Divided attention refers to the allocation of mental resources to multiple tasks or streams of information simultaneously. While we often feel capable of multitasking, research consistently shows that concurrent performance is limited — particularly when tasks compete for the same processing resources. Understanding divided attention has practical implications for driving, aviation, education, and any domain where people must manage multiple information streams.
Dual-Task Costs
The most basic finding in divided attention research is the dual-task cost: performance on each task is worse when done concurrently than when done alone. This cost is measured as the difference in speed or accuracy between single-task and dual-task conditions. The magnitude of dual-task costs depends on the difficulty of each task, the similarity between tasks, and the level of practice.
Bottleneck and Capacity Models
Two broad classes of theory explain dual-task limitations. Bottleneck models propose that certain processing stages — particularly response selection — can only handle one task at a time, creating a serial bottleneck. The psychological refractory period (PRP) effect demonstrates this: when two tasks must be performed in rapid succession, the response to the second task is delayed because response selection for the second must wait until response selection for the first is complete.
Capacity models, by contrast, propose a limited pool of processing resources that must be divided among concurrent tasks. Daniel Kahneman's (1973) capacity theory proposed a single undifferentiated pool, while Christopher Wickens' multiple resource theory proposed that resources are structured along dimensions of processing stage, modality, and code, with greater interference between tasks that share resources.
Modern research on media multitasking — simultaneously using multiple media devices — has found that heavy media multitaskers actually perform worse on laboratory measures of attention switching and distractor filtering, contrary to the intuition that practice should improve multitasking ability. This may reflect either self-selection (people prone to distraction are drawn to multitasking) or a genuine cognitive cost of habitual attention splitting. Regardless, the evidence consistently shows that human multitasking is far more limited than subjective experience suggests.
Automaticity and Practice
Dual-task costs can be dramatically reduced through practice when at least one task becomes automatized. Highly practiced skills require fewer attentional resources, freeing capacity for concurrent tasks. Schneider and Shiffrin's distinction between controlled processing (effortful, capacity-limited, serial) and automatic processing (effortless, unlimited capacity, parallel) explains why experienced drivers can hold a conversation while navigating familiar routes but struggle to do so in novel or demanding driving situations.
Neural Basis
Neuroimaging studies show that dual-task performance activates the prefrontal cortex more than either single task alone, reflecting the additional executive control demands of task coordination. The lateral prefrontal cortex appears to play a key role in maintaining and switching between task sets, while the anterior cingulate cortex monitors for conflict between competing task demands. These findings link divided attention to the broader construct of executive function.