“Perceptual Speed Does Not Cause Intelligence, and Intelligence Does Not Cause Perceptual Speed”, 2005-09 (; backlinks):
There is ongoing debate whether the efficiency of local cognitive processes leads to global cognitive ability or whether global ability feeds the efficiency of basic processes. A prominent example is the well-replicated association between inspection time (IT), a measure of perceptual discrimination speed, and intelligence (IQ), where it is not known whether increased speed is a cause or consequence of high IQ.
We investigated the direction of causation between IT and IQ in 2,012 genetically related subjects from Australia and The Netherlands. Models in which the reliable variance of each observed variable was specified as a latent trait showed IT correlations of −0.44 and −0.33 with respective Performance and Verbal IQ; heritabilities were 57% (IT), 83% (PIQ) and 77% (VIQ).
Directional causation models provided poor fits to the data, with covariation best explained by pleiotropic genes (influencing variation in both IT and IQ). This finding of a common genetic factor provides a better target for identifying genes involved in cognition than genes which are unique to specific traits.
[Keywords: inspection time, IQ, direction of causation, genetic modeling, twin study]
…An alternative method to establish direction of causation is to use data from genetically related individuals such as monozygotic (MZ; genetically identical) and dizygotic (fraternal) twins (DZ; sharing roughly half of their genes). When two correlated traits display relatively different sources of variation, eg. variation in one trait is mainly genetic and in the second trait mainly non-genetic, it is possible to resolve the direction of causation between them ( et al 1993). For example, if Trait A is mostly influenced by genes, whereas Trait B is mostly influenced by the common environment, then the direction of causation from A to B predicts the cross-covariance between Twin 1 Trait A and Twin 2 Trait B to be predominantly genetic (ie. larger MZ than DZ covariance), whereas if B causes A, the cross-covariance will be largely environmental. Different expectations for the co-twin cross-covariances apply as long as the genetic influence on each variable is sufficiently different in magnitude and if the sample of twins is sufficiently large (1994).