107 / 218
105
Reward modulation of cognitive function: aging
(i.e. due to 160 vs. 240 trials) varied between studies (from €8.80 to €13.20). To account for
this confound, we validated the age-related effects in a group of participants who received
identical low and high reward amounts and the same number of trials (and thus identical
total amounts of reward). To this end, we directly compared performance of a subset of 68
participants (from study A, C and D, matched in terms of these factors,
table 5.1
) using an
independent samples Mann-Whitney U test (reporting the standardized test statistic, denoted
by U), with age (using a median split) as a between subject factor. We consider p < 0.05 as
significant.
Results
With age, overall accuracy and response times increased (Age, accuracy across trial-types: r
(ρ) = 0.531, p < 0.001; Age, RTs across trial-types: r (ρ) = 0.607, p < 0.001). In terms of the
SAT score, increasing age was associated with a decrease in speed over accuracy, so that older
participants traded speed for accuracy across the task as a whole (Age, SAT across trial-types:
r (ρ) = -0.636, p < 0.001).
Effects of reward, task-switching and their interaction
There was a main effect of Reward in terms of SAT: participants increased speed over accuracy
on high reward trials, compared with low reward trials (Reward: W = 6.519, p < 0.001). In
addition, there was a main effect of Task switching: Participants also increased speed over
accuracy on switch trials compared with repeat trials (Task switching: W = 2.241, p= 0.025).
Moreover, there was an interaction between Reward and Task switching (Reward x Task
switching: W = 6.452, p < 0.001), due to an increase in speed over accuracy on high versus
low reward
switch
trials (Reward effect on switch trials: W = 7.276, p < 0.001;
table S5.1
), but
a decrease in speed over accuracy on high versus low reward
repeat
trials (Reward effect on
repeat trials: W = 4.610, p < 0.001).
The effects in terms of SAT were due to a main effect of Reward in the response times, but
not accuracy: Participants responded faster on high reward compared with low reward trials
(Reward in terms of RTs: W = 6.519, p < 0.001; Reward in terms of accuracy: W = 1.656, p
= 0.098;
table S5.1
). In addition, there was a main effect of Task switching both in terms of
response times and accuracy: Participants responded more slowly on switch trials compared
with repeat trials (Task switching: W = 4.612, p < 0.001) and participants responded less
accurately on switch compared with repeat trials (Task switching: W = 7.205, p < 0.001).
There were no interactions between Reward and Task switching in terms of response times
(Reward x Task switching: W = 0.337, p = 0.7) or accuracy (Reward x Task switching: W =
0.168, p = 0.8).
In sum, we observed faster but equally accurate responding under high versus low reward,
which translated into a reward-related increase in the speed-over-accuracy score. In terms of