“A Dynamical Model of General Intelligence: The Positive Manifold of Intelligence by Mutualism”, 2006 (; backlinks):
Scores on cognitive tasks used in intelligence tests correlate positively with each other, that is, they display a positive manifold of correlations. The positive manifold is often explained by positing a dominant latent variable, the g factor, associated with a single quantitative cognitive or biological process or capacity.
In this article, a new explanation of the positive manifold based on a dynamical model is proposed, in which reciprocal causation or mutualism plays a central role. It is shown that the positive manifold emerges purely by positive beneficial interactions between cognitive processes during development. A single underlying g factor plays no role in the model.
The model offers explanations of important findings in intelligence research, such as the hierarchical factor structure of intelligence, the low predictability of intelligence from early childhood performance, the integration/differentiation effect, the increase in heritability of g, and the Jensen effect, and is consistent with current explanations of the Flynn effect.
[Keywords: intelligence, g-factor, dynamical systems, mutualism, reciprocal causation]
…A century ago, Spearman (1904, 1927) introduced the notion of mental energy as the main cause or origin of g. Many current explanations are of this ’single quantitative latent factor’ type. We denote this the g explanation. For instance, it has been argued that individual differences in g are due to individual differences in an underlying cognitive factor, such as speed or efficiency of information processing, working memory, or the capacity to handle cognitive complexity (for reviews, see 2002; 2002; 1998). Alternatively, g is identified with underlying biologically related factors such as brain size, neural efficiency or pruning, or neural plasticity (2002; 2002; 2004). Although there is ample evidence that these factors play a major role in intelligence, none of these factors is generally accepted as the unitary cause of g ( et al 2005; et al 2005).
…1927 and 1951 proposed one such alternative mechanism, namely, sampling. In this sampling theory, carrying out cognitive tasks requires the use of many lower order uncorrelated modules or neural processes (so-called bonds). They hypothesized that the samples of modules or bonds used for different cognitive tests partly overlap, causing a positive correlation between the test scores. In this view, the positive manifold is due to a measurement problem in the sense that it is very difficult to obtain independent measures of the lower order processes. 1998 and Eysenck 1987 identified 3 problems with this sampling theory. First, whereas some complex mental tests, as predicted by sampling theory, highly load on the g factor, some very narrowly defined tests also display high g loadings. Second, some seemingly completely unrelated tests, such as visual and memory scan tasks, are consistently highly correlated, whereas related tests, such as forward and backward digit span, are only modestly correlated. Third, in some cases brain damage leads to very specific impairments, whereas sampling theory predicts general impairments. These 3 facts are difficult to explain with sampling theory, which as a consequence has not gained much acceptance.
…The aim of this article is to outline a third possibility, a new explanation of the positive manifold [mutualism]. This explanation is based on a mathematically formulated developmental model with mutualism or positive beneficial relationships between cognitive processes. This explanation identifies a plausible mechanism that gives rise to the positive manifold but that does not include g as a latent quantitative variable. At the very least, this demonstrates that a latent variable, which is well established psychometrically (ie. in factor analyses), need not correspond to an actual quantitative variable, such as speed of processing or brain size.
…Our dynamical explanation of the positive manifold of cognitive tasks is based on this type of interaction in multivariate dynamical systems (cf. van 1991). We argue that the positive manifold may be a by-product of the positive interactions between the different cognitive processes of the system. In our proposal, all processes of the system are initially undeveloped and uncorrelated. During the development of the system, the dynamical interactions give rise to correlations among the processes of the system.
…The next major step is the assumption that these cognitive processes have mutual beneficial or facilitating relations. Each process supports the development of other processes. This view of relations in developing complex systems is in accordance with modern views of dynamical systems (for discussion, see 1999). These positive relations can be direct (bidirectional or reciprocal) or indirect (via other processes). Reciprocal causal relations are well known in the psychological literature. For instance, better short-term memory helps to develop better cognitive strategies, and better strategies make it possible to increase the efficiency of short-term memory (2005). There are many examples of positive influences of language on cognition, and visa versa. Examples are syntactic bootstrapping ( et al 1994), and semantic bootstrapping (1994). Similar examples are the relations between cognition and meta-cognition (1998), between action and perception (1986, The Ecological Approach To Visual Perception), and between performance and motivation (1986). Clearly, these positive reciprocal relations are not limited to the intellectual domain. For instance, abstract thinking may help to find creative solutions for interpersonal social or emotional problems (2002), whereas good control over emotional and social life are beneficial to academic success (2003). Ideally such positive influences are demonstrated in experimental research, in which the independent variable is manipulated experimentally. It is of course possible that there are no facilitating interactions between certain processes, or even competitive or debilitating interactions. A simple example of the latter is the time constraint on cognitive expertise. Becoming an expert in say, chess, may not allow other specializations. Below we demonstrate that the model can include a good degree of zero or competitive interaction without affecting the fundamental result of the positive manifold of correlations. In short, we propose to view the cognitive system as a developing ecosystem (or society) with primarily cooperative relations between cognitive processes. Note that this model does not make use of latent variables.
In his classification of stereotypical influence patterns that may describe correlation data, 1965 called this model structure the general reticule (see 1984). Cattell never investigated this structure.
…First, the model provides a plausible explanation of hierarchical factor structures. Variability in the interaction weights in M, provided the average of M is positive, leads to complex positive manifolds, as observed in real data. Second, the model explains a number of developmental effects. The low correlation between infant test performance and adulthood IQ can be explained by the fact that the asymptotic states are independent of the growth parameters and the initial values, which together determine the model’s behavior in the initial phase of development. The correlation between test performance and adulthood IQ increases quickly because the limited resources and mutualism influence both the growth speed and the asymptotic states. Third, the mutualism model allows for interrelated integration/differentiation effects. In contrast to other models, differentiation can occur in the model without invoking any additional mechanism. However, when we assume an increase in the variance of mutualistic interactions during development, differentiation and the decline of a limited set of cognitive processes in adulthood can be explained. Fourth, results obtained with the method of correlated vectors do not pose a problem. They can be explained without further assumptions. Fifth, the increase in heritability of intelligence follows from the mutualism model if we are willing to assume that genetic effects are (primarily) on the limited resources K. Sixth, provided the genetic contributions to individual differences in K are minimally correlated (ie. correlations in the order of 0.01–0.09; see Figure 9), we can explain the Jensen effect, that is, the correlation between factor loadings and heritabilities of subtests. This assumption of low correlations between genetic contributions to individual differences in K is in accordance with the low correlation between singles genes and g in QTL research (2004). Finally, the model may be extended, with reciprocal causal relations between phenotypic intelligence and environmental factors leading to a gene-environment correlation that masks the potency of the environment. According to 2001, this accounts for the coexistence of a high heritability of psychometric g and a large environmental (Flynn) effect.
It is important to note that the mutualism model is consistent with many other models and theories of psychological development. It is a nonlinear dynamical model, and, as such, related to much recent work in developmental psychology (2003; van der 1992). Especially relevant are the applications of van Geert (van 1991, van 1994 [Dynamic systems of development: Change between complexity and chaos]) in developmental psychology and the model of 2001.
…The objections raised to the sampling theory (see introduction) are also less relevant to the mutualism model, because functional independence does not imply a developmental independence. Performance on simple reaction time tasks and performance on intelligence tests, such as the Raven’s Progressive Matrices test, may not have much in common. They may be functionally independent. Yet, in the development of reasoning processes that are important in the Raven test performance, speed of processing could well have been very important. Another example is the relation between short-term memory and many cognitive skills. In the first phase of skill acquisition, short term memory is essential, but later, when processes are automatized, short-term memory is no longer involved in performance (1982). Also brain damage might selectively impair performance on one type of test without impairing other, highly correlated, performances in the population, because the correlation is not based on current functional dependency or overlap in processes but rather on developmental dependency. Moreover, correlation between processes can be based on many Mij through indirect pathways.