Adapted from Liu et al. (2012) (Open Access).
Figure 6.4 fMRI in pianists with varying degrees of improvisation experience. More training is related to less brain activity (blue) during creative expression and to increased functional connectivity among other areas (red).
Reprinted with permission from Pinho et al. (2014, their figure 3). Free Access.
A comprehensive review of 45 functional and structural neuroimaging studies of creative cognition (not limited to musical improvisation)reached a similar conclusion (Arden et al., 2010). There was a range of different creativity measures across the studies, often only one test score per study, and different imaging methods were used. Perhaps not surprisingly, the results showed disappointingly little overlap among the studies. Figure 6.5, for example, shows the inconsistencies among seven fMRI studies. The authors concluded that a more standardized approach to creativity assessment was necessary for any progress. They proposed eight suggested goals and actions to accomplish them: “(1) Goal: discover whether creative cognition is domain-specific. Action: test people phenotypically across many domains of creative production to quantify the common variance. (2) Goal: increase reliability of the measure. Action: use exploratory factor analysis – administer diverse creative cognition test batteries to large samples (N > 2,000). (3) Goal: improve discriminant validity. Action: include intelligence (indexed by a reliable IQ-type test) and openness to experience (assessed by a reliable personality test) as covariates. (4) Goal: improve ecological validity of the criterion. Action: use evolutionary theory to inform or guide test development. (5) Goal: explore the etiology of creative cognition. Action: administer creative cognition tests to genetically informative samples such as twins. (6) Goal: improve confidence in our results. Action: increase sample sizes. (7) Goal: increase comparability across studies. Action: converge on a common brain nomenclature. (8) Goal: increase power of detecting effects. Action: move to study designs that use continuous measures rather than dichotomies such as case-control.” Another contemporaneous comprehensive review (Dietrich & Kanso, 2010) noted some general consistencies among creativity studies, including a pattern of activations and deactivations involving frontal areas as well as other areas distributed across the brain in both hemispheres (contrary to the popular idea that creativity is principally a right hemisphere function). A subsequent critical review came to similar conclusions and listed suggestions for future research that emphasized the important role for collaboration between creativity researchers and cognitive neuroscientists (Sawyer, 2011).
Figure 6.5 Different creativity findings from seven MRI studies. Each colored symbol shows activated brain areas related to creativity from a different study. There is little overlap of areas across studies.
Reprinted with permission from Arden et al. (2010, their figure 1).
Rex Jung and I attempted to integrate neuroimaging findings from intelligence studies and creativity studies and relate them to genius (Jung & Haier, 2013). We focused on consistencies from structural imaging and lesion studies of creativity because they avoid problems of task-specific results that confound functional imaging studies and are a major source of inconsistent results. One study of 40 lesion patients who completed creativity tests was particularly informative because lesions in some areas were associated with deficits in different aspects of creativity (Shamay-Tsoory et al., 2011). Studies of FTD, noted previously in this chapter, were also informative. Based on a combination of these studies, we proposed the Frontal Dis-inhibition Model (F-DIM) of creativity (Jung & Haier, 2013). Figure 6.6 shows this model, designed for easy comparison to the intelligence PFIT model (see Figure 3.7). Only four F-DIM areas overlap with PFIT (BAs 18/19, 39, and 32), suggesting mostly independent networks for intelligence and creativity. In comparison to the 37 studies that were reviewed for the PFIT, the F-DIM is more tentative because it is based on a smaller number of structural-only imaging studies. The essence of the F-DIM is that networks related to creativity are mostly dis-inhibitory, especially in frontal and temporal areas that affect other parts of the frontal lobes, the basal ganglia (part of the dopamine system), and the thalamus (an important relay station for information flow) through white matter connections.
Figure 6.6 Frontal Dis-inhibition Model (F-DIM) of creativity. Numbers indicate Brodmann areas associated with increased (up arrows) or decreased (down arrows) brain activity based on a review of studies. Blue is left lateralized; green is medial; purple is bilateral; yellow arrow is anterior thalamic radiation white matter tract.
Reprinted with permission, Jung & Haier (2013).
With respect to how the F-DIM and the PFIT might relate to genius, we speculated that, “…we must look not only to increased neural tissue or activity in key brain regions (e.g., frontal lobes), but perhaps also to some mismatch between mutually excitatory and inhibitory brain regions (e.g., temporal lobes) that form a network sub-serving such complex human behaviors as creativity (e.g., planning, insight, inspiration). This notion of a delicate interplay of both increases and decreases in neural mass, white matter organization, biochemical composition, and even functional activations within and between brain lobes and hemispheres, is an important concept. Indeed, it is the rare brain that has highly developed networks of brain regions sub-serving intelligence and (concurrently) the somewhat underdeveloped network of brain regions associated with dis-inhibitory brain processes associated with creative cognition. Such a finely tuned seesaw of complex higher and lower brain fidelity, balanced in dynamic opposition, would almost guarantee the rare occurrence of genius” (Jung & Haier, 2013). Or as we say privately, we really don’t know how intelligence and creativity are related to genius on the brain level.
Another comprehensive review of creativity research was based on a meta-analysis of 34 functional neuroimaging studies that included 622 healthy adults (Gonen-Yaacovi et al., 2013). A main analysis examined whether there were brain areas consistently activated despite the diversity of creativity tasks performed during the imaging. The analysis, however, was limited because it did not include areas of deactivation. The activation results for all studies together indicate some consistency, as shown in Figure 6.7. The resulting creativity map is consistent with the F-DIM and other studies, showing a distribution of salient areas including frontal and parieto-temporal regions, especially the lateral prefrontal cortex. Some of the creativity tasks required generation of ideas and other tasks required combination of elements. Separate analyses for both kinds of task suggested anterior regions were involved in combining ideas creatively and more posterior regions were involved in freely generating novel ideas. There were also some differences between verbal and non-verbal tasks. In the case of both the shared creativity map (Figure 6.7) and the findings of the two kinds of tasks (not shown), areas in both the right and left hemispheres are associated with creativity, providing additional evidence that creativity is not an exclusive function of the right-sided brain. A re-analysis that includes areas of deactivation would be informative for a more complete picture given other findings related to dis-inhibition. In fact, a newer meta-analysis of ten small-sample fMRI studies of divergent thinking published as this book is finalized shows widespread areas of deactivation although, inexplicably, the Gonen-Yaacovi analysis is not cited (Wu et al., 2015). Also just published, a structural MRI study in 135 adults reported correlations between gray matter and a test of creative fluency and a test of creative originality. Each test was correlated with gray matter in different areas and there was an interaction with intelligence only for fluency (Jauk et al., 2015). This field is attracting new attention and the number of imaging studies of creativity is increasing rapidly. This area is moving closer to claiming a weight of evidence for some findings, so stay tuned.
Figure 6.7 Summary findings from 34 functional imaging studies of creativity. Common brain areas of activation are shown revealing distributed networks related to creativity. From Gonen-Yaacovi et al. (2013, their figure 1). (Open Access).
H
ere is a final speculation for this section. If a deep level of dis-inhibition in certain brain circuits results in unconsciousness, perhaps a bit less dis-inhibition may increase creativity. The perception of increased creativity is often a subjective response to “mind-expanding” drugs like LSD. Dis-inhibition of the frontal cortex is also associated with dreaming during sleep (Muzur et al., 2002). Obviously, sleep is an unconscious state and dreams frequently are quite creative in content and narrative. Tying creativity and consciousness research together based on neuro-circuits would further demonstrate there is a neural basis for creativity. There also is some genetic evidence regarding creativity (Ukkola-Vuoti et al., 2013), suggesting there might be a potential for enhancing creativity by affecting brain mechanisms. Many drugs have been described subjectivity as creativity enhancers, but I am unaware of compelling empirical research that substantiates such observations. Increased creativity has been reported in a few studies that manipulate the brain without drugs (Fink et al., 2010) and there is a small tDCS study (Mayseless & Shamay-Tsoory, 2015), but so far there is no weight of evidence to support these preliminary reports. How intelligence may be related to creativity and consciousness on a neural level is an intriguing question that raises opportunities for imaginative research designs and innovative neuroscientists. Students, that means you.
6.6 Neuro-poverty and Neuro-Social–Economic Status (SES): Implications for Public Policy Based on the Neuroscience of Intelligence
The confounding of SES with intelligence was introduced in Section 2.1. Now we consider it further because it remains an important problem that often results in misleading conclusions from research studies. Here is a common train of thought about the importance of SES: Higher income allows upward mobility, especially the ability to move from poor environments to better ones. Better neighborhoods typically include better schools and more resources to foster children’s development so that children now have many advantages. If the children have high intelligence and greater academic and economic success, it could be concluded that higher SES was the key factor driving this chain of events. Here is an alternative train of thought: Generally, people with higher intelligence get jobs that require more of the g-factor and these jobs tend to pay more money. There are many factors involved, but empirical research shows g is the single strongest predictive factor for obtaining high-paying jobs that require complex thinking. Higher income allows upward mobility, especially the ability to move from poor environments to better ones. This often includes better schools and more resources to foster children’s development so that children now have many advantages. If the children have high intelligence and greater academic and economic success, it could be concluded that higher parental intelligence was the key factor driving this chain of events, due in large part to the strong genetic influences on intelligence.
The latter train of thought is hardly new. It was made clear more than 40 years ago in a controversial book mentioned earlier in Chapters 1 and 2, IQ in the Meritocracy (Herrnstein, 1973). The argument was reduced to its simplest form in a syllogism: “(1) If differences in mental abilities are inherited, and (2) if success requires those abilities, and (3) if earnings and prestige depend on success, (4) then social standing (which reflects earnings and prestige) will be based to some extent on inherited differences among people” (pp. 197–198, italics added). When this was published in 1973, the evidence for a genetic role in intelligence was strong but not overwhelming and there was room for skepticism; today the evidence is overwhelming and compelling (see Sections 2.5, 2.6, 4.5, and 4.6).
Dr. David Lubinski has written a comprehensive review of the SES/intelligence confounding issue (Lubinski, 2009). Although the context for his paper is cognitive epidemiology, the argument applies to all research using SES as a variable. Essentially, if a study incorporates measures of both SES and intelligence, statistical methods can help disentangle their respective effects. The interpretation of results from any study of SES cannot disentangle which factor is driving the result unless a measure of intelligence is included in the study. Studies of intelligence without considering SES are also problematic. When both variables are included in multivariate studies in large samples, the results typically show that general cognitive ability measures correlate with a particular variable of interest even after the effects of SES are statistically removed. For example, in a study of 641 Brazilian school children, SES did not predict scholastic achievement, but intelligence test scores did (Colom & Flores-Mendoza, 2007). An even larger classic study had data on 155,191 students from 41 American colleges and universities. Their analyses showed that SAT scores predicted academic performance about the same even after SES was controlled; that is, SES added no additional predictive power (Sackett et al., 2009). Another study of 3,233 adolescents in Portugal found that parents’ level of education predicted intelligence in the children regardless of family income (Lemos et al., 2011). These researchers stated their conclusion straightforwardly: “Adolescents from more affluent families tend to be brighter because their parents are brighter, not because they enjoy better family environments.”
Studies with equally large samples showing SES effects remain after removing effects of intelligence are less frequent, although one meta-analysis suggested that SES independently predicts economic success about as well as intelligence (Strenze, 2007). An illustrative example of using both SES and IQ is a study of 110 disadvantaged middle-school children. It included maternal IQ along with composite measures of parental nurturance and environmental stimulation (Farah et al., 2008). In the main analysis, parental nurturance was related to memory and environmental stimulation was related to language, after any effects of maternal IQ were statistically removed. The range of maternal IQ, however, was restricted to the lower end of the normal distribution (mean = 83, standard deviation = 9), possibly explaining the lack of an IQ finding, but this study does illustrate why it is important to include IQ measures when investigating specific SES factors. Replication in another sample of disadvantaged children would be important along with obtaining fathers’ IQ. Replication in a sample of children in higher SES levels would also be informative, as would studies of children at different ages since the effects of SES on the heritability of intelligence may vary with age (Hanscombe et al., 2012). It is particularly interesting that there is emerging evidence that the SES itself has a strong genetic component (Trzaskowski et al., 2014). Obviously, there are many questions to pursue for establishing a weight of evidence regarding how SES and IQ relate to each other.
One common view in cognitive psychology is that SES/cognitive relationships are mediated by how SES variables influence brain development during early childhood. Other researchers see such relationships as more related to neuroscience, especially when trying to relate such findings to education (Sigman et al., 2014). As you might imagine, the line between cognitive psychology and neurobiology is permeable (Hackman et al., 2010; Neville et al., 2013). The term “cognitive neuroscience” refers to both. Nothing about a major genetic component to intelligence and related neurobiological mechanisms negates or minimizes the importance of SES influences on cognitive psychology variables. Surely, SES is a consequence of many factors, but let’s consider just the portion of SES that is confounded with the genetic portion of intelligence. I designate this portion by the term “neuro-SES” and in my view it should be recognized as a matter for research and discussion.
To repeat the main point, studies that make claims about SES variables without including measures of intelligence are difficult to interpret and need to at least acknowledge the confounding problem before concluding or implying that SES has a causal role. This was a primary point made two decades ago in The Bell Curve. Nonetheless, bias toward SES-only explanations remains prevalent. Two recent high-profile examples illustrate the issue. Both studies use neuroimaging with structural MRI. The first paper is from MIT, reported by Dr. Mackey and colleagues (2015) (Mackey also reported a 10-point IQ increase in disadvantaged children followi
ng brief computer game playing in school; see Section 5.3). These researchers set out to study neuroanatomical correlates of the academic achievement gap between higher- and lower-income students (n = 35 and 23, respectively). The higher group average yearly family income was $145,465 (95% confidence interval between $122,461 and $168,470). The lower group family average was $46,353 (95% CI between $22,665 and $70,041). It is arguable whether family incomes of over $50,000 constitute a disadvantaged household, but the key finding is still of interest. Structural MRIs showed greater cortical thickness in several areas for the high-income group, although other brain measures did not (e.g., cortical surface area, cortical white matter volume). Cortical thickness differences between the groups in some areas were related to standard test score differences. The authors concluded, “Future studies will show how effective educational practices support academic gains and whether these practices alter cortical anatomy.” This is fair enough and certainly supports a commonly held view. However, without assessing the cognitive ability of the parents, we cannot be sure whether the cortical thickness difference is related to family income or to the genetics of intelligence. The results from this study would be far more compelling had some estimate or measure of parental intelligence been included to help disentangle SES effects from intelligence effects.
The second paper is a multicenter collaboration reported in Nature Neuroscience by Dr. Noble and colleagues (2015). This MRI study had a large sample of 1,099 children and adolescents. Data included family income, parental education, and genetic ancestry. Income was related to brain surface area even after controlling for parental education. Parental education related to other structural brain characteristics even after controlling for income. These associations were found irrespective of genetic ancestry. The authors state that “… in our correlational, non-experimental results, it is unclear what is driving the links between SES and brain structure. Such associations could stem from ongoing disparities in postnatal experience or exposures, such as family stress, cognitive stimulation, environmental toxins or nutrition, or from corresponding differences in the prenatal environment. If this correlational evidence reflects a possible underlying causal relationship, then policies targeting families at the low end of the income distribution may be most likely to lead to observable differences in children’s brain and cognitive development.” This is not an unreasonable statement, but one implication of this train of thought might be an experiment that provided modest or large monthly payments to low-income families to improve everyday life with the expectation that the resulting life changes might have subsequent effects on their children’s brain and cognitive development. Some recognition and discussion of the neuroscience aspects of intelligence and its intertwining with SES would be important considerations if such an experiment was undertaken. Intelligence was not mentioned in the discussion of these MRI results.
The Neuroscience of Intelligence Page 26