A genetic relationship between information processing speed and psychometric intelligence has been previously demonstrated. However, there have been no behavioural genetic studies of working memory, a cognitive process which is theorised to play a major role in mediating the relationship between processing speed and intelligence. According to this theory, a faster processing speed would enable information in working memory to be better accessed and manipulated before it decays from temporary storage, leading to better integration of old and new information and more effective problem solving.
This study aimed to extend the investigation of the covariance between processing speed and intelligence by including a measure of working memory in a genetically informative analysis. Based on previous theories that emphasise the interdependence of processing speed and working memory in predicting IQ, it was hypothesised that a single general genetic factor would influence the phenotypic associations between processing speed, working memory and IQ. Other specific genetic factors were hypothesised to contribute to the individual (or unique) variance in each of the three measures of cognitive function.
Data were collected from 183 monozygotic and 207 dizygotic adolescent twin pairs. Processing speed was measured using a Choice Reaction Time task (mean Reaction Time, Standard Deviation) and an Inspection Time task. A visuospatial Delayed Response task (accuracy: percentage correct, position error of response; speed: response initiation time, movement time) was used to measure working memory. The Multidimensional Aptitude Battery provided IQ measures (Full Scale IQ, Verbal IQ, Performance IQ, subtest scores). Fifty twin pairs were retested three months after their initial visit so that estimates of test-retest reliability for the measures could be derived. Univariate genetic modelling of all measures incorporated retest data so that a component of test unreliability could be estimated in addition to additive genetic, common environmental and unique environmental sources of variance. Heritability estimates ranged from 0.41 to 0.83 for IQ measures, from 0.20 to 0.70 for processing speed measures, and from 0.00 to 0.57 for working memory measures. Common environmental effects were mostly nonsignificant. The proportion of test unreliability variance ranged from a high 0.49 for Delayed Response percentage correct to a low 0 10 for Full Scale IQ, with unique environmental components generally accounting for less of the total variance than test unreliability.
An initial multivariate phenotypic analysis of the IQ subtests showed the presence of correlated verbal and performance group factors, a dichotomy which has been previously supported in the Multidimensional Aptitude Battery. A genetic decomposition of this same factor structure showed the increased importance of a genetic general factor rather than genetic group factors, with test-specific genetic effects also being prominent. The effects of common environment were general, while unique environmental effects were mainly specific. These genetic findings were in agreement with studies of the Wechsler IQ subtests.
Multivariate analysis of Choice Reaction Time (three choice conditions) and Full Scale IQ revealed two genetic factors (loadings ranged from 0.37 to 0.73) mediating their relationships, plus a third genetic factor which only affected IQ (accounting for 38% of its variance). Similarly, a bivariate analysis with Inspection Time and IQ showed that this relationship was mediated by a genetic factor (explaining 40% of variance in Inspection Time & 31% of variance in IQ), but that a further factor explained 52% of the genetic variance in IQ. An analysis of Choice Reaction Time, Inspection Time, Performance IQ and Verbal IQ showed that a psychometric factor primarily determined by genes (92%) mediated their interrelationship. Further modelling of the processing speed measures and the IQ subtest scores indicated that in addition to a general genetic factor (loadings ranged from 0.43 to 0.72), Choice Reaction Time shared an additional separate genetic relationship with the performance subtest measures (loadings ranged from 0.16 to 0.54), suggesting a psychomotor function rather than a pure processing speed function as reflected in the general factor.
Model fitting results of the covariation among Delayed Response measures (memory percentage correct, sensory position error, initiation time, movement time) and Full Scale IQ revealed the presence of a genetic factor related to memory accuracy and IQ but not related to sensory accuracy or speed variables. Extending these analyses (memory accuracy, initiation tune, Full Scale IQ) to include measures of processing speed (Choice Reaction Time, Inspection Time) showed the presence of a general genetic factor (loadings ranged from 0.19 to 0.72), a genetic factor influencing Choice Reaction Time, Delayed Response speed and accuracy, and IQ (loadings ranged from 0.26 to 0.81), and specific genetic factors influencing Delayed Response speed and accuracy, and IQ. As processing speed and working memory measures did not show independent genetic associations with IQ, this supported an interdependent role for processing speed and working memory capacity in predicting intelligence. In line with the limited capacity theory asserted by Jensen and Vernon, the general genetic factor is likened to a biological speed of processing in the brain, while the genetic factor influencing the subset of variables is considered to reflect information processes related to psychomotor ability.
Genetic correlations between the processing speed and working memory measures with IQ were substantial (range: 0.35 to 0.71), indicating that the variation in genes which produce faster Inspection Times and Choice Reaction Times, and more accurate Delayed Response performance, is strongly related to the variation in genes which promote higher IQs. Hence, these lower-order cognitive indices may prove invaluable in the molecular detection of genes for intelligence.