We all 'mean' well, but are we missing something? Alternative methods of reaction time data analysis to explore the relationship between intelligence and inhibitory control of response in the Stroop task

Abstract Background: A century’s worth of data has shown the correlation between reaction time (RT) and intelligence to be consistent, but weak. This paper investigates whether inhibitory processes may be the reason why ‘faster is not always smarter’, with special attention paid to the analysis of RT beyond traditional parameters of central tendency. Aims: Using a cognitive model of the decision making process, we combined traditional measures of speed (mean RT), consistency (RT variance) and accuracy to estimate parameters representing quality of evidence accumulation over time and response conservativeness. We also used the exponential-Gaussian distribution to describe participants’ RT in terms of centrality, dispersion and positive skew. Method: A sample of university students (N=119) completed two and four choice versions of the Stroop Colour Word Task. Congruent, incongruent and neutral stimuli were presented on a computer screen and manual responses via a button box were made. Each task comprised 288 trials, 96 for each Stroop condition. IQ was measured using the Wide Range Intelligence Test. Results & Conclusions: The Stroop effect on mean RT shared just 9% of variance with IQ, with a similar magnitude of relationship for the ex-Gaussian parameters being found. Combining traditional parameters in the cognitive model, we found the Stroop effect on the response conservativeness parameter was more strongly related to IQ, sharing 27.5% of variance. Results not only show that inhibitory control may be more important to cognitive ability than the literature suggests, but also that the way RT is analysed can affect the nature of results.