What is g?

William T. Dickens
William T. Dickens University Distinguished Professor of Economics & Social Policy - Northeastern University

May 3, 2007

Large secular gains in cognitive ability (the Flynn Effect) show that large, environmentally induced changes in measured cognitive ability are possible, but several studies have suggested that secular gains are not gains in general cognitive ability because the gains across different tests are not proportional to the degree to which each test reflects general cognitive ability.

Many have argued that secular gains are therefore not gains in real cognitive ability, but represent only changing measurement error. In his new paper which Dickens presented at the 2007 meetings of the American Education Research Association he extends the model of a single cognitive ability presented by Dickens and Flynn (2001) to multiple abilities. It shows that such a model can account for all the important facts about general cognitive ability without postulating any common underlying physiological cause for different mental abilities. A general intelligence factor (the tendency for someone who is good at any cognitive skill to be good at the full range of cognitive skills) arises in the model because people who are better at any cognitive skill are more likely to end up in environments that cause them to develop all skills. Scores on the resulting general ability factor can be highly heritable even while they are potentially subject to considerable environmental influence. Loadings of subtest scores on the general ability factor can be positively correlated with subtest heritabilities. In the model, discrimination against a social group in access to cognitively demanding environments can produce subtest score differences from other groups that are strongly correlated with both the g loadings and heritabilities of those subtests. Despite this, there is no reason to expect that meaningful secular gains should be correlated with g loadings across subtests. Thus there is no reason to conclude that secular gains do not represent real increases in cognitive capabilities.