Education and Neuroscience Symposium

Paper presented at the British Educational Research Association Annual Conference, University of Warwick, 6-9 September 2006

Education and Neuroscience Symposium

Educating to improve working memory function

Lessons from cognitive neuroscience research into giftedness and high intelligence

John G. Geake

OxfordBrookesUniversity

Abstract

Neuroimaging studies which compare the neural function of high-IQ adult subjects with those of average IQ report that high-IQ subjects display relatively enhanced inferior lateral pre-frontal cortical (ILPFC) activations, together with relatively enhanced activations in a network of other cortical regions, notably inferior parietal cortex. Consistently, studies comparing the neural functioning of gifted children with age-matched peers consistently report that gifted subjects display enhanced frontal cortical activation, among other neural features, notably in mathematics and music. The salience of ILPFC activations is supported by neuroanatomical studies in which the grey matter densities of high-IQ subjects in frontal regions are significantly higher than those of average IQ.

Meta-analyses of these results point to high-level ILPFC functioning as an instantiation of a more efficacious working memory (WM), a characteristic of gifted people. Specific WM processes contribute to the various high level abilities displayed by gifted people, e.g., creative analogising. Evidence for neural dissociations of some WM processes raises the question of whether targeted educational interventions could improve these various components of WM function for all children.

Introduction

The most prevailing metaphor in educational neuroscience, thanks to John Bruer (1997), is that any attempt to construct links between laboratory investigations into the internal functions of nerve cells and classroom practices is a bridge too far. Within a reductionist paradigm, bridges of partial variance can be conceptualised to categorise research into individual differences at multiple levels of focus between the classroom and the laboratory.

Individual differences in educational outcomes, as determined by various forms of assessment, but most notably, public examinations and basic skills tests (CATS, SATS, etc), are partially determined by individual differences in learning abilities, which might be measured by gain scores. Learning abilities in turn are a function of general intelligence, as measured by IQ or other g-loaded tests such as the Raven’s Progressive Matrices. At a neuropsychological level, levels of general intelligence can be predicted to a reasonable extent by levels of executive function, as determined by neuropsychological tests such as the Stroop Test. One of the key aspects of executive function is Working Memory (WM), where individual differences in WM capacity can be quantified by the n-back and similar WM-load tests. Functional neuroimaging studies, particularly using functional magnetic resonance imaging (fMRI), have shown that the functional neural correlates of WM lie in the prefrontal cortex (PFC). Consistently, structural neuroimaging studies, particularly MRI with VBM (voxel-based morphometry; a quantitative assessment of structural MRI data), have revealed individual differences in PFC structure which correlate positively with differences in WM capacity.

Of course one can replace the conceptual microscope with a conceptual telescope and reverse the direction of causality, whereby educational experiences affect learning abilities, which in turn affect measures of intelligence and so on, down to affecting neural growth in key areas of the brain such as the PFC.

A wide-span bridge experiment: from neural structure to education

In either case, such conceptual engineering suggests a wide-span bridge experiment to investigate whatever direct links might exist between neural structure and education. Subjects would be current or recent students, age range between 5 and 25 years. The neuroimaging data would consist of whole-brain MRI with VBM. The aim would be to correlate variance in neural structures with educational outcomes and see if there are any significant relationships. In order to control for widespread developmental and societal differences, N would have to be very large – between 5 to 10 thousand. Practically, this would not be an obstacle these days, where every medium to large size university has an MRI scanner, and a good portion of the general public are fascinated by the prospect of having a look at their brain. Meanwhile, we can review recent studies which bridge some of that span.

Neuroscientific bridges between intelligence and frontal cortical structure

An early bridge between intelligence and brain structure was built by Orzhekhovskaia (1996) with a post-mortem study in Soviet Russia, where IQ testing had been carried out on the whole school population over many years. Microscopic examination of neuronal anatomy revealed that the density of both neurons and gliocytes, and the thickness of cortical layers in the PFC in gifted (high IQ) people of different ages was more than double those in brains of average IQ people. That is, gifted people exhibit superior neurophysiology in frontal areas. Given that the average connectivity of a frontal neuron is of the order of 30,000 synapses, a doubled density represents a massive advantage in the information processing power of this region, perhaps analogous to comparing the computer processor in an old IBM 286 with a modern Pentium.

A more recent bridge of partial variance between frontal cortical structure and IQ was constructed by Shaw et al (2006) in their longitudinal MRI study, Intellectual ability and cortical development in children and adolescents, published inNaturein March. These researchers reported that changes in the thickness of the cerebral cortex, particularly in the prefrontal cortex, better correlated with measures of intelligence than cortical thickness itself, especially for high-IQ subjects. That is, the neuroanatomical expression of intelligence in children is dynamic. This is undoubtedly an important conclusion. Unfortunately, this study did not investigate putative sources of contributory variance, especially genetic predispositions, and social factors including education, and how they might interact.

Consistently with a dynamic interpretation, Haier et al (2004) used VBM to map which grey matter volumes correlate with IQ. They found that about 6% of the grey matter predicts IQ, and that these regions are distributed throughout the brain. Most are in frontal lobes. However, there were significant IQ regions in the parietal lobes for older subjects, and temporal lobes for younger subjects, and with different patterns for males and females.

Neuroscientific bridges between intelligence and frontal cortical function

Some years prior to the Shaw et al (2006) study, Duncan and colleagues at the University of Cambridge constructed a neuroscientific bridge of partial variance between general intelligence and frontal cortical function (Duncan et al, 2000). By comparing PET activations (later replicated with fMRI) for responses to high-g loaded IQ test items (the difficult ones) with responses to low-g loaded IQ test items (the easy ones), the over-arching research question was if the brain had dedicated areas for processing problems requiring high intelligence. The answer was that it does: the study showed bilateral inferior prefrontal cortex (PFC) activations for high-g over low-g items, for both spatial AND verbal tasks.

Moreover, a subsequent meta-analysis of 20 neuroimaging studies of various types of higher-level cognition (arithmetic, language, inferential, inductive and so on) found that the centres of activation all converged on the inferior PFC (Duncan & Owen, 2001). So much for multiple intelligences. Duncan explained this result by suggesting that neurons in the PFC are sensitively task-adaptive, enabling them to process information via afferents from multiple areas throughout the brain, while actively ignoring task-irrelevant information.

From Duncan’s account of frontal functioning, one could predict that a neural characteristic of gifted people should be highly efficacious frontal functioning. Evidence for this suggestion comes from a different approach to bridging intelligence with frontal cortical function, wherein the experimental criteria were not IQ items but fluid analogy questions (Geake & Hansen, 2005). An important point of departure was that fluid analogies have multiple plausible rather than one correct answer. Nevertheless, two areas of the left PFC were found where neural activity during fluid analogising correlated with verbal IQ, a result more significant given the restricted range of IQ from the subject pool (all above average) and the low ceiling (max IQ = 130) of the test used (National Adult Reading Test). Given that activations in these PFC areas were previously associated with working memory (WM) load tasks, it could be concluded that, at a cognitive level of description, it is WM capacity that stands in proportion to measures of IQ (Rypma et al, 1999).

Neuroscientific bridges between education and frontal cortical function

The Iowa study of mathematically precocious youth (ISMPC) has produced several neuroscientific bridges of partial variance between educational achievement and PFC functioning. The ISMPC involves the annual intake of highly math-gifted school students into a special University of Iowa summer school. Although their mean age is less than 13 years old, these subjects typically reached near ceiling on the SAT-M, mean score > 1300/1500, a similar performance to that of college math majors. This prompted Haier and Benbow (1995) to conduct a PET study of college math majors on the assumption that behavioural equivalence might be indicative of neural equivalence of the younger ISMPC cohort. Compared with college non-math majors, the PET subjects displayed greater activity in their frontal lobes, suggesting that the frontal lobes mediate high-level intelligence, a conclusion which Duncan’s group also came to several years later.

It might be recalled that Duncan’s interpretation of the role of frontal functioning was, amongst other things, to enable the processing of information from various regions of the brain. This was demonstrated in an ERP study comparing the alpha wave power of ISMPC subjects with age-matched peers on a chimeric faces task to assess relative lateralisation of spatial processing (Alexander, O'Boyle & Benbow, 1996). Here it was suggested that gifted subjects might have an unusually rapid and high-level development of inter-hemispheric interactions.

Neuroscientific bridges between learning abilities and frontal cortical function

Prodigiously talented young musicians typically show high rates of learning. In an EEG comparison study of musically talented vs. musically untalented college students listening to computer-generated pseudo-classical, pseudo-jazz and pseudo-rock music (necessary to control for preference effects), the Hausdorff dimension (complexity) of the frontal EEG was highest in the musically talented subjects while listening to the most musically complex stimuli (pseudo-classical), and lowest in musically untalented subjects while listening to pseudo-rock music, the least musically complexstimuli (Birbaumer et al, 1994). This interaction effect was consistent with the claimthat: “Complex music produces complex brain activity in complex people, simple music excites simple brain activity in simple people” which, although appearing to be a rather harsh summation of individual differences, in fact is just a succinct encapsulation of the positive feedback effect in operation in any learning situation – the range of learning outcomes in a group grows larger over time, as those with a small advantage in aptitude in the beginning learn faster through more efficacious positive feedback. Again, the site of this difference was in the frontal cortex.

Neuroscientific bridges between learning abilities and executive function

Similarly, in a study of modern-day Mozarts vs. normal children estimating degrees of musical self-coherence (autocorrelation), Geake (1996) found that it was their PFC executive functions, such as inward-directed attention, that contributed the most towards the remarkable abilities of the gifted young musicians.

More recently, Miyake et al (2000) investigated the fractionation of executive functions, with three separable but moderately correlated attributes: mental set shifting, information updating and monitoring, and inhibition of pre-potent responses. Such fractionation of executive functions in educational settings has been used to predict children’s mathematical ability (Bull & Scerif, 2001). Lower ability seems due to: poor working memory affecting relevant goal setting and task switching, and lack of inhibition of irrelevant information. This dichotomy whereby differences in intelligence involves differences in strategies has been replicated in several studies, where higher ability was due to reliance on constructive matching of information to task goal, whereas lower ability was due to limited matching of proximate information to task goal (Bunge et al, 2005; Vigneua, Caissie & Bors, 2005).

Educational neuroscience research: In the lab and in the classroom?

All of the above suggests that those students with a greater working memory capacity can enjoy better educational outcomes. It follows, then, that to improve educational outcomes for all, a worthwhile approach might be to develop pedagogic strategies to optimise WM load processing in classrooms. Despite informative psychological models of WM (e.g., Baddeley & Sala, 1998) we don’t fully understand how WM is fractionated or instantiated at a neural functional level, so to complement ongoing neuroimaging investigations into the neural basis of WM, the efficacy of classroom interventions aimed to optimise WM could be informative. Amongst some of the research foci in which teachers and cognitive neuroscientists could join forces to research pedagogies to optimise WM functioning include:

• short term memory capacity;
• accessing appropriate LTM store;
• making creative analogical connections (Geake & Dobson, 2005);
• delaying closure (Carson, Peterson & Higgins, 2003);
• evaluating relevance (Geake & Dobson, 2005).

Meanwhile, there are some immediate classroom implications from the research summarised above:

• ban disconnected approaches to teaching by

- ignoring L and R brain nonsense;

- expelling MI; and

- eradicating VAK ‘learning’ styles (Geake, 2004).

• set explicit goals for new work by

- providing answers with questions for new work (Geake, 2003).

• reward creative analogising by

- encouraging inter-subject connections (Geake & Dobson, 2005); and

- promoting novel joined-up thinking (Carlsson, Wendt & Risberg, 2000).

Summary: The neural bases of intelligence

Our brains enable us to be intelligent through a suite of interacting attributes (Gray & Thompson, 2004), including:

• the efficacy of modular functions, especially those enabling working memory in the lateral prefrontal cortex;
• the density and distribution of grey matter (neurons) and white matter (mylenation) that enable information processing;
• the cytoarchitecture (gyri and sulci) of cortex in key areas;
• the interconnectivity of functional modules.

Thus we find that there are common brain functions for all acts of intelligence (Gray et al, 2003):

• working memory involving the lateral frontal cortex activations (Duncan et al, 2000);
• long term memory involving the hippocampus + distributed cortical areas;
• decision making involving the orbitofrontal cortex;
• emotional mediation involving the limbic subcortex + the OFC;
• sequencing of symbolic representation involving the fusiform gyrus + temporal lobes (Colom et al, 2006);
• conceptual inter-relationships involving the parietal lobe (Haier et al, 2004);
• conceptual rehearsal in the cerebellum (Frangou, Chitins & Williams, 2004).

all coordinated through executive functioning involving the PFC.

References

Alexander, J. E., O'Boyle, M. W. & Benbow, C. P. (1996). Developmentally advanced EEG alpha power in gifted male and female adolescents. International Journal of Psychophysiology, 23, 25-31.

Baddeley, A. & Sala, S. D. (1998). Working memory and executive control. In A. C. Roberts, T. W. Robbins & L. Weiskrantz (Eds.) The Prefrontal Cortex: Executive and Cognitive Functions, OxfordUniversity Press, pp. 9-21.

Birbaumer, N., Lutzenberger, W., Rau, H., Braun, C. & Mayer-Kress, G. (1996). Perception of music and dimensional complexity of brain activity.International Journal of Bifurcation and Chaos, 6(2), 267-278.

Bruer, J. (1997). Education and teh brain: A bridge too far. Educational Researcher, 26(8), 4-16.

Bull, R. & Scerif, G. (2001). Executive functioning as a predictor of children’s mathematical ability: Inhibition, switching, and working memory. Developmental Neuropsychology, 19(3), 273-293.

Bunge, S. A., Wendelken, C., Badre, D. & Wagner, A. D. (2005). Analogical reasoning and prefrontal cortex: Evidence for separable retrieval and integration mechanisms. Cerebral Cortex, 15(3), 239-249.

Carlsson, I., Wendt, P. E. & Risberg, J. (2000). On the neurobiology of creativity. Differences in frontal activity between high and low creative subjects. Neuropsychologia, 38, 873-885.

Carson, S. H., Peterson, J. B. & Higgins, D. M. (2003). Decreased latent inhibition is associated with high-functioning individuals.Journal of Personality and Social Psychology, 85(3), 499-506.

Colom, R., Jung, R. E. & Haier, R. J. (2006). Distributed sites for the g-factor in intelligence. NeuroImage (in press).

Duncan, J., Seitz, R. J., Kolodny, J., Bor, D., Herzog, H., Ahmed, A., Newell, F. N. & Emslie, H. (2000). A neural basis for intelligence. Science, 289, 457-460.

Duncan, J. & Owen, A. M. (2001). Common regions of the human frontal lobe recruited by diverse cognitive demands. Trends in Neuroscience, 23, 475-483.

Frangou, S., Chitins, X., & Williams, S. C. (2004). Mapping IQ and gray matter density in healthy young people. NeuroImage, 23(3), 800-5.

Geake, J. G. (1996). Why Mozart? An information processing account of musical abilities. Research Studies in Music Education Journal, 7, 28-45.

Geake, J. G. (2003). Adapting Middle Level educational practices to current research on brain functioning. Journal of the New England League of Middle Schools, 15(2), 6-12.

Geake, J. G. (2004). How Children’s Brains Think: Not left or right but both together. Education 3-13, 32(3), 65-72.

Geake, J. G. & Hansen, P. (2005). Neural correlates of intelligence as revealed by fMRI of fluid analogies,NeuroImage, 26(2),555-564.

Geake, J. G. & Dodson, C. S. (2005). A neuro-psychological model of the creative intelligence of gifted children. Gifted & Talented International, 20(1), 4-16.

Gray, J. R., Chabris, C. F. & Braver, T. S. (2003). Neural mechanisms of general fluid intelligence. Nature Neuroscience, 6, 316-322.

Gray, J. R. & P. M. Thompson (2004). Neurobiology of intelligence: Science and ethics. Nature Neuroscience, 5, 471-482.

Haier, R. J. & Benbow, C. P. (1995). Sex differences and lateralisation in temporal lobe glucose metabolism during mathematical reasoning. Developmental Neuropsychology, 11(4), 405-414.

Haier, R. J., Jung, R. E., Yeo, R. A., Head, K. & Alkire, M. T. (2004). Structural brain variation and general intelligence. NeuroImage, 23(1), 425-433.