Can fluid intelligence be reduced to ‘simple’ short-term storage?

Abstract

Much is written regarding the associations between human intelligence and cognition. However, it is unusual to find comprehensive studies. Here twenty four measures tapping eight cognitive abilities and skills are considered for assessing a sample of one hundred and eighty five young adults. The simultaneous relationships among fluid, crystallized, and spatial intelligence, along with short-term memory, working memory capacity, executive updating, attention, and processing speed are analyzed using a latent-variable approach. The key findings show that (a) short-term storage, working memory, and updating are hardly distinguishable, and (b) fluid intelligence is near-perfectly correlated with these three cognitive functions. It is concluded that this nuclear intelligence component can be largely identified with basic and general short-term storage processes, namely, encoding, maintenance, and retrieval.

Introduction

Since the very beginning of the scientific study of human behavior two main disciplines were identified. The experimental approach is based on the manipulation of independent variables to find out its effect over a previously specified dependent variable, whereas the correlation approach analyzes patterns of co-variation among diverse psychological measures. The still widespread separation of the experimental and correlation approaches slows down progress in the scientific study of human behavior, as noted by several researchers across the years (Cronbach, 1957, Cronbach, 1975, Eysenck, 1967, Terman, 1924, Underwood, 1975).

Perhaps human intelligence is the psychological factor most frequently analyzed at the intersection between the two disciplines of scientific psychology (Wilhelm & Engle, 2005). Because of its relevance for predicting everyday life behaviors, such as scholastic achievement, work performance, health, longevity, or crime (Deary and Der, 2005, Neisser et al., 1996) researchers have asked about the causes underlying the documented intelligence differences at the population level.

In the second part of the XX century hundreds of empirical studies exploring the relationships between intelligence differences, as assessed by standardized tests, and performance differences on a range of cognitive tasks, were published (Deary, 2000, Hunt, 2011, Jensen, 1998, Jensen, 2006, Lohman, 2000). Perhaps the main conclusion that might be derived from these research efforts is that some cognitive functions are much more relevant than others for understanding intelligence differences.

Some theories underline processing speed (Jensen, 2006), while other models appeal to attention scope (Cowan, 2005), dual mechanisms for information maintenance and retrieval (Unsworth, Redick, Heitz, Broadway, & Engle, 2009), or simple short-term storage (Colom et al., 2008, Colom et al., 2006). Unfortunately, the debate is still open because, as noted by Unsworth and Engle, 2007, Unsworth et al., 2009 there are several issues not properly addressed in most of the published studies: (a) limited range of tasks for measuring the constructs of interest, (b) arguable scores derived from these tasks, (c) analyses of raw correlations instead of latent factors, and (d) use of small and improperly recruited samples.

Here an effort is made for overcoming many of these limitations. Following required guidelines for latent variable analyses (Ackerman et al., 2005, Johnson and Bouchard, 2005) a total of eight psychological constructs, both from the correlation and experimental approaches, are tapped by three diverse measures each. The constructs from the correlation approach were fluid, crystallized, and spatial intelligence, whereas the constructs from the experimental approach were short-term memory, working memory capacity, executive updating, attention, and processing speed.

The main interest focuses on the constructs, not on specific measures. Note that these measures tap the constructs as comprehensively as possible. A large sample comprising one hundred and eighty five young adults is assessed by this set of measures. Taking into account our own previous findings (Colom et al., 2005, Colom et al., 2004, Colom and Shih, 2004, Colom et al., 2006, Colom et al., 2008) along with diverse results from other research laboratories (Beier and Ackerman, 2004, Hornung et al., 2011, Krumm et al., 2009, Unsworth and Engle, 2007) it is predicted that memory span will be highly related to fluid intelligence. By using the term ‘memory span’ we are also predicting that short-term memory, working memory capacity, and executive updating will be hardly distinguishable (Colom et al., 2008, Unsworth and Engle, 2007). Importantly, these predictions will be tested concurrently against several potential candidates.