The relationship between intelligence and reaction time varies with age: Results from three representative narrow-age age cohorts at 30, 50 and 69 years

Background: Reaction time (RT) has played a prominent part in research on mental ability for over a century. Throughout this time a number of questions have been repeatedly posed: what is the relationship of RT to general mental ability, and is this the same for simple and choice RT? Does the relationship change with age? How important is RT variability compared with mean values? Here we examine these questions in three population representative cohorts. Methods: Participants were drawn from the West of Scotland Twenty-07 study, a longitudinal population based study designed to investigate socially structured health inequalities. At the fourth wave of data collection, part I of the Alice Heim 4 (AH4) test of general intelligence was administered, and reaction times were measured using a portable device. Means and standard deviations were recorded for simple and 4-choice reaction time. Full data were available for 2196 participants, comprising 714 aged 30 years, 813 aged 50, and 669 aged 69. Results: Correlations of simple RT means with AH4 scores were -0.27, -0.30 and -0.32, for age 30, 50 and 69, respectively; and -0.44, -0.47 and -0.53 for 4-choice RT. The underlying relationships showed evidence of non-linearity, particularly for simple RT, with stronger association at lower AH4 scores. This was more pronounced with age. RT variability was correlated with the mean at 0.57, 0.57, 0.58 for simple RT; and 0.53, 0.53, 0.47 for choice RT. Residuals from regressing the RT variability on the mean showed no association with AH4 in the case of simple RT but a weak association for choice RT which decreased with age. Conclusions: There is a strong correlation of RT means with general mental ability which increases with age. The underlying relationship is complex for SRT. RT variability shows little association with mental ability when its dependence on the mean is removed. Combining samples with disparate ages may overestimate the association.