In clinical research employing well-known psychoneurological tests as measures of cognitive functioning, it is common to combine tests to form composite cognitive domain scores. Although there is some small variation, there is quite strong agreement among well known and respected neuropsychological textbooks on which tests should be allocated to particular domains when forming the composite scores.
However, for some domains, the results of factor analyses do not suggest the combining of tests in the same way. That is particularly true of tests commonly combined to form the Executive Function cognitive domain. These include tests of working memory, reasoning and problem-solving, semantic verbal fluency, speeded tests involving attention switching (eg TMTB) or suppression of automatic responding (eg the Stroop test), etc. However, in well established factor-analytically based models of human mental intelligence, such as Gf/Gc theory (the Theory of Fluid and Crystallized Intelligence), such tests do not load on the same broad second-order factors. And no factor has been identified with marker tests corresponding to the range of tests commonly used by neuropsychologists to define the Executive Function cognitive domain.
There are other examples of a mismatch between the allocation of tests to cognitive domains according to common neuropsychological practice, and what is suggested by factor analysis, such as the formation of a composite Attention/processing speed cognitive domain.
Can anyone shed light on the reasons why such differences exist? Is it that the grouping of tests according to common practice by neuropsychologist has particular benefit from a clinical perspective, even those it is not consistent with the grouping of tests based on patterns of covariance as displayed in the results of factor-analysis? If so, how might this be explained?
All samples are not similar and do not give the same results. In case this question relates to your personal research, go with your factor analysis and interpret your results from there. Then you can correlate your results with other known results and give a plausible explanation why the results are likely to deviate.
theseThe results of factor (or other multivariate) analysis always reflect correlations among the input variables. However, these correlations exist between the variables within a cognitive domain but also across different domains. The progress in neuroscience will gradually clarify how the factor-analytic results are related to the neuropsychologically defined cognitive functions. Nevertheless, the research results suggest a substantial extent of correspondence even so far.
Let see, for example, the relation between fluid intelligence and executive functioning. According to research results, including quite recent (see address below), both are very substantially related.
https://pubmed.ncbi.nlm.nih.gov/26281913/
Janek, thank you for your response to my question, and those references. Regarding Executive Function, I agree that a number of tasks that are commonly used as markers by neuropsychologists for this cognitive domain have also been found upon factor analysis to be markers of Fluid intelligence. Such tests include those that have been described as measuring working memory, complex rule following, attention switching.
However, there are also other tests that are commonly treated by neuropsychologists as measures of Executive Function, which do not load on Gf, but which are markers of other second order factors. For example, the phonemic verbal fluency test, F-A-S, is widely used as an Executive Function marker, while in the factor-analytically based Gf/Gc theory, it would not be a marker of Gf, but share loading between the general fluency factor, F, and Crystallized Intelligence, Gc.
T
here are other examples of a mismatch between the grouping of tests according to common practice of neuropsychologists and how they would load upon factor analysis. As I mentioned earlier, this also occurs with the Attention/processing speed cognitive domain. Neuropsychologists commonly use both Digit Span Forward, and mental speed tests as markers of this cognitive domain. However, in Gf/Gc theory the Digit Span test would be a marker of the memory second order factor, SAR, which the mental speed tests would help define the general speed factor, Gs.
Obviously to the cognitive domains in neuropsychological practice MUCH MORE than in the results of the factor analysis ?. Neuropsychological and Psychopathological praxis requires an interpersonal interaction with the patient WHICH IS ESSENCE, while the other is based on "mere" statistical standards "blind and impersonal", NECESSARY ... BUT NOT ENOUGH !!
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India