a reaction time device that shared as many physical and
response characteristics as possible with the comparison device
(Numbers; Cox et al., 1993) and demographically and cognitively
well-characterised samples of young, middle-aged and older
people. The individual age group samples were modest in size,
though the overall sample size was moderately large. The
Numbers device did not retain individual trial data, which
means that any extreme responses would not be excluded, and
that means were used as an individual's score. It would have
been ideal to be able to exclude any outlying responses, and also
to take account more fully of the reaction time distributions.
However, it was important that the present studywas conducted
using the procedures and parameters of the Numbers task that
have produced so much age- and intelligence-relevant findings,
and that the Lights task was similarly set up and used. The likely
effect of not having individual trial data is the addition of some
noise to the reaction time parameters, and some lowering of
effect sizes. Despite that, the results were clear and cross-device
effect sizes were large.
In the present study we argued that the Lights device is
less complex than the Numbers procedure, and that relations
with the Lights device may be less influenced by requirements
associated with processing and translation. Although this appears
to be reasonable, there are other differences between the
two devices, and the role of translation or other processing
requirements could be investigated in future research much
more directly within any one device. For example, instructions
across conditions could be varied, as in research on stimulus–
response compatibility effects. Our largely in-principle arguments
about the involvement of theoretical processes involved
in the tasks would be helped in future research with more
direct, experimentally-manipulated comparisons in which nearly
everything is the same except for a critical manipulation.
There will continue to be discussion about why people
differ in intelligence and what reaction time can tell us about
those differences. The present study at least eliminates some
higher-level cognitive confounders from the results obtained
from the device that has, to date, provided the strongest
evidence for that association.