The General Factor of Intelligence

How General Is It?

Robert J. Sternberg, Elena L. Grigorenko, Robert J. Sternberg, Elena L. Grigorenko

• Publisher: Psychology Press
• Year: 2002
• ISBN:
  9781135655143
• Edition: 1
• Topic: Psychology
• Subtopic: History & Theory in Psychology
• Index: Psychology

This edited volume presents a balanced approach to the ongoing debate of just how general the "general factor" of intelligence is. To accomplish this goal, the editors chose a number of distinct approaches to the study of intelligence--psychometric, genetic-epistemological, cognitive, biological, behavior-genetic, sociocultural, systems--and asked distinguished scholars to write from the standpoint of these approaches. Each approach comprises two chapters, one by a scholar leaning toward a view arguing for the greater generality of g, and the other by a scholar leaning toward a view arguing for the lesser generality of g. The scholars are not simply "for" or "against" these outlooks, rather they provide a more textured view of the general factor, attempting to explain it in psychological terms that are easily understandable. Intended for psychologists in all areas, including clinical, consulting, educational, cognitive, school, developmental, and industrial-organizational, this book will also be of interest to educators, sociologists, anthropologists, and those interested in the nature of intelligence.

Part I
PSYCHOMETRIC APPROACH

Chapter 1
Tracing Psychology’s Invisible
giant and Its Visible Guards

Andreas Demetriou
University of Cyprus

This book presents a fascinating array of thorough and exhaustive chapters on the nature of human intelligence. The authors of these chapters, some of the world’s leading authorities in the field, were asked to evaluate the evidence available and tell us whether there is a common set of processes and abilities permeating all kinds of human understanding and problem solving or whether no such set exists. In more technical terms, the contributors to this volume were invited to elaborate on whether intelligence includes a general factor, the famous g, or independent domain-specific faculties of intelligence.

The volume was meant to be exhaustive. In this respect, it involves sections concerned with all of the important approaches to intelligence that have generated a satisfactory body of research and theory on the question of interest (i.e., the psychometric, the genetic-epistemological, the cognitive, the biological, the behavior-genetic, the sociocultural, and the systems approach). Moreover, the volume was meant to be fair and balanced in its treatment of the issue under discussion. In this respect, each section includes a chapter written by a scholar who is known to take a pro-g stance and another chapter was written by an anti-g (or almost) stance.

As a discussant, according to the editors’ interpretation of my stance, I was invited as a member of the pro-g camp, most likely because in recent years I tried to identify, in my research, general abilities of the mind (Demetriou, Efklides, & Platsidou, 1993). However, a note of caution is in order here. That is, I do not ally myself with either of the two camps— either in theory or research. In fact, I have spent many years trying to specify specialized capacity spheres or abilities in the developing mind (Demetriou & Efklides, 1985, 1987). At present, my aim is to advance a theory of intellectual development that would do justice to both the general and the specific aspects of the mind, and this from the point of view of the psychometric, the cognitive, and the genetic-epistemological approaches (Demetriou & Kazi, 2001). My discussion bears this integrative approach. Thus, I try to answer the following questions:

Does a general factor really exist?
If it exists, what does the general factor include?
What is the architecture of the mind?
How is it related to development?

DOES A GENERAL FACTOR REALLY EXIST?

As an opinion poll about the existence of the g factor, this volume is certainly strongly in favor of it. In fact, only one of the 16 chapters (Berg, chap. 14, this volume) explicitly states that general intelligence does not exist. The rest of the anti-g contributors accept the presence of general abilities but believe that specialized or modular processes and abilities may be more important than the general ones to account for different aspects of intellectual functioning and understanding (see, e.g., Grigorenko, chap. 12, this volume; Kray & Frensch, chap. 8, this volume).

The pro-g contributors, as expected, stress the prominence of general abilities and processes over the specialized ones and thus argue that g is strong, ever-present, and omni-present. However, it should be noted that these contributors also made steps in the direction of the anti-g camp. That is, these contributors do not deny that other, more specialized, functions may be needed, in addition to g, to account for intellectual functioning in particula r domains. Of course, the importance they ascribe to these specialized processes varies. Some believe that they add little in terms of incremental predictive validity (Humphreys & Stark, chap. 5, this volume; Jensen, chap. 3, this volume). Others believe that these specialized functions are equally important as predictors and explanations of intellectual functioning (e.g., Deary, chap. 7, this volume).

The same conclusion is suggested by the chapters concerned with the biological or genetic aspects of intelligence. That is, starting at the level of genes, it is generally accepted that a large part of variance in intelligence (50% or more) is accounted for by shared genes. At the same time, there is strong evidence that there is genetic variance in intelligence that is independent of g (Grigorenko, chap. 12, this volume; Petrill, chap. 11, this volume; Wahlsten, chap. 10, this volume). At the level of the brain, there are aspects of brain functioning (e.g., energy consumption as indicated by glucose metabolism) and parts of it (e.g., the frontal lobes) that seem to be directly and causally associated with all aspects of cognitive functioning. These co-exist with aspects of brain functioning (e.g., different types of neurotransmitters) and areas of the brain that are associated with different types of cognitive functioning (Naglieri & Das, chap. 4, this volume; Wahlsten, chap. 10, this volume). However, few could disagree with Wahlsten, on the basis of principle or empirical evidence, that the complexity of gene action at the molecular level does not say much about intelligence, general or domain specific, at the psychological level. This is so because, according to Wahlsten, intelligence is a property that exists at the psychological level, the level of the thinking individual who possesses a complex brain tutored in a human society. Thus, the “final theory of intelligence,” if there is ever going to be one, would have to involve descriptions and premises running from the genes, to nerves, to thoughts, to society, and to history and evolutionary theory. We explore this point latter, when reference is made to the uses of intelligence to attain complex goals (Sternberg, chap. 16, this volume).

Thus, if this book represents the state of the art in the field of the psychology of intelligence, a clear conclusion is warranted: That is, a century of research on intelligence suggests that there are abilities and processes in intellectual functioning that are truly general, very strong in their effects, and always present. We can see their effects in domains as remote from each other as traffic behavior, functioning in marriage, and job performance, let alone, of course, scholastic and academic achievement. At the same time, nobody denies that there is variation across domains so that excellence and high levels of expertise in one domain may co-exist with more humble achievements in other domains. It is thus equally clear that general abilities are embedded or interleaved with domain or problem-specific abilities and processes.

This conclusion suggests that the battle between the pro-g camp (led by Spearman in the old days and Jensen, 1998, nowadays) and the anti g camp (led by Thurstone and Guildford in the old days and Gardner, 1983, nowadays) has ended with no winner. To the satisfaction of the pro-g camp, g did stand up to the test of time. To their dismay, it cannot do the job of understanding and dealing with the world on its own. It needs completion. To the satisfaction of the anti-g camp, the factors needed for completion do behave systematically and consistently enough to warrant the status of independent dimensions of intelligence on a par with g. To their dismay, these factors may not even be able to be activated and used without the concomitant activation and use of the processes included in g. Therefore, let us first specify what is included in g and the specialized factors and then specify how all of them are organized to allow intelligent functioning.

WHAT DOES THE g-FACTOR INCLUDE?

The g-factor emerges from a matrix of correlations between diverse cognitive tests as a result of the so-called “positive manifold.” That is, it reflects the fact that all tests are positively correlated (see mainly Detterman, chap. 9, this volume; Humphreys & Stark, chap. 5, this volume; Jensen, chap. 3, this volume). The more variable the tests, the stronger the g-factor (Humphreys & Stark, chap. 5, this volume). This statistical construct is supposed to reflect the operation of an invisible power, which, like gravity (Detterman, chap. 9, this volume), underlies and constrains performance on all of the tests and is responsible for the positive manifold.

How is this invisible power defined? No fully agreed upon answer exists. However, there is considerable agreement on a number of functions and characteristics. First of all, there is overwhelming agreement that g can be defined in terms of a number of parameters concerned with processing efficiency and capacity. Three parameters have been systematically studied. Speed of processing is mentioned by everybody. That is, the higher the speed, the more efficient processing proves to be and this leads to better performance on cognitive tasks of all kinds. Different authors emphasize different aspects of processing speed, such as speed of searching or consulting items in short-term memory and rate of gaining information. Efficiency of inhibition is also considered to be a component of g because it is related to the thinker’s ability to stay focused on goal and minimize the effects of interference. A third function, which is considered by many to be involved in g, is working memory. In fact, some authors go as far as to argue that g is working memory (Kyllonen, chap. 15, this volume; Detterman’s views about working memory are very close to this strong interpretation of the role of working memory). This function refers to the capacity of the system to hold information in an active state for the sake of processing. The more the capacity of working memory, the better for cognitive performance.

There is a second, more active, set of processes associated with g. These processes refer to the management of both the processing resources available and the task demands and goals. Attention is one of these processes. In fact, as discussed in the present context, attention may be conceived of as the subjectively and personally felt and controlled aspect of the inhibition processes referred to earlier. It is noted here that Stankov and Roberts (1997) have recently shown that the importance of speed of processing is not due to speed itself but to the fact that speeded tasks require selective attention. Under this perspective, directed attention is the crucial factor.

Planning is another, more complex, process. Planning refers to goal management strategies that enable the person to analyze the task demands in a time-dependent order of goals and subgoals and match them with the resources available so that optimum performance can be achieved, given the problem, the resources available at the time, and the time constraints as such. It is interesting that these two functions, that is, attention and planning, are invoked as the building blocks of intelligence by contributors who take the anti-g stance (Naglieri & Das, chap. 4, this volume). Others (Deary, chap. 7, this volume) note that complex planning and management of complexity depends on the person’s ability to generate and manage goals and subgoals in working memory. Therefore, even at this level, one finds signs of the more fundamental processes of attention, processing control, and representational capacity.

Spearman, the father of the concept of g, defined g as the ability to uncover and deal with relations at different levels of complexity and abstraction. Eduction of relations and correlates were the two basic processes he invoked to specify this ability. Nowadays very few explicitly associate g with these or other reasoning processes as such. This trend is clearly represented in this volume. Only Deary (chap. 7, this volume) has explicitly referred to analogical reasoning as a component of g and this only in order to show that this line of research did not lead to anything of substance in our attempt to crack the mysteries of the mind. It should also be noted that g is not associated with knowledge. Thus, it is natural that the model that splits the general factor into crystallized and fluid abilities is not very popular. In fact, of the 16 chapters in this volume, only one has explicitly invoked these dimensions as building constructs of g (Jensen, chap. 3, this volume).

Moreover, it is the anti-g rather than the pro-g contributors who focus on knowledge and reasoning processes as such. These contributors tend of course to think that these processes are modular and domain- or even task-specific. This is interesting because it suggests that once we come to processes that interface with real world directly, intraindividual variability and instability of performance become the rule and homogeneity and stability the exception (Berg, chap. 14, this volume; Sternberg, chap. 16, this volume). This creates the apparent paradox of having people who are high in g and have very variable performance across different domains, ranging from excellence in the domain of their expertise to moderate in many other domains. In fact, this paradox is the main reason that the struggle between the pro-g and the anti-g camps went on for about a century. I believe that we are ready to resolve the paradox and unite the two camps in one: The higher a person scores in g, the better this person could perform in his or her chosen domain. Thus, by definition, g must co-exist with a wide range of variability. We return to this point latter in the discussion.

ARCHITECTURES OF INTELLECT

Cognitive architecture may be specified in a number of ways. One way, which is common in the psychometric tradition, is to decompose the amount of variance accounted for by different kinds of factors. This is a kind of structural architecture that depicts...