Divided attention refers to the ability to process two or more tasks simultaneously. Interference between tasks is predicted by capacity theories (Kahneman), which posit a limited central resource, and by multiple-resource theories (Wickens), which posit modality-specific pools. Practice and automaticity reduce resource demands, allowing previously effortful processes to run in parallel without significant interference.
Try dual-task experiments such as tapping while reading and note when interference is high versus low. Distinguish tasks that share modalities from those that do not — cross-modal pairings tend to interfere less.
From your study of selective attention, you know that the cognitive system has filters and bottlenecks that limit what information reaches conscious processing. Divided attention takes the complementary question: when you must process two things *simultaneously*, what determines how well you can do it? The answer turns out to depend on what resources the two tasks require and whether they can be drawn from separate pools.
The single-resource model, developed by Kahneman in the 1970s, proposes a single undifferentiated pool of mental effort or capacity. On this view, any two tasks compete for the same limited supply — like two appliances sharing one electrical circuit. Total demand cannot exceed total capacity, so as one task increases in difficulty, the other suffers. This model predicts that any two tasks will interfere with each other, with worse performance as total demand rises. It explains why driving in heavy traffic makes it hard to maintain a conversation: both tasks are drawing from the same central pool.
Multiple-resource theory (Wickens, 1980s) offers a more nuanced account: attention is not one pool but several, organized along three dimensions — processing stage (perceptual/cognitive vs. response output), perceptual modality (auditory vs. visual), and response type (verbal vs. manual/spatial). Tasks that draw from the *same* resources interfere strongly; tasks that draw from *different* resources can be performed simultaneously with little cost. This explains why driving (visual spatial perception, manual response) interferes heavily with reading a sign (visual verbal perception) but interferes less with listening to the radio (auditory verbal processing). The prediction is that cross-modal, cross-code task pairings will show less dual-task interference than within-modal pairings.
Practice and automaticity fundamentally change the interference equation. A task that initially requires effortful controlled processing — consuming attentional resources and susceptible to interference — can, with extensive practice, become automatic: running with minimal resource demands, no longer requiring attention, and no longer susceptible to ordinary dual-task interference. This is how expert typists can sustain a conversation while typing, how musicians can improvise while reading a score, how drivers navigate familiar routes while thinking about something else entirely. Automaticity is the cognitive signature of expertise: the practice-driven transfer of processing from the effortful, capacity-limited controlled system to the efficient, capacity-free automatic system. However, automaticity is task-specific and fragile under truly novel demands — an expert driver suddenly facing an unexpected road hazard reinvokes controlled attention immediately.
The practical implications are significant. What people call multitasking is almost always rapid sequential task-switching rather than genuine parallel processing — the brain alternates attention between tasks at the cost of switching overhead, not truly handling both simultaneously. Cell phone use while driving is dangerous precisely because both involve cognitive-verbal processing (reasoning about a conversation) combined with spatial-manual processing, and the verbal component is not as separable from driving as people assume. Cognitive load in one domain consistently degrades performance in any other demanding domain, which is why pilots use checklists, surgeons minimize distractions, and interface designers minimize cognitive load at critical decision points. The research on divided attention is less about human limitation and more about understanding the architecture of those limits — and using that understanding to design tasks, tools, and training that work with rather than against cognitive capacity.