Comparing Child and Adult Brains: How to Account for Performance Differences?

In an ideal world, we’d be able to study maturational brain changes by scanning a group of adults, a group of children, and comparing the brain images. Unfortunately, there are complications.



One complication is that these studies usually require doing some kind of task in the scanner, and children usually have lower accuracy and longer reaction times on this task. These differences, especially reaction time differences, can have a significant effect on brain activation. (Activation is averaged over seconds, so the longer your reaction time, the higher the brain activation, simply because you spend more time on the task). So how do we know what differences are due to actual brain maturation and what differences are due to poor performance inside the scanner?

In a 2002 study, Schlaggar and colleagues addressed the issue of performance differences by comparing only those children and adults that performed similarly on the task. They were interested in word processing in children (aged 7-10) and adults (age 18-35). In their experiment, participants saw single words on a screen and had to say a response word based on a cue (for example, a rhyming word, or the opposite word).

Instead of comparing all adults and all children, Schlaggar and colleagues divided the participants into two subgroups. The top scoring children and lower scoring adults to formed a Performance matched subgroup, where the children’s performance did not differ significantly from adults. The rest formed a Performance Non-matched subgroup, where there were clear differences in accuracy and response time between adults and children.

Schlaggar and colleagues looked at several regions of interest in the left frontal and left extrastriate regions, all traditional language and word processing areas. When there were differences between age groups, the children almost always had greater activation.

But the interesting results occur when you compare the subgroups. Some regions showed differences in the Non-Matched subgroup that disappeared when you look at the Performance Matched groups, suggesting that the brain differences here were due to performance differences inside the scanner.

Other regions, however showed differences between children and adults in both the Performance Matched and the Performance Non-matched subgroups. In these regions, one can safely assume that there’s more to these differences than simple in scanner performance.

What does this tell us? For one thing, it tells us that in-scanner performance differences between children and adults should not be ignored. There were several “developmental” differences here that disappeared as soon as you controlled for in-scanner performance. On the other hand, there do appear to be differences that remain even when you have children and adults that perform equally.

A few things to think about with this study. First, it’s admirable that the authors control for performance, but doing so also introduces opposite selection biases in children and adults. You have to wonder what population of children would perform as well as adults almost twice their age, and conversely, what population of adults would perform at the level of 7-10 year olds. Is it fair to compare these two groups and generalize these comparisons to the entire population?

Second, what does it mean to control for in-scanner performance? If we treat it as a confounding factor and control for it, we’re assuming that performance differences on this word generation task are irrelevant to the process we’re studying. However, that can’t be completely true. We’re interested in word processing differences between children and adults, so if we look for children and adults that perform similarly on a word generation task, we’re filtering out some of the differences that we set out to study.

Third, it might be helpful, as mentioned in BJ Casey’s commentary on the study, to differentiate between differences from maturation alone and differences due to skill level. In a field such as reading, this might be hard to tease apart. While children’s brains mature between ages 7 and 18, they also undergo thousands of hours of reading instruction that introduce changes in the brain. When we study reading acquisition in children, therefore, we should think about kind of brain changes we’re interested in, and that will affect the comparisons and analyses we do.

Schlaggar BL, Brown TT, Lugar HM, Visscher KM, Miezin FM, & Petersen SE (2002). Functional neuroanatomical differences between adults and school-age children in the processing of single words. Science (New York, N.Y.), 296 (5572), 1476-9 PMID: 12029136

Casey, B. (2002). NEUROSCIENCE: Windows into the Human Brain Science, 296 (5572), 1408-1409 DOI: 10.1126/science.1072684