An ‘instantaneous’ estimate of a lifetime's cognitive change
Introduction
When examining a person for signs of cognitive decline, it is important either to know or at least to estimate their prior level of cognitive ability. In the ideal situation, a person's current cognitive function score would be compared with their own baseline. However, such prior cognitive ability data are rarely available, especially from a previous period of good health. Therefore, prior or premorbid mental ability is commonly estimated using the National Adult Reading Test (NART; Crawford, 1992, Nelson & Willison, 1991). The NART tests the pronunciation of 50 phonologically irregular English words. NART performance is resistant to the effects of normal aging (Crawford, Stewart, Garthwaite, Parker, & Besson, 1988) and is relatively resistant to the effects of neurological or psychiatric morbidity (see Crawford, 1992, Crawford, 2003, O'Carroll, 1995 for reviews). We have recently shown that the NART has good retrospective validity in healthy old people. Taken at age 77 years, it correlated .73 (P<.001) with Moray House Test (MHT) scores from the same subjects taken at age 11 (Crawford, Deary, Starr, & Whalley, 2001).
Use of the NART as an indicator of prior or premorbid intelligence rests upon the empirical evidence that the ability to read words does not decline, as other cognitive functions do, with age and some medical conditions Crawford, 1992, Crawford et al., 2001. One influential theory of human intelligence differences, which is highly relevant to the use of NART and to aging and cognitive functions generally, is Cattell's theory of fluid and crystallised intelligence Cattell, 1998, Horn & Cattell, 1966. Crystallised intelligence is related to a person's stored information and cultural influences (Baltes, Staudinger, & Lindenberger, 1999). It is often tested using vocabulary- and general knowledge-type tests. It shows little change with healthy aging. Fluid intelligence is related to basic information processing and is assessed using tests comprising novel materials, often under time pressure. Scores on tests of fluid intelligence decline with age and as a result of some medical conditions. Raven's Progressive Matrices (RPM or Raven) is often used as a measure of general fluid reasoning (Carroll, 1993). In summary, NART score is often used to represent a relatively stable record of the prior best level of cognitive functioning, whereas Raven score is used to represent a relatively labile index of the current level of cognitive functioning.
At the level of the population, the difference between fluid-type and crystallised-type intelligence test scores increases with age. It is increasingly seen as important to employ this difference in the individual person to estimate potentially clinically significant cognitive change Crawford, 2003, Sawrie et al., 1999. The difference between an individual's NART and Raven performances has been investigated as an estimate of their cognitive change Davis et al., 2000, Freeman & Godfrey, 2000, van den Broek & Bradshaw, 1994, but no study until the present report has criterion validated this estimate by correlating it with their actual cognitive change. Here we report such a validation study after conducting two waves of follow-up on survivors of the Scottish Mental Survey of 1932 (SMS 1932).
Section snippets
Participants
For the main analyses, subjects were 87 non-demented people from the Aberdeen area who took part in the SMS 1932 at age 11 years Deary et al., 2000, Scottish Council for Research in Education., 1933. All were born in 1921. The selection of subjects for the first wave of follow-up, at age 77, was described previously (Crawford et al., 2001). The present sample comprised those subjects who undertook a further wave of follow-up cognitive testing and medical examination. Therefore, these subjects
Results
The first results concern the approach that uses Raven scores in both the actual and estimated cognitive change measures. The mean scores for Raven and NART were similar at ages 77 and 78 years (Table 1). The one-year stability of the two tests was very high, with NART at .89 and Raven at .78. Correlations between NART and Raven were between .42 and .55. Estimated cognitive change (NART age 77–Raven age 77) correlated .638 (P<.001) with actual cognitive change (MHT age 11–Raven age 78) (Fig. 1)
Discussion
The ‘instantaneous’ difference between NART and Raven scores has good criterion validity and stability as a measure of relative lifetime cognitive change. Thus, we validated a cognitive change estimate taking about 30 min at a single sitting against actual cognitive change that took about 67 years to measure. Our study involved relatively healthy old people within a narrow age range. Future studies should provide normative data on NART–Raven differences on other age and medical groups for use
Acknowledgements
This research was supported by grants from Henry Smith's Charities, the Scottish Executive Department of Health, Chief Scientist's Office, and the Biotechnology and Biological Sciences Research Council. We thank Mariesha Struth, Helen Lemmon, and Steve Leaper for gathering and compiling data. Ian J. Deary is the recipient of a Royal Society-Wolfson Research Merit Award. Lawrence J. Whalley holds a Wellcome Trust Career Development Award.
References (19)
- et al.
The stability of individual differences in mental ability from childhood to old age: Follow-up of the 1932 Scottish Mental Survey
Intelligence
(2000) - et al.
Lifespan psychology: Theory and application to intellectual functioning
Annual Review of Psychology
(1999) Human cognitive abilities: A survey of factor-analytic studies
(1993)Where is intelligence? Some answers from the triadic theory
Current and premorbid intelligence measures in neuropsychological assessment
Psychometric foundations of neuropsychological assessment
- et al.
The NART as an index of prior intellectual functioning: A retrospective validity study covering a 66 year interval
Psychological Medicine
(2001) - et al.
The relationship between demographic variables and NART performance in normal subjects
British Journal of Clinical Psychology
(1988) - et al.
Estimation of premorbid performance on Raven's Standard Progressive Matrices using reading-test performance: New normative data
Clinical Neuropsychological Assessment
(2000)
Cited by (41)
Aging trajectories of fluid intelligence in late life: The influence of age, practice and childhood IQ on Raven's Progressive Matrices
2014, IntelligenceCitation Excerpt :Childhood intelligence data were kindly provided by the Scottish Council for Research in Education (SCRE) from data archived in the Scottish Mental Surveys of 1932 or 1947. All children born in either 1921 or 1936 and at school in Scotland on the 1st of June 1932 or the 4th of June 1947 when aged about 11 years sat a group-administered intelligence test (The Moray House Test (MHT)) (Deary, Whalley, & Crawford, 2004). Current adult fluid intelligence was assessed using Raven's Standardized Progressive Matrices (RPM) (Raven, 2000), in which each test item consists of a matrix of geometric patterns with one missing cell, and the task for the participant is to select the best completion of the missing cell from a set of alternatives.
Assessment of cognitive function across pregnancy using CANTAB: A longitudinal study
2014, Brain and CognitionCitation Excerpt :A score greater than 10 on the EPDS was considered an indication of increased risk of anxiety and depression. The NART (Nelson, 1991) is used to assess intelligence and provides an IQ (Deary, Whalley, & Crawford, 2004). The test consists of 50 words that participants must read and pronounce.
Does the direction and magnitude of cognitive change depend on initial level of ability?
2012, IntelligenceCitation Excerpt :In order to explore the generalizability of the weak ability–change relations, two more general measures of cognitive ability were also examined. One was the score on the 35-item version of the North American Adult Reading Test (NAART, Uttl, 2002), that has been considered an indicator of the individual's peak ability early in life (e.g., Deary, Whalley, & Crawford, 2004). The other general measure of cognitive ability was the individual's score on the first principal component (1st PC) based on the T11 scores because the 1st PC is sometimes used as an estimate of general cognitive ability (Jensen, 1998).
Signatures of multiple processes contributing to fluid reasoning performance
2018, IntelligenceCitation Excerpt :Accordingly, fluid reasoning appears to represent a collection of processes that are common to a wide range of cognitive domains, specifically those that emphasize controlled processing (Salthouse, 2005). Standardized tests such as Raven's Progressive Matrices (RPM; Raven, Court, & Raven, 1988) have been successful in a variety of demographic groups to study individual differences in fluid reasoning (Deary, Whalley, & Crawford, 2004; Herrnstein & Murray, 1994). When fluid reasoning is considered as a multi-process cognitive construct, the contributions of the underlying processes to the individual differences in performance remain unclear.
Why is Mini-Mental state examination performance correlated with estimated premorbid cognitive ability?
2016, Psychological MedicineThe Stability of Cognitive Abilities: A Meta-Analytic Review of Longitudinal Studies
2024, Psychological Bulletin