Although they are vastly outnumbered by simpler control systems,
complex, ``deliberative'' control systems have evolved on our
planet. Hence, agents with deliberative capabilities must have
adaptive advantages over agents with simpler control systems in some
environments. This paper examines the tradeoffs between the costs
of control systems and the benefits they offer in a variety of
environments and discusses the implications of these tradeoffs on
evolutionary trajectories from ``reactive''
control
systems, through somewhat more complex
``affective''
systems, to genuine deliberative
systems.