Recently we have been employing a task cuing paradigm to investigate task expectancy effects on real-time task processing. Participants are cued in advance to expect either a lexical decision task or a magnitude comparison task between two digits. Responses to both tasks are slower on invalidly cued trials than on validly cued trials. We also find that cuing effects are underadditive with effects of symbolic distance in the magnitude comparison task, suggesting that the operations needed to recover from an incorrect task set create a processing bottleneck. How similar is this task-set bottleneck to the bottleneck that has been hypothesized to cause Task 2 slowing at short stimulus-onset-asynchronies (SOAs) in the psychological refractory period (PRP) paradigm? To find out, we conducted several PRP studies in which Task 2 was magnitude comparison and Task 1 was a 2AFC tone frequency discrimination. Symbolic distance effects in the magnitude comparison task were additive with SOA effects in the standard PRP preparation. In an effort to bring the PRP and the task-cuing paradigm closer together procedurally, we then conducted a Go No-Go version of the PRP paradigm, where responses are withheld to one of the two Task 1 stimuli. Symbolic distance effects were additive with SOA effects even on No-Go trials. We then brought the two paradigms even closer together in a new study that omitted the Task 1 stimulus on a small proportion of the trials. Subjects were slower to perform the magnitude comparison task on these withholding trials than they were on standard dual-task trials with a long SOA. That is, just the expectancy of having to perform Task 1 slowed performance of the magnitude comparison task. However, symbolic-distance effects were still additive with the slowdown. Apparently, single and dual task expectancies generate processing bottlenecks, but the bottlenecks are not the same.