Here is an interesting new paper in PNAS. You can read a news article about it here. Ten native Chinese speakers (of Chinese ethnicity) and ten native English speakers (white) were put through four different tasks while their brains were scanned by fMRI: (1) Symbol, deciding whether the third figure in a triplet of figural stimuli is presented in the same orientation as the previous two; (2) Number, same as Symbol except the stimuli are numbers rather than non-semantic figures; (3) Addition, deciding whether the third number is equal to the sum of the first two; and (4) Comparison, deciding whether the third number is larger than both of the first two. The control task consisted of deciding whether the third dot in a triplet of dots was the same color as the previous two. All subjects were comparable in age and of the same handedness. I would include here the slick figure summarizing the results, but Blogger is acting up again. I will have to settle for, uh, merely verbal representations (ha-ha).
While the Symbol condition evoked no differences of note between groups, significant differences emerged in the other three.
The activation in NES [native English speakers] is greater in the left SMA [supplementary motor area], Broca area, and Wernicke area (Wn) [the so-called “language centers”], compared with the corresponding areas in NCS [native Chinese speakers]…. Importantly, much larger brain activation was found at a region in-between BA6, BA8, and BA9 in NCS. We termed this region as a premotor assocation area (PMA), which has previously been associated with visuo-spatial processing …
Interestingly, within-group comparisons also reveal a similar activation pattern between Symbol and Number conditions in NCS…. Such similarity may imply the utilization of a visual-symbol system for representing Arabic digits in Chinese speakers….
For the other three conditions [Number, Addition, Comparison], although similar activated networks were found in the occipito-parietal areas, perisylvian area, and PMA area, the perisylvian activations were significantly larger in NES than those in NCS….
The larger perisylvian activation in NES alone may suggest that the brain representation of numbers is influenced by different language processes. However, across all of the four conditions as the arithmetic loading increased, there was a trend of increase in the premotor activation in NCS but not in NES. Such a trend was also found at the perisylvian area in NES but not in NCS. Therefore, between NCS and NES, there was a double dissociation in the brain activation during these taks, which suggests that the differences may not be merely due to different languages but also due to specific mathematic processes. In other words, whereas the numbers are represented in different brain regions from those involved in languages, people speaking Chinese or English may engage different neural pathways in numerical processing.
I am reminded of a passage from The Pleasure of Finding Things Out where Richard Feynman describes his informal experiments with counting silently in one’s head. He found out that it is possible to count in two different ways: (1) Feynman himself seemed to say the numbers “one, two, three,” and so on to himself, sotto voce as it were; while (2) his friend, the statistican John Tukey, visualized a mental number line and moved along it tick by tick. It has indeed been found in more recent studies that even “elementary” cognitive tasks can be represented in different ways (i.e., verbal-propositional v. spatial-pictorial) across individuals (e.g., Neubauer & Freudenthaler, 1994).
So it seems that the present study has revealed another such difference scaled up to the level of group averages. The authors speculate that “the strong involvement of visuo-premotor association in NCS may be related to the experience of reading Chinese characters…. The use of the abacus in many Asian schools also suggests that, in one way or the other, the engagement of a ‘mental image’ for arithmetic could be related to the differences in brain activation.” Well, maybe so. The logical next step then is to scan the brains of English speakers of Chinese ethnicity, preferably with both a larger sample and a larger control group. It should be possible to find Chinese Americans who do not read or speak Chinese. It is also desirable to ensure that the presents findings are not an artifact of an ability difference between the two groups. A pseudo-race group of white English speakers with an ability profile similar to that of typical East Asians (average or slightly above-average SAT-V scores, high SAT-M scores, high spatial scores) should be compared to a more typical group of whites; if differences in brain activation between whites and the pseudo-Asian group are similar to those in the present study, then the difference is not a true race- or culture-specific difference but rather a mere ability difference. In the latter case, of course, we would still want to know what causes the particular ability profile of East Asians in the first place.
Can the extant literature give us any hints as to such a follow-up would turn out? Check out p. 169 of Nicholas Mackintosh’s textbook IQ and Human Intelligence, which provides mean WISC-R subtest scores of the Japanese standardization sample. (The content of these subtests is described here.) The subtests can be classified as follows by performance of the Japanese relative to whites:
much better than whites: Block Design
better than whites: Arithmetic, Digit Span, Picture Completion, Picture Arrangement, Object Assembly
comparable to whites: Similarities, Digit Symbol
worse than whites: Information, Comprehension
It is clear that Japanese children excel their white peers in mental tests that do not load on the verbal factor. This may be due to an advantage in g, but the Japanese-white gap on Block Design is so large that at this impressionistic level we are forced to also invoke a Japanese advantage in spatial-visualization ability. It is tempting to speculate that perhaps East Asians preferentially employ some form of analogical representation of numbers precisely because of this advantage.
Regardless of its bearing on the processing of number, is the profile difference between whites and East Asians genetic or cultural? I know of one adoption study of East Asian mental abilities that has employed the WISC-R (Frydman & Lynn, 1989). After a correction for the Flynn Effect, it was found that 19 Korean orphans adopted into Belgian families between the ages of 3 to 72 months obtained an average IQ 10 points higher than the Belgian population mean (p < 0.01). Unfortunately, the subtest scores reported in this paper do not appear to be corrected for the Flynn Effect. I have listed the mean scores of these Korean adoptees in decreasing order: Block Design: 13.89
Picture Arrangement: 13.32
Object Assembly: 13.26
Picture Completion: 13.00
Similarities: 12.95
Arithmetic: 12.79
Coding: 12.58
Comprehension: 12.00
Information: 10.37
Vocabulary: 9.63
It is hard to say how much these average scores are shifted rightward of the whie means by the Flynn Effect. But the relative profile of these Korean children raised in Belgium is unmistakably similar to the profile of the native Japanese children.
The nature of white-East Asian differences in mental abilities is complex and has not received nearly enough psychometric research effort as it deserves. But we have here evidence that the typical East Asian profile (high spatial, weak verbal) follows East Asian children regardless of the culture in which they are reared. Hmm. Something to chew on.