a jalet’s but label was “flurp,” or vice-versa. Then, most of the adults went with the label (we accept that a dolphin is a mammal, even though it looks and swims like a fish). The children relied on the head for identification. Regardless of its name, a thing with a jalet’s head is a jalet.

This is interesting for a number of reasons. First, it seems to call into further question the rote skill-and-drill-flashcard-style learning utilized in behavioral models of intervention. If, developmentally, children are processing the sensory, motor, and affective attributes and affordances of an object in order to learn how to classify it, this would be best facilitated in the context of guided, hands-on exploration with a supportive adult. The study’s authors agree:

Sloutsky says the findings could inform teaching and communicating with children. “If saying something is a dog does not communicate what it is any more than saying it is brown, then labeling it is necessary but by no means sufficient for a child to understand.” Talking with young children, “we need to do more than just label things.” 

The second reason I find this interesting is more speculative; I have often wondered whether many of the hallmark symptoms of autism and the secondary deficits observed in sensory processing disorder, such as language delay and alleged theory of mind differences might be traced back to poorly integrated processing (i.e. rather than thinking of these as primary deficits, that they are the result of piecemeal processing and the resulting spotty connectivity between sensory memory stores). For example, in this case, if a child is taking in the visual shape of the duck, but not forming pathways that connect and integrate that information with the texture of the feathers, the quack, or the characteristic waddle, flap, and swimming motions of a duck, how much more challenging would it be for them to acquire the functional, working hypothesis of “duckness” required to label a novel creature with the label, “duck?”  

Read the full article here.