There are probably two reasons for the paper’s high number of citations: First, the article describes a comprehensive meta-analysis of many disparate studies that have been reported in the literature, and it provides a definitive answer to the question as to whether working memory and intelligence are the same or different constructs. Second, the answer (that these are different constructs) is controversial; especially to some experimental psychologists—even though it is not very controversial to most researchers whose main work is in the area of human intelligence.

“Experimental psychologists are sometimes ignorant of how to conduct studies and analyze data involving individual differences.”

It describes a synthesis of knowledge. The meta-analysis synthesizes a body of research concerning the relations between several different aspects of working memory abilities and different cognitive/intellectual abilities. The article also provides an historical context to the research and central findings on the structure of intellectual abilities.

In the history of the modern study of human intelligence, there have been many attempts to "explain" intelligence by appealing to some single underlying concept that is currently of interest. However, general intelligence as a construct has been very difficult to reduce to a single entity, whether it is sensory acuity, speed of simple reactions, nerve conduction speed, or glucose uptake in the brain, and so on.

Recent research suggested that individual differences in general intelligence could be reduced to working memory—which is conceptualized mainly as a "central executive involved…in planning, monitoring of stimuli" in the chapter entitled: "The working memory model in adult aging research," from the book: "Working memory in perspective," Hove, UK: Psychology Press Ltd., Andrade, J. (Ed.), pp. 101-125, Phillips, L. H., & Hamilton, C., 2001.

Intelligence and related measures are used in a variety of different applications, such as in academic selection, placement, and occupational selection. If working memory captures the essence of general intelligence, it might be possible to discard the traditional measures—those that involve not just memory, but also judgment, knowledge, comprehension, broad content abilities of verbal, numerical and spatial domains, and so on—in favor of tasks that only involve the brief storage and processing of relatively simple stimuli (such as words and numbers).

However, our meta-analysis has shown that one cannot equate working memory abilities with general intelligence. That is, there is far more to general intelligence than is captured by working memory measures.

I have been working in the field of individual differences in human intelligence for the past 20 or so years. The central focus of my research has been in two different areas, namely individual differences in the ability determinants of skill acquisition, and the nature of adult intellectual development.

However, in recent years, there has been much discussion about the importance of the construct of working memory—which is currently a central focus of some domains of experimental psychology—and its relation to intelligence.

The first major paper on this topic appeared in 1990 (Intelligence 14, 389-433, "Reasoning is (little more than) working-memory capacity," Kyllonen, P. C. & Christal, R. E., 1990). These authors suggested a major role of working memory in determining individual differences in abstract reasoning abilities.

Through much of the 1990s and more recently, some investigators went beyond the work by Kyllonen and Christal and started to claim that working memory abilities were in fact the "essence" of not just reasoning abilities, but also general intelligence. One investigator, for example, claimed that working memory was "isomorphic" to general intelligence.

There were several articles in the literature where such thinking was taken as a well-founded result, even though a cursory review of the literature suggested that this might be a vast overstatement.

In some of my own research, I initially thought that I might be missing something by not directly attending to working-memory concepts. So, in an investigation of individual differences in skill acquisition, my students and I incorporated several of these measures to evaluate the role of working memory and its relation to general intelligence.

What we found in that study was that working memory measures neither fully accounted for individual differences in intelligence, nor were the sufficiently separate from individual differences in perceptual speed abilities (Journal of Experimental Psychology: General 131, 567-589, "Individual differences in working memory within a nomological network of cognitive and perceptual speed abilities." Ackerman, P. L., Beier, M. E., & Boyle, M. O., 2002).

Although these results were diagnostic of the lack of complete overlap between working memory and general intelligence, many other investigations appeared to claim that there was high overlap. Thus, we decided to conduct a thorough quantitative meta-analytic review of the literature, in order to settle the controversy. What we found was that there was no valid evidence to support the conclusion that working memory and general intelligence represented the same underlying constructs.

The problems encountered were mainly political. For the past 80 years or so, there has been a tension between traditional fields of experimental psychology and differential psychology (the study of individual and group differences).

Experimental psychologists are sometimes ignorant of how to conduct studies and analyze data involving individual differences. For example, in studying abilities they sometimes use extreme-group designs, where the data from half of the participants in a study are discarded in order to highlight differences between the top 25% and bottom 25% of the distribution.

Such studies frequently result in overly optimistic claims of relationships that do not exist in the larger sample, or exist only in a substantially diminished fashion. Many experimental investigators often fail to properly triangulate their results, by using a single, relatively narrow, measure of ability and calling it "general intelligence," rather than using an adequate sample of tests that represent the broader construct.

The outcome is a confusion of reliability with validity. A fundamental principle of measurement in psychology is that great reliability is necessary, but reliability is not sufficient for achieving validity.

Properly conducting an individual-differences study requires a large number of participants and a substantial amount of time and effort in triangulating the estimation of abilities through multiple tests. The meta-analysis we conducted showed that when several of these issues are accounted for, it is impossible to equate working memory with general intelligence.

Fundamentally, general intelligence is much more complex than any single narrow measure. It involves content abilities (such as verbal, spatial, and numerical), it involves both abstract reasoning and domain knowledge, and it involves processing speed.

Working memory is also related to constructs and tests that have existed for over 100 years —such as short-term memory and perceptual speed—that don’t benefit much from relabeling.

There will probably always be attempts to reduce general intelligence to some fundamental or singular property of the individual. However, even though traditional measures of intelligence have been, in many ways, relatively unchanged in scope and structure since the time of Alfred Binet—whose "New Methods for the Diagnosis of the Intellectual Level of Subnormals" first appeared in L'Année Psychologique, 12, 191-244, in 1905—they still represent the gold standard for predicting individual differences in academic performance, especially for children and adolescents.

One can hope that the work described in this article will better focus research efforts to domains that might yield more effective measures of intelligence, rather than focusing on a construct that has much in common with older measures, but little new to recommend it in terms revealing the fundamental essence of human intelligence or for assessing human intelligence.