Working Memory and Fluid Intelligence : The Role of Executive Processes , Age and School Type in Children *

r e s u m e n El objetivo de este estudio fue constatar qué componentes del modelo de memoria de trabajo (MT): fonológico, visuoespacial y ejecutivo central, predicen el rendimiento en la inteligencia fluida (IF), teniendo en cuenta edad, nivel de educación y tipo de escuela. Participaron 419 niños, entre seis y 12 años, del primero al sexto grado de primaria de escuelas públicas y privadas de Río Grande del Sur (Brasil). Se aplicaron los subtests de MT del Instrumento de Evaluación Neuropsicológica Breve Infantil (NEUPSILIN-Inf) y el Test de Matrices Progresivas de Colores de Raven como medida de IF. En el análisis de regresión lineal, el componente ejecutivo explicó principalmente la relación entre MT y IF en la infancia, en lugar del componente fonológico. Cuando se incluyeron variables sociodemográficas, edad, tipo de escuela y el componente ejecutivo explicaron 47 % de la varianza de la IF, pero hubo una reducción en el poder predictivo del componente ejecutivo. Los resultados refuerzan la relación entre el procesamiento ejecutivo y la FI, así como la importancia de tener en cuenta las variables sociodemográficas, de modo que la magnitud de la relación entre estos constructos no se sobrevalore. Palabras clave memoria operativa; inteligencia; neuropsicología doi:10.11144/Javeriana.UPSY13-3.umfi Para citar este artículo: Burgues, J., da Rosa, L., Paz, R., & Fumagalli, J. (2014). Working memory and fluid intelligence: The role executive processes, age and school type in children. Universitas Psychologica, 13(3), 934-946. http://dx.doi.org/10.11144/ Javeriana.UPSY13-3.umfi * Artículo de investigación ** Universidade Federal do Rio Grande do Sul, Brasil. Miembro del Centro de Estudios e Investigación en Neuropsicología Cognitiva. E-mail: [email protected] *** Universidade Federal do Rio Grande do Sul, Brasil. E-mail: [email protected] **** Pontifícia Universidade Católica do Rio Grande do Sul, Brasil. Associate Professor of Faculty of Psychology and Graduate Program in Psychology. E-mail : [email protected] ***** Universidade Federal do Rio Grande do Sul, Brasil. Associate Professor of Faculty of Psychology and Graduate Program in Psychology. E-mail: [email protected] Juliana Burges sBicigo, luciane da rosa Piccolo, rochele Paz Fonseca, Jerusa Fumagalli de salles 936 Un i v e r s i ta s P s yc hol o g i c a V. 13 No. 3 j U l i o s e P t i e m Br e 2014 Evidence suggests that working memory (WM) and general intelligence are related constructs (Barbey, Colom, Paul, & Grafman, 2013; Friedman et al., 2006; Roca et al., 2009). More specifically, a relationship between the central executive of WM and fluid intelligence (FI) has been confirmed in studies with adults (e.g. Colom & Flores-Mendoza, 2006; Fukuda, Vogel, Mayr, & Awh, 2010; Unsworth, Redick, Heitz, Broadway, & Engle, 2009). In the last years, findings of that relationship were also found in childhood (Belacchi, Carretti, & Cornoldi, 2010; Cornoldi, Giofrè, Calgaro, & Stupiggia, 2013; Engel de Abreu, Conway, & Gathercole, 2010; Hornung, Brunner, Reuter, & Martin, 2011; Tilman, Nyberg, & Bohlin, 2008). However, one limitation of these investigations is that sociodemographic factors that may interfere in that relationship have not been controlled or manipulated. In the present study the relationship between WM and FI is investigated in children, taking into account some sociodemographic factors that may also contribute to cognitive performance, such as schooling, socioeconomic status and school type (Ardila, Rosselli, Matute, & Guajardo, 2005; Villaseñor, Martín, Díaz, Rosselli, & Ardila, 2009). FI has been defined by Cattell (1971) as the cognitive capacity of adaptive and flexible thinking when there are no resources already classified in memory to respond to complex tasks. Such capacity includes mental operations like the recognition and formation of concepts, problem resolution, extrapolation and transformation of information. In other words, FI is the capacity to reason under new conditions, which is opposite to the performance based on learned knowledge, i.e., crystallized intelligence (Horn & Cattell, 1966). In order to investigate FI, non-verbal tasks and tests are usually employed, as they are less dependent on culture and language (Engel de Abreu et al., 2010; Fonseca, Salles, & Parente, 2007). Raven’s Progressive Matrices Test is among the most used ones (Cornoldi et al., 2013; Engel de Abreu et al., 2010; Hornung et al., 2011; Tilman et al., 2008). FI is closely associated with WM (Belacchi et al., 2010; Colom & Flores-Mendoza, 2006), defined by Baddeley (Baddeley, 2007, 2012; Baddeley & Hitch, 1974), in his multi-component model, as a limited capacity system that temporarily stores information and also processes it, making it possible for the individual to perform complex activities, such as reasoning, learning and understanding. The multi-component model assumes that WM is divided in four components: the phonological component (phonological loop), the visuospatial component (visuospatial sketchpad), the central executive and the episodic buffer. The phonological component is responsible for the storage and temporary maintenance of sequences of acoustic elements or elements based on discourse. The visuospatial component is specialized in the temporary maintenance of visuospatial element information. The central executive is an attentional system with limited capacity that selects and manipulates the material coming from previous components, acting like a WM global system controller. Finally, the episodic buffer is a multimodal code mechanism that allows the interaction among various WM subcomponents and long-term memory. In this study, the last component will not be directly assessed. Phonological and visuospatial components, considered as storage subsystems in the multi-component model, are assessed through simple span tasks that simultaneously require the retention and additional processing of the stimulus (Conway, Jarrold, Kane, Miyake, & Towse, 2008; Engel de Abreu et al., 2010). Simple span tasks (e.g. forward digit span, pseudowords and/or nonword span, Corsi blocks forward) assess the storage capacity for short periods of time in situations that do not impose other cognitive demands (Gathercole, Alloway, Willis, & Adams, 2006). Complex span tasks require a double task (e.g. word span in sentences, backward digit span, Corsi blocks backward) and thus evaluate the executive capacity of simultaneously storing and processing the information, that is, central executive component (Alloway, Gathercole, Willis, & Adams, 2004). Even though it is possible to establish a differentiation among WM components and assess them in separate tasks, it is important to emphasize that no task is a pure measure of those capacities (Conway et al., 2008). Some studies have indicated that simple and complex Working MeMory and Fluid intelligence Un i v e r s i ta s P s yc hol o g i c a V. 13 No. 3 j U l i o s e P t i e m Br e 2014 937 span tasks likely require both storage capacity and central executive, but in different degrees. Complex span tasks may primarily involve central executive and secondary storage capacity, while simple span tasks may require more storage capacity and less central executive component (Conway, Macnamara, Getz, & Engel de Abreu, in press; Kane et al., 2004; Unsworth & Engle, 2007). Studies testing structural models have strongly supported the view that storage (simple tasks) and executive (complex tasks) components are already distinguishable, although related, in children (e.g. Alloway et al., 2004; Engel de Abreu et al., 2010; Hornung et al., 2011). These findings are consistent with the Baddeleỳ s model for adults (Baddeley, 2012). Throughout the last years, researchers have been trying to understand how and why WM and FI are related (e.g. Belacchi et al., 2010; Colom, Abad, Quiroga, Shih, & Flores-Mendoza, 2008; Cornoldi et al., 2013; Dang, Braeken, Ferrer, & Liu, 2012; Engel de Abreu et al., 2010; Engle, Kane, & Tuholski 1999; Hornung et al., 2011; Oberauer, Süß, Wilhelm, & Wittmann, 2008; Redick, Unsworth, Kelly, & Engle, 2012; Tilman et al., 2008; Unsworth & Engle, 2007). The results of studies with both adults and children are controversial in what concerns the role of WM components in the explanation of the performance in FI tests (Belacchi et al., 2010; Colom, Rebollo, Abad, & Shih, 2006; Dang et al., 2012; Engle et al., 1999; Engel de Abreu et al., 2010; Hornung et al., 2011; Tilman et al., 2008). In studies with adults, there is evidence that when the common variance between the storage (simple tasks) and executive (complex tasks) components is removed, only the executive component is associated with FI (Engle et al., 1999). Although the association between FI and executive component has been also found in other studies (Conway, Cowan, Bunting, Therriault, & Minkoff, 2002; Dang et al., 2012; Kane et al., 2004), FI was significantly associated as well with both storage and executive components (Colom et al., 2006; Conway et al., 2008). In addition, storage was a better predictor of FI in some of these studies (Colom et al., 2008; Colom et al., 2006). These inconsistent findings are also observed among the few studies with children (Belacchi et al., 2010; Cornoldi et al., 2013; Engel de Abreu et al., 2010; Hornung et al., 2011; Tilman et al., 2008). Engel de Abreu et al. (2010)’s study with 119 children aged between 5 and 9 years indicated that, after the control of shared variance between phonological storage and executive component, only the executive component assessed through complex tasks was predictor of FI. This relation between executive component and FI was also found in other studies (Cornoldi et al., 2013; Swanson, 2008). However, Tilman et al. (2008), in a study with 196 children aged between 6 and 13 years, have verified that both components contributed in the prediction of FI and, thus, they had considered that the executive component does not necessarily have a primary role in the relationship with FI. Hornung et al. (2011)’s study indicated that short-term storage capacity (composed of the shared variance of simple and complex span tasks – general-domain) primarily explicated the relationship between WM and fluid intelligence. Therefore, more investigations are needed to clarify that question. In addition to the controversial findings, the available studies rarely take sociodemographic variables into account to explore the relationship between WM and FI. For example, in some studies reported earlier, children had come from families of different socioeconomic levels and this factor was not considered in the analysis (Hornung et al., 2011; Tilman et al., 2008). In some studies, socioeconomic information is not even mentioned (e.g. Cornoldi et al., 2013). Ignoring these aspect can lead to an overestimation of the relationship between WM and IF. There is evidence that age, schooling and socioeconomic level (measure correlated with school type) interfere in cognitive performance (Foss, Vale, & Speciali, 2005; Villaseñor et al., 2009). In Brazilian studies, it is especially relevant to consider school type, having in mind that national examinations have revealed a superior school performance from private school students (Instituto Nacional de Estudos e Pesquisas Educacionais Anísio Teixeira [Inep], 2006). The difference in performance may be explained by qualitative aspects in the schoolJuliana Burges sBicigo, luciane da rosa Piccolo, rochele Paz Fonseca, Jerusa Fumagalli de salles 938 Un i v e r s i ta s P s yc hol o g i c a V. 13 No. 3 j U l i o s e P t i e m Br e 2014 ing process, such as teaching methods and level of stimulation, quality of the relations maintained in the school context with colleagues and teachers, among others (Gardinal & Marturano, 2007). Based upon those considerations, the relationship between WM and FI in children was investigated. More specifically, the aim was to verify which components of WM best explain FI, taking sociodemographic factors into account.