Skip to main content

What Goes Together? – Young Infants and Abstract Visual Learning

Figuring out how objects and ideas relate to each other is at the core of learning; this capacity helps the preschooler solve her very first puzzle, the fledgling scientist see the connection between an atom and the solar system, and the young dancer stomp his feet in time to the beat. When you hand your 3-year-old a stack of Legos alternating blue-green-blue and she grabs a green Lego and enthusiastically plunks it on top, even this simplest of actions reveals her ability to discover the increasingly complex relational structure in the world around her.

But what if you handed your 3-month-old the same stack of Legos? Well, for starters you’d be surprised if he did anything at all given his limited motor abilities. But you might also think that no matter how you hard you tried to measure his ability to detect abstract relationships, he’s just too young too see the world in terms of patterns, relations, and structures. A new study from the Project on Child Development, however, suggests you might be wrong – at least in the latter case!

Developmental psychologists have known for a while that infants this young can detect patterns in the speech they hear–that is, if you play speech sounds that follow a specific pattern (e.g., ABB, ba-po-po, go-la-la, etc.) for an infant repeatedly and, when they are sufficiently bored, switch to an ABA sound (e.g., ba-po-ba), the baby will perk up when the new pattern is introduced. This shows that they recognize that these new sounds don’t follow the same relational pattern as before, and consequently increase their attention in an attempt to figure out what’s going on. What’s been harder to establish, however, is whether this relational learning is restricted to speech (after all, infants are precocious language learners!) or whether they can detect the same relational structures in other domains as well, such as vision.

Psychologists have investigated this ability to detect abstract visual relations in a manner similar to the speech studies. First, infants see a series of shapes looming on a big screen in front of them in an ABB pattern (e.g., circle, triangle, triangle) until they get bored of this pattern, and then the pattern switches to ABA (e.g., circle, triangle, circle), or vice versa.  The results have been mixed, with some studies showing that infants can learn some patterns at some ages, but not reliably, and others reporting more consistently positive results by the time infants are about 7 months old. At the Project on Child Development, we set out to answer a couple of key questions: Why were babies better at learning abstract visual rules in some cases than others? And if we could reproduce these more supportive conditions, would we possibly see visual rule learning in even younger infants?

We noted that in the studies suggesting babies had a difficult time reliably learning visual rules, the infants were shown each image one at a time in a sequence, which may have over-burdened their memory capacities. We also speculated that showing the objects in sequence made it more difficult to compare them. By contrast, the one study showing that 7-month-olds successfully learned ABA and ABB patterns in the visual domain – reported by Jenny Saffran and colleagues at the University of Wisconsin at Madison – presented the pictures in a sequence such that they each remained on the screen until all three had appeared, and for a bonus second afterward. Might this simple change in presentation style have supported their relational learning? And, if it did, would it also support even younger infants’ learning of the same rules?

To address these questions, in our study, we tested 3- and 4-month-olds using the same method. The infants were shown trios of dog images in either an ABA (e.g., Dachsund, Beagle, Dachsund) or ABB (e.g., Dachsund, Beagle, Beagle) pattern until they became sufficiently bored, at which point we showed them several sets of three new kinds of dogs that followed either the familiar pattern or a new one. Remarkably, infants as young as 3 months old successfully learned the abstract rule in these conditions—they significantly preferred looking to the pattern they had previously been exposed to. Notice that this preference is different than the one we see in other studies, where babies prefer something new and surprising. It suggests that these much younger infants wanted to learn more about the rule they had seen before, as if they were still in the process of committing it to memory. And indeed, the older the infants were, the more likely they were to look at the novel rule. This suggests that they were beginning to learn the rule quickly enough to say, “Yeah, I got this–now show me something new!”

So although your 3-month-olds may not be putting Legos together anytime soon, they are already capable of learning how objects relate to each other! At the Project on Child Development, we look forward to continuing to investigate how these extraordinary little scientists learn to make sense of the world around them.

Ferguson, B. & Waxman, S. R. (in press). Visual abstract rule learning by 3- and 4-month-old infants. In R. Dale, C. Jennings, P. Maglio, T. Matlock, D. Noelle, A. Warlaumont, & J. Yoshimi (Eds.), Proceedings of the 37th Annual Conference of the Cognitive Science Society. Austin, TX: Cognitive Science Society.