déclin cognitif.pdf

Aperçu du fichier PDF declin-cognitif.pdf - page 3/38

Page 1 2 34538

Aperçu texte

M. Ramscar et al. / Topics in Cognitive Science 6 (2014)


(Salthouse, 2011). Increasing reaction times are a primary marker for age-related cognitive decline (Deary, Johnson, & Starr, 2010) and are even considered its cause
(Salthouse, 1996), yet they are puzzling: Practice improves speed and performance on
individual cognitive tasks at all ages (Dew & Giovanello, 2010). Since we continually
practice using our cognitive capacities as we age, why does our performance on tests of
them decline?
We suggest that the answer to this question lies in the way that psychometric tests
neglect learning and its relationship to the statistical patterns that characterize human
experience. Learning is a discriminative process that serves to locally reduce the information processing demands associated with specific forms of knowledge and skill (Ramscar,
Yarlett, Dye, Denny, & Thorpe, 2010; Rescorla & Wagner, 1972). However, age and
experience will inevitably increase the overall range of knowledge and skills any individual possesses, increasing the amount information in (and complexity of) his or her cognitive systems. Processing all this extra information must inevitably have a cost (Shannon,
2.1. Learning and the long tail of experience
Statistically, the distribution of human experience is highly skewed: Much of our dayto-day life is fairly repetitive, involving a small repertoire of common occurrences, such
as reading the newspaper and going to work. At the same time, we encounter a far more
diverse range of infrequent or even unique occurrences (as Wittgenstein, 1953, noted, one
rarely reads the exact same newspaper twice). When data are distributed in this way,
comparisons of means are often meaningless (Baayen, 2001). Consider the problem of
recalling birthdays: We are usually reminded of the birthdays of family members on an
annual basis, and this usually makes us good at remembering them. However, as we
move through life, we learn about other birthdays. Sometimes we hear these dates only
once, such as when we attend a party for someone we barely know. As we learn each
new birthday, the mean exposure we have had to all the birthdates we know declines,
and the task of recalling a particular birthday becomes more complex. Accordingly, it
does not necessarily follow that someone who can recall 600 birthdays with 95% accuracy has a worse memory than someone who can recall just six with 99% accuracy.
Psychometric tests do not take account of the statistical skew of human experience, or
the way knowledge increases with experience. As a consequence, when these tests are
used to compare age groups, they paint a misleading picture of cognitive development.
This point can be demonstrated most clearly and effectively in relation to language: It is
a central and largely unique aspect of human cognition, and thanks to recent developments in machine information processing, its statistics are more readily and objectively
quantified than other aspects of experience. Moreover, almost all psychometric tests
involve some form of linguistic information processing: On any test in which subjects
have to comprehend verbal instructions and then refer to them in memory in order to perform a task, performance can be influenced by, and may even simply reflect, individual
differences in linguistic information loads.