«Supporting Online Material for How Learning to Read Changes the Cortical Networks for Vision ...»
Supporting Online Material for
How Learning to Read Changes the Cortical Networks for Vision and
Stanislas Dehaene,* Felipe Pegado, Lucia W. Braga, Paulo Ventura, Gilberto Nunes
Filho, Antoinette Jobert, Ghislaine Dehaene-Lambertz, Régine Kolinsky, José Morais,
*To whom correspondence should be addressed. E-mail: email@example.com
Published 11 November 2010 on Science Express
DOI: 10.1126/science.1194140 This PDF file includes Materials and Methods SOM Text Figs. S1 to S8 Tables S1 to S3 References Supplementary materials to “How learning to read changes the cortical networks for vision and language”
I. Materials and methods
2) Definition of groups and participant selection
3) Behavioural tests
4) Behavioural profiles of the participants
5) fMRI data acquisition
6) fMRI analyses
II. Supplementary results
1) Behavioral results during fMRI
2) fMRI Activation specific to ex-illiterates
3) Regions of interest surrounding individual peaks of responsivity to faces and houses... 14
4) fMRI effects of lexicality during the lexical decision task
III. Supplementary Figures
1) Figure S1. Reading scores in the six groups of participants.
2) Figure S2. Examples of stimuli used in the localizer run (translated into English)........ 17
3) Figure S3. Examples of stimuli used in the visual runs
4) Figure S4. Brain-imaging evidence of participants’ compliance, comprehension, and literacy
5) Figure S5. Activation specific to ex-illiterates during sentence reading.
6) Figure S6. Analyses of cortical competition induced by literacy at and around the individual peak responses to faces and houses.
7) Figure S7. Plots of fMRI responses at a-priori peaks within left-hemispheric perisylvian language areas.
8) Figure S8. Comparison of activations to written and spoken sentences
IV. Supplementary tables
1) Table S1. Schooling and behavioral profiles of the six groups of participants............... 23
2) Table S2. Additional tests performed on peaks identified by the main SPM analysis.... 24 V. References for supplementary materials
I. Materials and methods
1) Terminology Throughout this study, the term literacy refers specifically to the acquired capacity to read and write. The term schooling refers to the broader experience acquired by attending school during childhood, which includes abilities such as literacy, numeracy, mathematics, social skills, executive control, etc. The factors of schooling and literacy are thus nested inside each other, but can be partially uncorrelated because literacy can also be acquired later in life, during adulthood, and therefore without early schooling. We refer to this experience as late literacy, and to those who achieved it as ex-illiterates. An important question for the present research is whether late literacy can fully mimic the effects of early schooling, or whether early schooling has a unique impact on some brain systems.
It is important to recognize that literacy is neither a one-dimensional, nor an all-or-none variable, as learning to read can have dissociated and graded effects on different tasks such as letter knowledge, rapid naming, phonemic awareness, etc (S1). Here, as a first approximation, we summarize the inter-individual variations in speed and ease of reading by a single measure of reading performance, defined as the number of words and pseudowords that could be read accurately per minute (see figure S1).
2) Definition of groups and participant selection Our aim was to assemble, both in Brazil and in Portugal, three groups of participants: literates, ex-illiterates, and illiterates. Ex-illiterate and illiterate subjects were defined as adults who had received no schooling during childhood, or extremely little. Among this population, ex-illiterates were defined as those who had received a sufficient amount of adult education and informal training to be able to read at least a few simple words, while illiterates were those who were still unable to read even simple words (but could identify some letters). Finally, literate subjects, who had received a normal education to literacy at an early age and were all normal readers, were matched to these groups in age and sex.
All our subjects were fully functional in their daily lives, socially integrated and in good health at the time of the tests. All anatomical MRIs were analyzed by an experienced clinical radiologist in order to rule out brain anomalies. One subject (from the Portugal literate group) presented MRI images compatible with Budd-Chiari syndrome, absolutely asymptomatic.
Portuguese subjects were mostly recruited in Portugal, pre-tested behaviorally at the Faculty of Psychology of the University of Lisbon, and flown to the NeuroSpin Center (Saclay, France) for neuroimaging. For simplicity, seven literate Portuguese subjects were also recruited in the greater Paris area through announcements on the radio and on internet sites dedicated to the Portuguese community. The Portuguese ex-illiterates were recruited through Portuguese governmental agencies. Brazilian illiterate and ex-illiterate subjects were recruited in the greater Brasilia area, either through community searches or through adult literacy courses. They were pre-tested at local community schools or at home and underwent neuroimaging scans at the Brasilia unit of the SARAH Network of Neurorehabilitation Hospitals. The literate Brazilian subjects constituted 2 groups: one recruited from the illiterate community (family or neighbors), matched to some extent on socio-economic and cultural characteristics; and another, consisting of college-graduate professionals from the SARAH Network.
Efforts were made to standardize data collection in Portugal, France and Brazil, using the same subject instructions, assessment criteria, fMRI and ERP equipment, MR sequences, and analysis scripts.
Participants tested in France received 80 Euros for their participation, and those flying from Portugal were housed for 3 nights and invited to a one-day tourist visit of Paris. Since Brazilian legislation does not permit remunerating participants of scientific studies, allowing only for reimbursement of related expenses, volunteers tested in Brasília were given food items equivalent to one day’s work. All participants were motivated by the study itself and will be shown the results of the research when it is published.
A total of sixty-three subjects (41 from Brazil and 22 from Portugal and France) participated in the study. Six groups were formed according to the participant’s literacy status and place of origin (Brazil or Portugal). Table S1 lists their characteristics. All subjects were right-handed (Edinburgh inventory), had normal or corrected-to-normal visual acuity (Snellen or Jagger chart), normal audition, and reported no neurological diseases. All participants gave their written informed consent (the consent form was read aloud and explained to illiterates). Clinical examination ensured that all subjects did not have any counter-indication to MRI scanning.
Illiterates. The illiterate participants (n=10) were from rural areas (5 subjects) or urban centers in the Brasilia vicinity. Participants were illiterates for social reasons, with no history of special difficulty other than lack of access to schools. The families of all these participants were originally from rural areas, and their parents had been illiterate rural workers. The average age of the illiterates was 53.3. Between 44 and 48 years ago, according to estimates in Brazilian records, the rate of illiteracy was 39.6% of the total population and was particularly prevalent in rural areas. There were almost no accessible schools in these communities, nor public transportation to bring children from these removed regions to areas with schools. As a result, most of the participants in our illiterates group did not attend school at all as children, and only two attended it very inconsistently, starting at 10 years of age (either once a week, or only 2 months a year for a number of years). One was able to identify 15 letters and the other, 16, but they were unable to read words or pseudowords. A few illiterates later reported attending adult alphabetization classes but only for a few months and with insufficient focus to impact on literacy (e.g. the maximum reported was 3.5 years of very inconsistently attended evening classes, but this person still could not read a single word or pseudoword). Four women were housewives, the others were all employed as rural workers, housemaid, street vendor, public servant, or construction worker.
In spite of our efforts, we did not manage to form a group of Portuguese adult illiterates of comparable size. In both Portugal and Brazil, many adults now receive some reading education once they are identified as illiterate by associations or social services. Furthermore, Portuguese participants had to engage in a long series of questionnaires and behavioral tests, pass several behavioral and medical criteria for fMRI (e.g. absence of ferrous metals in the body), and finally accept a three-day trip and a day of examination with unfamiliar fMRI and ERP techniques. These constraints exceedingly reduced the number of Portuguese illiterate subjects willing to participate.
Ex-illiterates. The Brazilian ex-illiterates (n=10) had backgrounds similar to those of the illiterates (illiterate parents from rural areas), but now lived in suburban settings in greater Brasilia. Like the illiterates, their early education was quite poor: eight had not attended school during childhood, and two attended it for 1 and 3 years, respectively. All had attended adult courses, and their total education was quite variable, ranging from 0.5 to 7 years. However, even those who reported a total education of less than 2 years (6 participants) reached the criterion for reading ability, suggesting they had found enough stimulation for learning to read outside of official courses. Three were housewives, four were construction workers while the others worked as electrician, taxi driver or dressmaker. In Portugal, among the unschooled subjects, we could only recruit ex-illiterate participants (n=11). Only one of them attended school in childhood, and in this case only during the first half of the school year. They all attended alphabetization classes for adults (from 1 to 5 years, mean 2.4) for a longer period than Brazilian ex-illiterates did (from 0.8 to 4 years, mean 1.4).
However, compared to these, they displayed lower and more variable reading skills as described below. Five were janitors or maids, two were unemployed, one was retired, the others worked as taxi driver, mason or receptionist.
Literates. The Brazilian literates (n=21) were separated into two groups: the LB1 group (n=10) comprised only college graduates (16 years or more of formal education), with a medium to high socio-economic status, while the LB2 group (n=11) comprised participants from the same social community as the illiterate group, but with 2 to 7 years of early education. Thus, while the LB1 group included two systems analysts, two psychologists, two nurses, one physician, one architect, one teacher, and one translator, the LB2 group included three construction workers, two maids, one housewife, one rural worker, one watchman, one yardman, one dressmaker, and one unemployed person. The literate Portuguese group (LP, n=11) comprised a mixture of participants from various SES groups: four of them had limited education (5 years of both early and total education) and came from the same social background as the Portuguese ex-illiterates; two had, respectively, 9 and 10 years of both early and total educational, and the remaining five had college-level education and upper background.
Group matching and potential confounding variables. When testing participants with distinct, often life-long histories of education, it is difficult to ensure that other social or environmental variables are perfectly matched. However, note that all of the Brazilian participants in the illiterate (ILB), ex-illiterate (EXB) and low-SES literate groups (LB2) belonged to the same social communities and occupational classification (natural resources, construction and maintenance), with an average minimum monthly salary of US$292 or EU$230. Thus, restricted comparisons involving these groups are tightly controlled. We cannot fully exclude that literacy led to changes in exposure to faces or houses (e.g. in magazines). However, all of the subjects in the 4 Brazilian groups, including the illiterates, had TVs at home. Watching TV is a common habit of all households in Brazil, irrespective of socio-economic status. Furthermore, previous studies have demonstrated that illiterates typically have normal scores in basic-level object naming and do not differ from ex-illiterates in a variety of psychophysical tests. On a part-verification task, as long as enantiomorphy (left-right discrimination) is avoided, there is a large effect of schooling but no effect of literacy (S2, S3). Similarly, in the domain of cognitive styles, the Frame-Line Test revealed a large effect of culture (Thaïland vs Portugal) and schooling in the Western world, but not of literacy (S4). As regards general cognitive abilities, the mini-mental state examination (MMSE), once discarding the two items that examine reading and writing abilities, typically reveals no difference between illiterates and ex-illiterates, but only between illiterates and schooled literates (S3).