Gerald Coles. Handbook of Early Childhood Literacy. Editor: Nigel Hall, Joanne Larson, Jackie Marsh. Sage Publication. 2003.
What is the relationship between brain functioning, genetics, and learning to read—and should reading educators care? Do reading educators really need to know about what’s going on in the brain and in DNA when devising effective reading instruction and helping students learn to read? Knowledge about these connections might be interesting, but isn’t it enough to focus on the children’s visible reading activity and outcomes when making decisions about instruction and its implementation? Or, addressing a more immediate question, why is this chapter in a handbook on literacy?
In recent years claims have insisted that if reading teachers are to be effective, they must know about the brain and genetics. As the editor of the newsletter of The Center for Education Reform stated, ‘Teachers are rarely if ever taught about how reading gets accommodated in the brain. And of course without that knowledge, we’ll never be a nation of readers, and the nearly 40 percent of children who are mainly disadvantaged will never reverse that label’ (2001: 1).
The need to know about the brain and genetics is also asserted on the website of the International Dyslexia Association (2002). Answering the question, ‘How do people get dyslexia?,’ the Association states, ‘The causes for dyslexia are neurobiological and genetic. Research shows that individuals inherit the genetic links for dyslexia. One of your grandparents, parents, aunts or uncles is dyslexic, and chances are that one or more of your children will be dyslexic’
A press release from the Office of Public Information of the University of Florida summarized for the nation’s media the findings of campus researchers: ‘Brain structure may play a role in children’s ability to learn to read’ (Ramey, 1998). And the Child Development Institute (1997) told readers of its ‘award winning’ website that ‘10 Years of Brain Imaging Research Shows The Brain Reads Sound By Sound.’ More specifically, it stated the instructional implications of the research: ‘The new brain research shows why intensive phonics is also the best way for everyone to learn to read.’ Buttressing these kinds of claims is a stream of media headlines. The New York Times (1999) informed readers that ‘Scientists Find the First Gene for Dyslexia.’ The Times reported, ‘Changes in Brain a Clue to Dyslexia.’ An implication of these claims was summarized in the Education Week headline, ‘Demands Grow to Link Neuroscience with Education’ (Jacobson, 2000).
These ‘demands’ have, in fact, already been transformed into reading policy that has mandated particular kinds of classroom instruction and excluded others (Coles, 2000; 2003). Consequently, reading educators, to the extent that they understand this research, would be better able to make informed judgements about the research linked to these demands and policy—which should be supported or opposed. The purpose of this chapter is to contribute to that understanding.
Brain Activity and Beginning Reading Skills
A discussion of the research on brain activity, genetics, and learning to read must begin with the observation that most of this work has been guided by a priori assumptions about how children should be taught to read. Foremost among these has been the judgement that beginning readers will learn best if they first master the basic skill of making sound-symbol relationships, then continue in a ‘building block’ sequence to erect skill upon skill, eventually gaining competence in the skill of comprehending text (National Institute for Literacy, 2001). The opposite has also been assumed: if beginning readers do not first master basic skills, they will be at risk of becoming reading disabled. Phonological awareness (distinguishing and manipulating sounds in words) is considered to be the first building block within the sequence, as well as the core deficit in reading disabilities—an assumption, according to reading researchers such as Sally Shaywitz, consistent ‘with what neuroscientists know about brain organization and function’ (1996: 99).
Much recent research on brain malfunctioning and reading acquisition aimed at demonstrating this primary belief has been funded by the Child Development and Behavior Branch of the National Institutes of Child Health and Human Development (NICHD), whose viewpoint was expressed in a paper reporting an NICHD-funded study co-authored by Reid Lyon, chief of the branch. The relationship between ‘brain activation patterns’ and tasks of ‘sounding out words,’ Lyon and his co-authors concluded, ‘provided neurobiologic evidence of an underlying disruption in the neural systems for reading in children with dyslexia and indicate that it is evident at a young age’ (Shaywitz et al., 2002: 101). Findings like these have, according to Lyon, contributed to his anticipating a time in the foreseeable future when scientists will ‘confidently’ be able to ‘design a classroom curriculum based completely on neuroscience’ (Hotz, 1998: 1).
The following section reviews representative studies associated with this interpretation of brain functioning and reading success. Although much of the research reviewed in this chapter used adults rather than children as subjects, the researchers of and commentators on this work have assumed that the findings are directly applicable to classroom reading instruction. In order to give the reader an adequate sense of the quality of the research and because of space limitations, I will offer in-depth discussions of key, representative studies, rather than brief summaries of many studies.
A ‘Brain Glitch’
Studying 29 ‘dyslexic’ readers (14 men and 15 women, ages 16–54 years), in an NICHD-supported investigation, Sally Shaywitz and her colleagues (1998) used functional magnetic resonance imagery (MRI), a technology that provides information about the structure and function of the brain, while the subjects engaged in a sequence of reading and reading-related tasks. Whether or not the adults actually met criteria for being ‘dyslexic’ cannot be determined because, except for their IQ score range, we are told nothing else about them. For the sake of the primary focus of my appraisal, however, I will accept this categorization.
The study used a sequence of five tasks, beginning with one that required no ability with written language (asking subjects to decide whether or not lines matched: V versus V).
The second task asked subjects to match patterns of upper-and lower-case letters (bbBb versus bbBb). This required letter, but not phonological, knowledge.
The third task asked the subjects if letters rhymed (‘Do T and V rhyme?’). This added ‘a phonological processing demand’ requiring knowing and comparing the sounds of the letters.
The fourth task asked the subjects if non-words rhymed (‘Do leat and bete rhyme?’). This task required ‘analysis of more complex structures.’
The fifth and last task required that the subjects know complex sound-symbol relationships and the meaning of words (‘Are corn and rice in the same category?’), requiring both phonological and semantic knowledge and processing.
Clearly, this study is not simply about the brain and reading, but is linked to skills-emphasis theory and pedagogy, of which phonological awareness is the centrepiece. That is, using this viewpoint, the researchers contrived a study around their conception of what is causal in beginning reading, namely, phonological awareness. In addition, to assume that this study enables one to draw conclusions about ‘reading’ begs the question because the attenuated reading-related tasks that were used, while providing activities for potentially useful functional MRI information about language, cannot be thought to represent ‘reading.’ At best, they pertain to delimited skills. Alternative definitions of ‘reading’ certainly would have led to the creation of very different ‘reading’ tasks—perhaps even one in which the subjects actually ‘read’ sentences! (For other definitions see, for example, Ruddell et al., 1994.) The study was also infused with an assumption that a neuropsychological deficit (or, as Shaywitz has described it, ‘a brain glitch’) can cause reading problems in otherwise normal children (Kolata, 1998). When we look at the results, we see how the researchers’ presumptions influenced interpretation of the data.
Shaywitz and her colleagues found group differences in brain patterns while the subjects were engaged in the various tasks. The good readers showed ‘a systematic increase in activation’ in the brain areas studied, when going from the second (matching letter patterns) to the fourth task (determining if non-words rhyme). That is, there was an increase in brain activation as the tasks increased demands for applying phonological awareness abilities. In contrast, the dyslexics showed a fairly steady level of brain activation, rather than an increase, in response to these tasks. Generally speaking, the brain activation for these tasks was higher for the good readers than for the dyslexics, although one area of the brain showed the reverse pattern.
Examining the activation in the brain hemispheres, the researchers found that for good readers the activation was greater in the left, and for dyslexics it was greater in the right. This pattern held across all tasks.
Shaywitz and her colleagues concluded that ‘for dyslexic readers, these brain activation patterns provide evidence of an imperfectly functioning system for segmenting words into their phonological constituents.’ This malfunctioning was ‘evident’ to the researchers when they asked the dyslexics to respond to increasing demands on phonological analysis. The dyslexic readers demonstrated ‘a functional disruption’ in the rear area of the brain in which visual and sound identification and associations are made during reading. These findings, according to the researchers, added ‘neurological support’ to evidence obtained through studies at the behavioural and cognitive levels that pointed ‘to the critical role of phonological analysis and its impairment in dyslexia’ (1998: 2640).
The problem with these interpretations is that functional MRI data themselves do not carry an imprint of their meaning, that is, they do not point to a cause of the specific brain activation. The specific activation linked to responses generated by the five tasks tells us nothing about the processes that produced the activation. The specific activation facts would actually make possible various reasonable explanations, and given the limited data in this study, they would all have equal legitimacy.
For example, the study disregarded problem solving approaches, learning experiences, personal meanings, emotions, motivation, and confidence, to name but a few potential influences that could have affected the group outcomes. Studies have shown that during tasks of this kind, altering any one of these background and processing factors could result in altered patterns of brain functioning (Coles, 1987). Furthermore, since there were ability differences between the groups, why would anyone assume that the brain activity for the two groups would be the same when doing these tasks?
Another NICHD-funded study, led by Bennett Shaywitz, with Sally Shaywitz and Reid Lyon among the co-authors, also used the functional MRI to identify brain areas that were active when good and poor readers did non-word and real word tasks. Normal readers showed more activation in the back of their brain, while the dyslexic group showed more activation in the front and side regions.
Continuing to be guided by the assumption that ‘converging evidence indicates a functional disruption in the neural systems for reading in adults with dyslexia’ (Shaywitz et al., 2002: 101), an NIH-NICHD (2002) press release stated, ‘Children who are poor readers appear to have a disruption in the part of their brain involved in reading phonetically.’
The study does not provide the means for supporting these conclusions, however, because it is only one more investigation containing methodological confusion of correlation and causation. That is, as was true for the NICHD-supported study just discussed, the fact that there is a difference in the brain activation between good and poor readers does not mean that the brain activation is the cause of the respective reading abilities. Rather, we know only that the activation is correlated with reading ability. Consider the following experiment. If two groups of normal people were asked to read a Czechoslovakian text, and if only one group could read Czechoslovakian, who would expect the brain activation of the two groups to be the same? And who would think that differences in brain activity revealed dysfunctions (Czechlexia), not differences? This failure to distinguish between correlation and causation fosters the single-minded interpretation of the data represented by this and similar studies. We do know from the study that the brain activity of good and poor readers differs when they do reading tasks, but that is all we know. Certainly nothing can be concluded about the cause of the reading problems or the best way to teach reading.
Learning and Brain Changes
A study that sheds light on the question of causation is one Leonide Goldstein and I did on differences in brain hemisphere activation in adult beginning readers as they were learning to read (Coles and Goldstein, 1985). We found that these adults did, indeed, have greater right hemisphere activation initially, when they were poor readers or non-readers, but as their reading improved the activation of their hemispheres also changed toward the greater left activation pattern common to good readers. We interpreted these data as evidence that new knowledge and competencies were linked to concomitant changes in brain structure and functioning. More generally, these brain changes associated with written language acquisition were representative of the kinds of changes that would occur through all kinds of learning. There was nothing in the data to suggest that these beginning readers started learning to read with anything but normal brains that were initially configured as they were because the students had not learned to read; no data suggested that the educational intervention somehow repaired or circumvented dysfunctional brain areas.
Unfortunately, in later studies using educational intervention that produced similar findings, the researchers concluded that the brain changes were evidence that a neurological dysfunction had caused the reading problems. Under the subtitle ‘New evidence on neurobiological causes,’ Reid Lyon and Jack Fletcher (2001) described a study in which 60 hours of intensive educational intervention produced brain changes in the left brain hemisphere. Before intervention, MRI analysis revealed unactivated portions of the left hemisphere—‘the standard activity pattern of children with reading disabilities’ noted Lyon and Fletcher. After intervention, they said, ‘brain activation patterns shifted to the normative profile seen in nonimpaired readers.’ The authors concluded that although environmental factors can influence brain organization and activity, the results were part of ‘a sizable body of evidence’ that indicates ‘poor readers exhibit disruption primarily, but not exclusively, in the neural circuitry of the brain’s left hemisphere, the part that serves language.’ Why they drew these conclusions is not apparent because they offered no evidence to demonstrate that the minimal activity in the circuitry that serves language was caused by a ‘disruption’ rather than by merely the absence of knowledge and skills that the educational intervention later provided.
In another educational intervention study with a comparable outcome, a similar interpretation was offered: the brain changes produced through educational intervention demonstrated faulty brain wiring in poor readers and the possibility of rewiring. ‘We now know that people with reading problems are using the wrong hardware in their brains, and if we can get them to switch to the right hardware, we might be able to improve their reading’ said NICHD-supported principal investigator Andrew Papanicolaou (Suriano, 2002). Again, there was no evidence that these readers began learning with the ‘wrong hardware.’ Of course it is true that the brain areas were rewired, but there was no evidence that this rewiring was different in any way from the rewiring that goes on throughout our learning lives.
Regrettably, although the functional MRI is a potentially valuable technology for literacy studies, it is rendered worthless when used with flawed theories, methods, and data interpretations. In making this criticism, I do not want to lose sight of the potential value this technology can provide for understanding the complex phenomenon of learning to read and for addressing questions such as, ‘Will alternative teaching approaches configure brain activity in alternative ways?’ Clearly, the reading field could benefit from objective studies that employ good teaching and use a developmental method that appraises brain activity as reading ability evolves in the reading acquisition process.
Genetics and Reading
Explanations of genetic causes of reading problems parallel much of the brain research. In a House of Representatives Committee on Education and the Workforce hearing, for example, Reid Lyon (1997) stated, ‘our NICHD studies have taught us that the phonological differences we see in good and poor readers have a genetic basis.’ In another overview of ‘major findings’ from NICHD research programmes, Lyon declared, ‘There is strong evidence for a genetic basis for reading disabilities, with deficits in phonological awareness reflecting the greatest degree of heritability’ (1996: 65).
One example of this ‘strong evidence’ was an NICHD-financed study published in the prestigious journal Science (Cardon et al., 1994). Lon Cardon and his colleagues reported locating a gene for reading disability on chromosome 6, a finding that appeared especially compelling because the gene was shared to a much greater extent—an extraordinary extent—by subjects with ‘extreme deficits in reading performance’ than by a group of poor readers with serious but relatively fewer problems. Similarly, the families of the subjects with ‘extreme deficits’ were found to have the gene more frequently than the families of poor readers with fewer problems. In other words, the worse the reading deficit, the greater the evidence of a relationship to a gene on chromosome 6.
The results were even more striking when the researchers grouped the twins who had ‘more extreme deficits’ in reading and found that their genetic sharing on a portion of chromosome 6 soared to an astronomical 0.00001! In contrast, the same analysis for siblings with ‘more extreme deficits in reading’ found a correlation just short of significance (0.066). In short, Cardon and his colleagues presented what they regarded as considerable evidence suggesting a ‘broad heritability’ of reading disability.
Although these findings certainly seemed to provide exceptionally strong evidence, seven months after the article appeared, the researchers wrote a letter to Science offering a ‘correction’ of the previously reported results (Cardon et al., 1995). ‘Reanalyses of the twin data revealed that four identical twin pairs had been inadvertently included in the [fraternal pair] sample.’ This, it turned out, accounted for the highly statistically significant correlations. How this mistake occurred was not explained, but one can reasonably wonder how researchers with precision enough for mapping chromosomes and for performing intricate statistical analyses could have missed recognizing the inclusion of four pairs of identical twins.
When the four pairs of identical twins were removed from the fraternal twin group, the researchers’ correction letter explained, the extraordinarily high statistical results reported in the original paper essentially disappeared. Overall, the ‘correction’ letter repudiated the researchers’ original conclusions: ‘In order to confirm evidence for a possible [linkage] for reading disability on chromosome 6,’ the researchers conceded, ‘analyses of data from additional twin pairs will be required’ (1995: 1553). A later study by these researchers, using over 100 pairs of twins categorized by degrees of severity of reading problems, failed to salvage the initial claims (DeFries et al., 1997). Remarkably, current reviews of the genetics of reading disability continue to cite the findings of the original study and completely omit the reanalysis in the ‘correction’ letter or later failures by the same group of researchers (Olson and Gayan, 2001).
The decline in statistical significance that resulted from the removal of only four of 50 twin pairs in the original study demonstrates how the weight of only a small number of heavily loaded scores can dramatically shift correlations, and cautions against placing strong confidence in a single group of subjects in this kind of genetics research. This is no small matter because the evaporation of the evidence was exactly what happened in the earlier research on chromosome 15. At first there were ‘breakthrough’ findings linking the gene to ‘dyslexia,’ but added subjects, researchers later found, reduced the level of statistical significance. Furthermore, one investigator, using a different group of subjects, failed to duplicate the original research. Eventually, the researchers of the original ‘breakthrough’ study on this chromosome repudiated their own initial ‘findings’ (Coles, 1987; 1998).
Reading Disability Genes in Families
Another study that looked for a reading disability gene on chromosome 6 examined six families of adults who had had serious reading problems as children (Grigorenko et al., 1997). The families were divided according to five measures or phenotypes: (1) phonological awareness, (2) phonological decoding, (3) single-word reading, (4) rapid naming automaticity, and (5) a discrepancy score between IQ and reading score.
The investigators reported that an association with chromosome 6 varied for each phenotype, with phonological awareness having the highest association and single-word reading the least. Presumably, this was a striking piece of evidence for the phonological awareness explanation of reading success and failure.
To assess these findings we need to look closely at the results by themselves and in relationship to previous studies on genetics and reading, particularly the ones just discussed. Key outcomes of such an assessment are these:
- The statistical significance for a portion of chromosome 6 found in the first twin group study reported by Cardon et al. (1994) was not found in this research.
- For another portion of chromosome 6, this study found a strong correlation for the phonological awareness phenotype, but the correlation was largely due to the strong contribution by one family, which made up for the lack of any such correlation for phonological awareness in two families at this or any other portion of chromosome 6. If there is a powerful association between reading and chromosome 6, why did one-third of the six families show no consequential association at all?
- The investigators of this study claimed that their results for ‘chromosome 6 are consistent with the results’ of the previous research (Cardon et al., 1994) I have discussed. There is, however, little overlap of significant findings for any of the regions explored on chromosome 6. In fact, in contrast to the researchers’ assertions, comparing the two results demonstrates that this is not a replication study!
Other questions may be raised about this family study with respect to the now familiar theory of reading disabilities that underpins it and to the actual reading abilities of the subjects. The investigators acknowledged that because ‘a number of individuals with a phonological awareness deficit—including all affected cases in the largest family members)—exhibit normal single-word scores, it follows that this hypothetical gene on chromosome 6 is not itself sufficient for the full syndrome of dyslexia’ (Grigorenko et al., 1997: 35). In other words, although the individuals scored poorly on tests of phonological awareness, their scores on word reading tests were those of normal readers. This discrepancy between purported cause and effect can be stated even more strongly: what is the actual impact of the supposedly genetically generated phonological awareness deficit if these adults, who had been classified as ‘reading disabled’ in childhood, did not in adulthood have single-word reading problems? The word recognition score is, of course, inadequate for a satisfactory picture of the reading abilities of these ‘affected cases,’ but the researchers provided no other reading profile information. Presumably, the word reading results indicated that the subjects could read sufficiently well, therefore making unapparent a clear link between phonological deficits and actual reading.
The question of whether a phonological awareness gene contributes to the creation of reading problems is further complicated by the fact, as the researchers themselves acknowledge, that it is phonological decoding skills, not phonological awareness, that has been identified as a ‘central, disproportionate deficit in dyslexia’ (1997: 29). What, therefore, do the phenotype results in this research mean, if they do not jibe with the decisive skills the researchers believed were associated with reading achievement?
When the researchers added two families to their subject pool (Grigorenko et al., 2000), they only duplicated the previous results and inconsistencies contained in the report of the six families. The phonological phenotype continued not to play the causal role that the researchers initially expected. No other phenotype stood out as having a determinant connection to poor reading. Furthermore, the outcome of this kind of investigation, the researchers acknowledged, was dependent upon the very analytical method of identifying an area of a gene and, of course, upon how frequently a ‘phenotype of interest’ appears ‘in a given sample’—an explanation apparently meant to explain why the phenotype the researchers expected to find was not found (2000: 721).
Findings in additional studies that, on the surface, appeared to support claims about a dyslexic gene on chromosome 6, were questionable because of inconsistencies over whether the word-identification phenotype was associated with chromosome 6 or 15; inconsistencies over the association between word identification and phonological phenotypes with reading problems; and contradictory conclusions about effect sizes and purported gene actions (Fisher et al., 1999; Gayan et al., 1999). In addition, the results of these studies were not replicated by researchers who reported finding an ‘absence of linkage of phonological coding dyslexia to chromosome 6’ (Field and Kaplan, 1998). Finally, all of these studies suffer from a complete lack of exploration of (1) the precise nature of the reading problems that were conveniently clumped together as ‘dyslexia’ in order to categorize the subjects for the studies, and (2) alternative experiential explanations, such as family and school influences, that could explain the subjects’ reading problems.
As is true for the brain function studies, sophisticated genetics techniques cannot compensate for the flawed theories, methods, and data interpretations. Are there genes that determine the effectiveness of portions of the brain that process sound-symbol relationships and are root causes of reading success or failure? To answer this question, or to answer any question about genes and reading, and brain functioning and reading, research must be based on accurate models of brain functioning. In the next sections, I will discuss these models.
How the Brain Works: Modules?
An understanding of the relationship between brain functioning, genes and reading acquisition requires examining the extent to which the premises of the research accord with current findings in neuroscience. A chief premise holds that the brain has specific modules for specialized operations that work in sequence and in coordination with other modules in learning written language. As we have seen in the research discussed above, one or more modules that process basic sound-symbol skills are believed to be fundamental in the hierarchy and organization of modular brain activities that underpin learning to read. That is, unless these fundamental modules first process written sounds and symbols, other brain modules involved in learning to read will not be able to function adequately.
Explanations of reading acquisition based on a modular model rely heavily on an assumption that the fundamental modules—those that process written sounds and symbols—must first be activated and stocked because a beginning reader has limited working memory that restricts the amount of attention that can be allocated to various aspects of written language. If, for instance, a beginning reader has to give equal attention to sound-symbol correspondence of words and to the meaning of what he or she is reading, working memory would be overloaded. To avoid this, the focus of beginning reading instruction must be consistently narrow, aiming the student’s learning on sound-symbol skills (Adams, 2001).
However, the basic assumptions of brain modules and working memory that underpin this view of the mental requirements in beginning reading are rejected by many neuroscience researchers. Merlin Donald, for example, a psychologist who has written extensively on human consciousness, argues that conclusions about limited working memory come primarily from laboratory studies that have used a brief time frame methodology in which ‘short-term memory, visual imagery, perceptual illusions and the allocation of attention, must be crammed’ (2001: 47). Because of ‘this built-in, albeit unintended, bias, such experiments look only at the lower limits of conscious experience’ (2001: 47). In real-life activities, Donald stresses, ‘the width and depth of working memory in such situations are much larger than those suggested by traditional laboratory techniques’ (2001: 50). Consequently, these laboratory models ‘have very limited real-world generality’ (2001: 52). He argues that the memory system is composed of both short-term and ‘intermediate-term awarenesss’ that constantly update working memory. Hence, there is much more that working memory can address, incorporate, and apply in these more elaborate mechanisms. If this is so, then even if the modular view of reading acquisition were correct, the beginning reader would have no cognitive need to focus almost exclusively on modules assumed to be foundational in the sequenced organization of modular activity.
With respect to the theory that there are modules that do ‘specialized operation,’ such as deciphering language (2001: 3), Donald argues that the columnar unit within the ‘now-mythical Broca’s region, once believed to be the language region of the dominant hemisphere’ (2001: 101) is only one among several hundred thousand in the brain that are interconnected, ‘woven into various brain-wide networks by millions of long communication fibers’ (2001: 101). In other words, learning written language—as learning all else—involves an extensive network (a polyphony) of brain areas activated and communicating simultaneously and interactively. It is not a predetermined network.
Donald notes that ‘many researchers trained in the sixties,’ including himself, ‘sought to discover the neurological magic module that might explain human language and symbolic thought,’ a tradition that extends ‘back to the early heroes of the Great Module Hunt’ such as Wernicke and Broca. The results of that search, Donald concludes, ‘were largely negative’ because there is no modular ‘table of elements’ that make humans more unique than chimpanzees in some ‘modular redesign of the nervous system’ (2001: 111).
Furthermore, Donald criticizes the ‘isolated mind’ bias in cognitive science that treats the ‘cognitive system’ as though it were a ‘self-contained entity or monad’—an isolated organ with the modules in place to acquire written language (2001: 150). Instead, although the brain has fundamental mechanisms for beginning to learn written language, it is learning and experience that shape the brain’s circuits and how they are used in learning to read. The brain ‘has no fixed pattern of connectivity to start.’ There are basic mechanisms that are innate, but the brain’s ‘connectivity pattern is set by experience’ with ‘countless interconnection points, or synapses, which connect neurons to one another in various patterns’ (2001: 103).
The view of a ‘connectivity pattern’ that emerges and is activated as children learn to read contrasts with the model of step-by-step progression from module to module. If the former is an accurate model of brain organization and functioning, it suggests that the connectivity pattern should be the focus of research because only by looking at the overall pattern can researchers begin to determine the functioning and interrelationships of any part, and the causal, consequential, or interactive function of that part within the entire pattern.
From the perspective of a connectivity pattern model, not only do the brain areas involved in grasping the sound-symbol correspondence not have to be primed first before other areas of the pattern can become effectively operable, but the functioning of these areas depends on connections within the entire pattern. And because the pattern is not innately fixed, if instruction were to stimulate certain areas more than others, a particular connectivity pattern would emerge. That specific pattern, however, might not necessarily be the sole one required for reading success or the one superior over other connectivity patterns. Moreover, if conceptions of limited working memory are incorrect, a more complex connectivity pattern could be created through richer written language learning.
Linguist Philip Lieberman (2000) too has criticized modular explanations, calling them ‘neophrenological theories,’ that is, theories that ‘map complex behaviors to localized regions of the brain, on the assumption that a particular part of the brain regulates an aspect of behavior’ (2000: 3). In these theories, the functional organization of the brain is run by ‘a set of petty bureaucrats each of which controls a behavior’ (2000: 2). Like Donald, he proposes that:
converging behavioral and neurobiological data indicate that human language is regulated by a distributed network that includes subcortical structures, the traditional cortical ‘language’ areas (Broca’s and Wernicke’s areas), and regions of the neocortex associated with ‘nonlinguistic’ aspects of cognition. (2000: 2)
Complex processes ‘are regulated by neural networks formed by circuits linking populations of neurons in neuroanatomical structures that may be distributed throughout the brain,’ not by a hierarchical system (2000: 4). Lieberman stresses:
although specific operations may be performed in particular parts of the brain, these operations must be integrated into a network that regulates an observable aspect of behavior. And so, a particular aspect of behavior usually involves activity in neuroanatomical structures distributed throughout the brain. (2000: 4, emphasis in original)
Such a view of functioning that is distributed, not localized, undercuts a fundamental premise upon which the research on brain activity and reading acquisition is grounded. Pertinent to explanations of genetic bases of reading and reading disabilities is Lieberman’s (1998) judgement that there is no genetic ‘blueprint’ for learning functional language or aspects of it, such as phonological awareness. The ‘details of syntax, speech, and the words of the languages that a person knows’ are not learned by specific genes for these details, but ‘appear to be learned by means of the associative processes that enable us to learn other complex aspects of behavior’ (1998: 132).
Putting this another way, there are neural systems that include language-related portions of the brain, but learning written language is not determined solely by the functioning of these specific parts or by the genes for these specific parts. Learning language, spoken and written, is based on the inferential aspects of our thinking that are part of a larger neural network that includes functions and systems used for other kinds of thinking. No one gene determines phonological awareness or word recognition because there is not that kind of specificity for the details of language. It is a larger thinking system that orchestrates language learning.
‘Cognition’ and Reading: The Absence of Emotions
The assumption that ‘cognition’ actually describes the brain processes associated with reading also needs to be examined. Skills-emphasis/‘building-block’ brain research has assumed that ‘cognition’—that is, the process of images, concepts, and mental operations—is an independent reality, not a construct, and in doing so has ignored ever-growing evidence suggesting that thinking is an inseparable interaction of both cognition and emotion (feelings, desires, enthusiasms, antipathies etc.), not cognition alone.
Neurologist Antonio Damasio (1994), for example, rejects the traditional distinction between cognition, thought to be neocortical, and emotions, thought to be subcortical. There are no ‘higher’ and ‘lower’ brain centres, he argues: the neocortex—the ‘high level’ part of the brain—does not handle reason, while the subcortex—the ‘low level’ part of the brain—handles emotions (1994: xiii). Rather, he maintains, the neural substrates for cognitive responses are associated with neural substrates for emotions: both so-called ‘high’ and ‘low’ levels are integrated in thinking processes. His work supports conceptualizing cognition and emotions, to use the metaphor of paediatric researcher Michael Lewis, as a ‘continuous and interwoven fugue’ (Lewis et al., 1984: 264) that is operative in every facet of learning to read.
The work of neural scientist Joseph E. LeDoux (1996), which has identified brain pathways that carry sensory signals to sites of emotion and of cognition, also reveals the error of focusing solely on ‘cognition’ in studies on brain activity and reading acquisition. LeDoux has found that the thalamus, an area that relays sensory information, conveys sensory stimuli to the amygdala (a site of basic emotional memory) and to the cortex, where ‘cognition’ occurs. From the cortex the stimuli go on to the hippocampus, a site involved in memory and linked to the amygdala. This interconnection of pathways means that an emotional response can, in terms of pathway activity, precede a cognitive perception and response, and that emotions and cognition are integrated and interactive. Consequently, reading researchers who focus only on cognition when studying the brain are ignoring the areas of networks whose emotional activation are part of ‘cognition.’ Faulty or insufficient activity identified in a portion of the brain of someone doing a reading task might be a consequence of an emotional response, in that emotional memories can exert a powerful influence on ‘thought processes.’ What these connections are remains for future research to determine, but there is no question that research on brain activity and reading that fails to account for the fugue of cognition and emotion is severely insufficient research.
Instruction and Thinking
Although most reading research does not delve into brain activity, some of it can offer insights into how thinking is organized in relation to learning to read, especially if the investigation explores the question, ‘Does the particular way in which reading is defined and taught shape the kind of readers students become?’ If different reading approaches result in different kinds of thinking among students who become competent readers, according to conventional definitions of reading success, this outcome would contradict the modular view of reading acquisition. That is, one could conclude that thinking related to reading acquisition is organized through learning, but does not have to be organized in one way only in order for someone to learn to read. On the other hand, if youngsters successfully learn to read with different approaches, but their thinking is organized the same way, that would suggest that the modular view is correct. A study by reading researcher Penny Freppon (1991) provides some insights into these issues.
Freppon compared reading outcomes for first-grade children taught with either skills-based or literature-based/whole-language instruction and found that the test results were similar for both groups. But her study went beyond these outcomes by looking closely at the way in which the children processed written language while reading and at the conceptions of reading they held. She found that even though the literature-based/whole-language instruction did not explicitly teach skills, the children in both forms of instruction ‘were knowledgeable about the importance of decoding’ and ‘successfully used’ it in reading (1991: 159). There was no evidence that whole-language instruction diminished children’s sense of the value of this aspect of reading.
Freppon’s finding accords with Lieberman’s view: as children learn to read they problem solve, and by doing so attain increased ability to understand causal and reciprocal relationships. As part of this problem solving, they grasp that a key problem to be solved in learning to read is the mastery of connections between graphemes and phonemes.
The similar group knowledge of decoding did not mean that each group used the strategy the same way—that there is an invariable mental organization necessary for learning to read. The skills-emphasis group used decoding as a primary strategy, while the whole-language group used it to a lesser degree because that group employed a greater variety of strategies, such as rereading, using context, and skipping words. An unexpected finding was that even though the whole-language children ‘attempted to sound out words less often’ when they did attempt it, they ‘achieved a higher success rate of correctly sounding out words.’ Their rate was 53% compared to 32% for the skills-emphasis children (1991: 139).
These findings suggest that a particular reading approach is likely to produce particular kinds of thought processes. Presumably, the neural networks created in learning to read included the necessary activation of a subnetwork facilitating learning of sound-symbol skills, but this subnetwork, as part of the larger one, was not any more foundational than other subnetworks that were used for learning to read. The implicit definition of ‘reading’ in whole-language instruction made decoding ‘a’ key, not ‘the’ key, in orchestrating the thought processes. For the skills group, the grapheme-phoneme task loomed larger both as a strategy and as the meaning of ‘reading’ and was more ‘the’ key than ‘a’ key. In the skills classroom, reading for meaning was included but it was ‘incidental’ to word skills instruction (1991: 144).
In the literature-based instruction, decoding skills were focused on as needed but more of the students’ attention was drawn to meaning, with the teacher encouraging the children to think about what was going on in the story. Interviews with the children found that the literature-based group expressed greater ‘understandings of the use of multiple strategies in reading’ and ‘associated reading with language’ (whether something makes sense or sounds like a sentence), whereas the skills-emphasis group ‘expressed understanding of sounding out as a primary reading strategy’ and ‘associated reading with getting words correct’ (1991: 152). Almost all of the children in the literature group ‘said that understanding the story or both understanding and getting words right is more important in reading.’ In contrast, only half the children in the skills group chose these explanations; nearly all of the remaining half chose ‘getting words right as most important’ (1991: 153).
Asked about the ‘characteristics of good readers,’ the skills group emphasized ‘knowing and learning words and sounding out words.’ In contrast, the literature-based group discussed characteristics such as ‘reading a lot’ and ‘understanding the story.’ The skills group included ‘paying attention to the teacher’ and ‘knowing their place in the book,’ characteristics that were not mentioned by the literature group (1991: 152).
These findings also put in serious doubt the assumption that children have a limited working memory requiring that they focus on only one kind of beginning reading strategy. It would appear that children can orchestrate successfully several strategies in working memory while not diminishing their ability to identify words and comprehend stories.
The Freppon study suggests an extremely important conclusion: instruction itself contributes to the construction of the thinking process to a considerable degree, and different instruction produces different thinking processes. And it also suggests that the assumptions about mental organization and activity that underpin most of the research on brain activity and learning are erroneous.
Philip Lieberman offers a caveat worth emphasizing in appraising contemporary research and conclusions about brain activity, genetics, and reading acquisition: ‘We must remember that we stand on the threshold of an understanding of how brains really work. The greatest danger perhaps rests in making claims that are not supported by data or that inherently cannot be subjected to rigorous tests’ (1998: 132). Unfortunately, not only have reading researchers who have undertaken this work seldom been guided by such a caveat, but they have tended to misconstrue the data and draw conclusions that serve to justify unwarranted beliefs and instructional policy that have driven the research in the first place.
The review in this chapter also suggests that there are fundamental theoretical problems in the assumptions about modular brain organization and ‘cognition’ that guide the research. Given these problems, the research does not allow us to conclude that a modular organization of the brain requires one form of instruction, that a dysfunction in a skills module creates reading disability, or that cognition, independently of affect, can explain reading acquisition. Genetics research thus far adds nothing to our understanding of reading outcomes both because of the faulty data of the studies and the highly questionable premise that there are genes that can cause dysfunctional modules.
Deficiencies like these do not mean that brain research cannot contribute to our understanding of reading acquisition. Continued understanding of whether brain functioning is organized as neural networks or as sequential modules can provide a sounder basis for appraising the logic of instructional approaches and for devising sound instruction. More specifically, such knowledge can help evaluate arguments that give special weight to particular aspects of beginning reading instruction. Understanding how emotions are involved in neural networks can help teachers appraise the degree and contributions of affect in classroom literacy instruction. With greater understanding of the brain we can also better determine the interaction between children’s personal knowledge and their literacy learning, and thereby better grasp and devise an interplay between the two. Finally, greater understanding of the relationship between brain functioning and reading acquisition can help promote ecological approaches that are grounded in an understanding of the unified interrelationships of brain, active child, and learning environment, and that eschew instructional views that are ‘brain based’ or that conceive of the brain as an extraneous ‘black box.’