Research on multitasking harkens back to the beginnings of cognitive psychology. The central question has always been how we manage to perform multiple actions at the same time. Here, we highlight the role of specific inputand output-modalities involved in coordinating multiple action demands (i.e., crossmodal action). For a long time, modalityand content-blind models of multitasking have dominated theory, but a variety of recent findings indicate that modalities and content substantially determine performance. Typically, the term ‘‘input modality’’ refers to sensory channels (e.g., visual input is treated differently from auditory input), and the term ‘‘output modality’’ is closely associated with effector systems (e.g., hand vs. foot movements). However, this definition may be too narrow. The term ‘‘input modality’’ sometimes refers to a dimension within a sensory channel (e.g., shape/color in vision). Furthermore, the linkage between output-modalities and effector systems may not be specific enough to illuminate some notorious twilight zones (e.g., to distinguish between hand and wrist movements). As a consequence, we will use ‘‘modality’’ as an umbrella term here to capture various sources of stimulus variability used to differentiate the task-relevant information and sources of motor variability used to differentiate responses. Many of the pioneering studies involved the observation of dual-task performance in two continuous tasks that typically consisted of complex action sequences (e.g., reading and writing, see Solomons & Stein, 1896; Spelke, Hirst, & Neisser, 1976). However, it soon became apparent that tighter experimental control was necessary to pinpoint the specific cognitive mechanisms supporting multitasking. The PRP paradigm: an experimental breakthrough. The development of the psychological refractory period (PRP) paradigm (Telford, 1931; Welford, 1952) provided a methodological breakthrough that allowed researchers to exactly control the flow of information in both tasks. The PRP paradigm involves two elementary tasks with a limited set of clearly defined stimuli and responses. The mechanisms underlying multitasking are studied by systematically manipulating the temporal overlap of the two tasks, which is achieved by varying the delay between the presentations of the stimuli for the two tasks (stimulus onset asynchrony, SOA). The PRP effect refers to the typical finding that reaction times (RTs) for the second task increase with decreasing SOA, an effect that has been replicated in numerous studies with a variety of stimulus and response modalities (see Bertelson, 1966; Pashler, 1994; Smith, 1967). The RSB model: a powerful explanatory concept? The most influential and elegant account of the PRP effect has been the response selection bottleneck (RSB) model (Telford, 1931; Welford, 1952). A starting assumption of the RSB model is that tasks at hand can be divided into three successive cognitive processing steps, namely perceptual processing (i.e., stimulus encoding/categorization), response selection (i.e., deciding which response corresponds to the stimulus according to the task rules), and response execution processes. In a number of experiments, the duration of each of these processing stages was systematically manipulated for each of the two tasks (see Pashler, 1994). As a result, the most convincing hypothesis to accommodate the corresponding findings was the assumption that perceptual processing and response L. Huestegge (&) RWTH Aachen University, Aachen, Germany e-mail: lynn.huestegge@psych.rwth-aachen.de