A theory of executive control is presented that proposes that executive pro
cesses control subordinate processes by manipulating their parameters, reco
nfiguring them to respond in accord with the current task set. It adopts C.
Bundesen's (1990) theory of visual attention (TVA) and R. M. Nosofsky and
T. J. Palmeri's (1997) exemplar-based random walk (EBRW) as the theory of s
ubordinate processes. It assumes that a task set is a set of TVA and EBRW p
arameters sufficient to perform a task and that set switching involves chan
ging those parameters. The theory solves 2 computational problems that emer
ge in dual-task situations: the binding problem and the serial order proble
m. It can perform dual tasks in series or in parallel but prefers the seria
l strategy because it is faster and it solves the binding problem naturally
. The theory accounts for concurrence cost, set-switching cost, crosstalk b
etween tasks, and the modulation of crosstalk by task set.