Monday 19 May 2014

Processing speed and ageing: Elliot Tucker-Drob


Elliot’s presentation began with a very useful recapitulation of the relevant two-process models (slide 2). Baltes distinguished between Mechanics & Pragmatics; Cattell between Fluid & Crystallized intelligence; and Salthouse between Process & Product. I particularly like that last distinction, because the interest in Process is to find exactly how it leads to Product.



As the slide shows, we move from the mechanics of basic information processing, which is content poor, universal, biological and genetic; to the pragmatics of acquired knowledge, which is content rich, culture dependent and experience based. The pragmatics (please call it wisdom) hold up with age, the mechanics fade after the age of maturity (25) and decay thereafter. For all I know, orgasm frequency may follow the same trajectory. Processing is the feeder system, pragmatics the store of knowledge.


Slide 3 above spells out the concepts and puts them onto the Venn diagram.

Tucker-Drob’s personal view is: Processing Speed is a marker of “system integrity”  (that Ian Deary phrase again) i.e. an Endo-phenotype for cognitive ability and cognitive decline. It is largely unrelated to conventional forms of environmental variation like social class, schooling and culture. On the other hand, it is related to biologically-relevant conditions such as physical trauma, disease, health and nutrition. It is strongly linked with cognitive ability and cognitive decline It might be a fundamental cause of intelligence, but it does not have to be. It is simply “a less contaminated window into neurobiological integrity.”

All cognitive development shows an age effect. What is interesting about processing speed is there is little difference between those with little and those with much education, whereas those groups differ significantly in abstract reasoning and, as you would expect, academic knowledge.

You can see the whole presentation here:

In order to increase the cognitive load on you, the process of transferring the Powerpoint presentation has inverted and reversed a few of the slides. (If you succeed in righting them, let me know how).

Slide 10 on Woodcock Johnson subtests is a little hard to read without the subtest names, but the startlingly horizontal red line Gc is Comprehension, a well preserved skill. In contrast, the bright green line Gs which falls precipitously is for Perceptual Speed. So,there is the big contrast in one picture. The other subtests:  Blue Gsm = short term memory; yellow Gf = Fluid Reasoning; brownish Gv = Visuo-Spatial; lightish purple Glr = Associative Memory; darker purple Ga = Auditory Processing. There you are, the whole panoply of mental decay in one cheerful picture. You retain comprehension, which allows you to comprehend that you are slowing up fast.

Slide 11 makes equally sad viewing, but suggests that rather than assuming a whole set of different declines in specific abilities, it might be parsimonious to assume there is a common factor of decline. Note that Digit Symbol seems to show the sharpest decline, which might be an indication that it is a good measure of that common factor, whatever it is.

Slide 12 shows that half of the decline is global, and the rest is rather more domain specific than test specific.

Slide 14 show substantial age effects on both g and processing speed, with a smaller effect on episodic memory.

Slide 15 superimposes the test specific declines, showing the slightly sharper decline for digit symbol. Incidentally, this shows why giving a two minute digit symbol test would be of great use in all psychology research.

Slide 18 on a longitudinal sample shows that there have been significant changes in processing speed over a 7 year period, but not as pronounced as the other changes.

Slide 21 shows significant effect on processing speed, even though only a 3 year period of ageing has elapsed.

Slide 23 also shows a speed change, this time over 6 years.

Slide 25 also shows a speed change over 16 years in Sweden.

Slide 27 is a meta-analysis of 12,000+ subjects, which shows no difference between processing speed and other mechanics, which shows that processing speed goes down as much as the other mechanics measures go down.

Tucker-Drob then ends up with some general conclusions: A single, partially heritable, dimension underlies considerable proportions of individual differences in aging-related cognitive declines across different abilities. Processing Speed is strongly linked with this dimension.

The talk had an impact on the assembled company, in that it was a stirring defence of the importance of processing speed as the upstream process which led to the downstream product of cognitive ability. Naturally, it is couched in cautious terms, because we do not yet know what brain mechanisms are involved in simple processing speed tasks. Whatever those mechanism are, they are either important in themselves, or closely related to something important. Worth finding out, was the general reaction. With some animation, we went off to lunch in the next door room.

Sharp eyed observers with note that after that stirring call to arms Tucker-Drob ended up with instructions about how to bake a cake. This is probably some sort of Texan ritual.

On the other hand, it may be an example of the processing task of using a microwave cooker, and thus of more universal significance, together with a test of remembering to take your tablets, and trying to make sense of a long phone bill. Tasks of daily living, they call them, as they evaluate the capabilities of their wise elders. You decide what you think of these, but only after having someone test you on the tasks yourself.

1 comment:

  1. Lion of the Judah-sphere9 May 2015 at 03:52


    I once heard that men on average have significantly lower processing speeds compared to women, even after controlling for age and general intellectual ability. Then I looked it up on Google and found some research to back this up:

    Do you know why this may be? I had always thought that men did relatively better on non-verbal tests like processing speed, but it turns out I'm wrong.