Neurobiology of Aging 30 (2009) 521–524
Open peer commentary
Challenging the notion of an early-onset of cognitive decline
Lars-Göran Nilsson
a,d,e,f,*
, Ola Sternäng
a,d
, Michael Rönnlund
b,g
, Lars Nyberg
c,d,e,f
a
Department of Psychology, Stockholm University, Sweden
b
Department of Psychology, Umeå University, Sweden
c
Department of Radiation Sciences (Diagnostic Radiology) and Integrative Medical Biology (Physiology Section),
Umeå University, Sweden
d
Stockholm Brain Institute, Umeå University, Sweden
e
Nordic Center of Excellence, Umeå University, Sweden
f
Aging and Living Conditions, Umeå University, Sweden
g
Center for Population Studies, Umeå University, Sweden
Received 20 October 2008; accepted 6 November 2008
Abstract
Salthouse claims that cognitive aging starts around 20 years of age. The basis for this claim is cross-sectional data. He dismisses longitudinal
data, which typically show the cognitive decline to start much later, around 60 years of age. He states that longitudinal data cannot be trusted
because they are flawed. There is a confounding between the effects of maturation and retest effects. We challenge Salthouse’s strong claim
on four accounts.
© 2009 Elsevier Inc. All rights reserved.
Keywords: Cognition; Cross-sectional design; Longitudinal design; Neurobiology
The present paper challenges Salthouseˇ ıs strong claim
about an early cognitive decline on the following four
grounds. First, development from childhood to mature
age is a matter of individual development and cognitive
change through different passages throughout the life span.
Conceptually, we believe, statements about cognitive decline
are highly related to cognitive change as seen in within-
person comparisons. Second, both cross-sectional and
longitudinal data are flawed if proper control is not applied.
For cross-sectional data there is always a risk of a confound
between true age effects and cohort effects. For longitudinal
data, there is always a risk that retest effects are confounded
with the true age effects. Various means have been suggested
to control for cohort effects in cross-sectional data and for
retest effects in longitudinal data. Third, Salthouse uses
selective evidence from neurobiology and animal studies
*
Corresponding author at: Department of Psychology, Stockholm Uni-
versity, SE-10691 Stockholm, Sweden. Tel.: +46 8 163940;
fax: +46 8 159342.
E-mail address: lgn@psychology.su.se (L.-G. Nilsson).
to support his claim of this early onset of cognitive decline.
Fourth, Salthouse is talking about cognitive functioning as if
this is a unitary entity. We will deal with these four issues in
turn.
1. Conceptually about cognitive decline
It seems obvious that cognitive decline, in whatever age,
is a matter of cognitive change. It seems self-evident to claim
that within-person comparisons (as in longitudinal data) are
more adequate to make than between-person comparisons
(as in all cross-sectional data) when making claims about
cognitive change of an individual. Salthouse seems to be
of the same opinion as he writes in the last paragraph of
the paper, “... only with longitudinal data can one examine
within-individual changes distinct from between-person dif-
ferences”. In light of this citation, the overall claim of the
paper that cross-sectional data are to be preferred is intel-
lectually confusing. The citation begs the question at the
conceptual level of why the within-individual comparisons
0197-4580/$ – see front matter © 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.neurobiolaging.2008.11.013