Jung & Haier (J&H) present an extensive review of brain activation during the performance of psychological tasks that have traditionally been assumed to reflect intelligence in general, such as IQ tests, logical reasoning, fluid-crystallized measures, and reasoning in games such as chess and Go. Although the authors document many important similarities in the location of relevant brain areas during the performance of many of these tasks, all the studies they reviewed are based on samples of untrained participants. This type of research cannot address the critical question of whether the same structures of parieto-frontal integration theory (P-FIT) would be activated, and thus mediate individual differences, in the execution of high levels of skilled performance.

Over the last two decades there has been remarkable progress in measuring expert performance scientifically (Ericsson et al. Reference Ericsson, Charness, Feltovich and Hoffman2006), particularly in finding valid metrics of superior achievement. Reproducibly superior performance has been successfully captured by recreating ecologically valid and representative tasks in the laboratory (Ericsson Reference Ericsson, Ericsson, Charness, Feltovich and Hoffman2006a; Reference Ericsson, Ericsson, Charness, Feltovich and Hoffman2006b) that typically predict objective performance in skill domains far better than do subject metrics, such as social judgments of expertise. When we limit our review to only objective measures of performance in domains of skill, we find that traditional intelligence tests are not related to individual differences among skilled and expert performers. In general, our reviews show that although intelligence is frequently correlated with initial performance on an unfamiliar task, after extended periods of skill acquisition the relation is no longer statistically reliable. For example, a recent longitudinal study of children's improvements in chess found that the predictive power of IQ diminishes dramatically as skill improves, and did not predict rate of improvement after accounting for practice activity (Bilalič Reference Bilalič2006). Research on expert performance has also found that full-scale IQ tests and heavily g-loaded tests (e.g., Raven's matrices) are not reliably correlated with expert performance in many types of intellectual domains, such as chess (Doll & Mayr Reference Doll and Mayr1987), Go (Masunaga & Horn Reference Masunaga and Horn2001), and several others (Ericsson & Lehmann Reference Ericsson and Lehmann1996). Although a common objection to these findings is that the entire sample of skilled chess players have an average IQ that is higher than the mean of the general population (e.g., Doll & Mayr Reference Doll and Mayr1987), the lack of a relation persists even in samples that vary widely in skill and IQ (Grabner et al. Reference Grabner, Neubauer and Stern2006; Unterrainer et al. Reference Unterrainer, Kaller, Halsband and Rahm2006), indicating that restriction of range cannot be the explanation. Additionally, IQ is not a requirement for high performance, given the numerous documented cases of individuals achieving extreme levels of achievement with IQs below 100. For example, some of the grandmaster chess players in Doll and Mayr's (1987) study had IQ scores below the normative mean; and even in the verbal board game SCRABBLE, some top players have below-average verbal ability (Tuffiash et al., submitted). That IQ and cognitive-ability tasks fail to reliably predict objective measures of domain achievement raises doubts as to whether the underlying mechanisms overlap.

Indeed, research on processes mediating reproducibly superior expert performance shows that the mediating mechanisms differ fundamentally from those used by novices (Ericsson Reference Ericsson, Ericsson, Charness, Feltovich and Hoffman2006a) – that is, during years of practice and training, experts acquire elaborate mechanisms for encoding and maintaining flexible access to critical task information that bypass basic capacities, such as short-term memory capacity (Ericsson & Kintsch Reference Ericsson and Kintsch1995). Moreover, there is an impressive body of evidence showing that with increasing level of skill there are changes in the patterns of neural activation (Hill & Schneider Reference Hill, Schneider, Ericsson, Charness, Feltovich and Hoffman2006); and some evidence even suggests that intense training can change functional and structural aspects of the brain (Ericsson Reference Ericsson, Ericsson, Charness, Feltovich and Hoffman2006b). For example, early and extended training has been shown to change the cortical mapping of the brain area controlling fingers of string players (Elbert et al. Reference Elbert, Pantev, Wienbruch, Rockstroh and Taub1995) and the flexibility of fingers (Ericsson & Lehmann Reference Ericsson and Lehmann1996). Also interesting is the recent finding that intense music practice influences the development of myelin around nerves in critical brain regions (Bengtsson et al. Reference Bengtsson, Nagy, Skare, Forsman, Forssberg and Ullen2005). Notably, several studies have also found that regions of brain activity may dramatically change as chess skill increases (Amidzic et al. Reference Amidzic, Riehle, Fehr, Wienbruch and Elbert2001; Grabner et al. Reference Grabner, Neubauer and Stern2006; Volke et al. Reference Volke, Dettmar, Richter, Rudolf and Buhss2002), which further challenges the results based on novice samples reviewed by J&H. Other studies have shown evidence that experts use different neural regions than the ones novices use in many other intellectual domains as well, including memory (Maguire et al. Reference Maguire, Valentine, Wilding and Kapur2003) and mental calculation (Pesenti et al. Reference Pesenti, Zago, Crivello, Mellet, Samson, Duroux, Seron, Mazoyer and Tzourio-Mazoyer2001), and even in taxi driving (Hartley Reference Hartley, Maguire, Spiers and Burgess2003). In general, achieving expert proficiency in a domain requires thousands of hours of effortful deliberate practice involving problem solving and intense concentration (Ericsson Reference Ericsson, Ericsson, Charness, Feltovich and Hoffman2006b; Ericsson et al. Reference Ericsson, Krampe and Tesch-Römer1993), and critical changes in neural substrates will accumulate over this period of time, in a manner that cannot be elicited during a brief training period with the tasks.

If authentic skilled performance activates different brain regions than does unskilled performance on lab tasks, the regions outlined by P-FIT may not influence or constrain skilled performance in naturalistic settings. If extended deliberate practice can transform and adapt physical, cognitive, and neural structures, then the search for the ultimate neurological circuit or brain structure underlying intelligence may be misguided. We would recommend that J&H extend their research to brain activation during experts' performance on ecologically valid and representative tasks and compare these patterns to those activated by conventional tests of intelligence. We also believe that longitudinal studies of the decade-long acquisition of skilled and expert performance would be necessary to allow us to assess the changes in brain activation and their relation to extensive training periods. Ultimately, a general theory of mental ability and intelligence must account for individual differences in cognitive performance throughout skill development, from unskilled beginners to the highest levels of domain achievement.