Multiple Resource Theory

Multiple resource theory is a theory of attention and dual-task performance developed by the engineering psychologist Christopher D. Wickens. It holds that human attention is not drawn from a single, general-purpose pool but from several distinct pools of processing resources, each tied to a particular kind of mental operation. Because of this structure, the difficulty of doing two things at once depends not only on how demanding each task is, but on whether the two tasks compete for the same resources. Two activities that draw on different pools can often be combined with little loss, whereas two that draw on the same pool interfere sharply even when neither is especially hard (Wickens, 2002).

The theory grew out of the resource, or capacity, tradition in attention research and was offered as a refinement of it. Where earlier accounts described attention as a single reservoir of mental effort, multiple resource theory kept the idea of limited capacity but argued that capacity is partitioned along a small number of dimensions (Wickens, 1984). Its most influential form, the four-dimensional multiple resource model, specifies those dimensions and pairs them with a computational "conflict matrix" that predicts how much two tasks will interfere from the degree to which their resource demands overlap (Wickens, 2002; Wickens, 2008).

Multiple resource theory is primarily a functional theory: it defines resources by the processing they support rather than by specific brain regions. It has been unusually productive in applied settings — aviation, driving, and health care among them — because its predictions translate directly into guidance for designing displays and tasks that spread demand across resources rather than overloading any one of them (Wickens, 2008; Wickens et al., 2022).

Historical Development

The theory belongs to a line of work that recast attention as a limited resource to be allocated rather than a gate that admits one channel at a time. A key early distinction came from Norman and Bobrow, who separated performance limits caused by a shortage of processing resources ("resource-limited" processes) from those caused by poor-quality input ("data-limited" processes) — establishing the vocabulary of resources on which later theories would build (Norman & Bobrow, 1975).

The single-resource version of this idea, in which one undifferentiated pool of effort is allocated across tasks, was given its most influential statement by Daniel Kahneman (Kahneman, 1973). Almost immediately, evidence began to strain it: some pairs of tasks interfered far more than others of equal difficulty, which a single pool cannot easily explain. Working independently, David Navon and Daniel Gopher proposed that the processing system is better understood as a set of distinct mechanisms, each with its own capacity, and introduced tools such as the performance operating characteristic for charting how two time-shared tasks trade off against each other (Navon & Gopher, 1979).

Wickens drew these threads together into a structured, predictive theory. He first set out the case that attentional resources are multiple rather than unitary (Wickens, 1980), then organized the resources along defined dimensions in his foundational chapter "Processing resources in attention" (Wickens, 1984). Over the following decades he consolidated the framework into the four-dimensional model and its computational conflict matrix (Wickens, 2002) and reviewed half a century of supporting evidence in a widely cited synthesis (Wickens, 2008).

The Four Resource Dimensions

The four-dimensional multiple resource model proposes that resources are separated along four dichotomous dimensions. Two tasks draw on more nearly separate resources — and therefore interfere less — the fewer of these dimensions they share (Wickens, 2002).

Dimension	Separable levels	What it distinguishes
Processing stages	Perceptual/cognitive vs. response	Perceiving and thinking draw on different resources than selecting and executing responses
Perceptual modalities	Visual vs. auditory	Seeing and hearing draw on separate pools
Processing codes	Spatial vs. verbal	Spatial/analog processing is separate from verbal/linguistic processing, at both the cognitive and the response level
Visual channels	Focal vs. ambient	Within vision, detailed foveal processing is separate from peripheral, ambient processing

The stage dimension separates the resources used for perception and central cognition from those used to select and carry out responses; a demanding perceptual task and a demanding response task can overlap relatively well because they tap different stages (Wickens, 1984). The modality dimension captures the familiar observation that it is easier to monitor a visual display while listening than while watching a second visual display (Wickens, 2002). The code dimension distinguishes spatial from verbal processing and applies across stages: spatial and verbal working memory are partly separate, as are manual (spatial) and vocal (verbal) responses (Wickens, 1984). The visual channel dimension, the most recent addition, separates focal vision — used for object recognition and reading detail — from ambient vision, used for orientation and movement through space (Wickens, 2002).

Predicting Dual-Task Interference

The practical heart of the theory is its claim that interference between two tasks is a joint function of two things: the demand each task places on resources, and the overlap in the specific resources the two tasks require. Holding difficulty constant, two tasks that share resource dimensions will interfere more than two that do not (Wickens, 2002).

To make this quantitative, Wickens rendered the model as a computational conflict matrix. Each task is coded for its demands along the four dimensions; for any pair of tasks, the model derives a conflict value reflecting how much their demands coincide, and combines this with the tasks' overall demand to predict the size of the dual-task decrement. The result is a tool that does not merely describe interference after the fact but predicts, in advance, which task combinations will prove costly and which will be relatively safe — and therefore which display and interface designs will overload an operator (Wickens, 2002; Wickens, 2008).

This predictive use is what distinguishes multiple resource theory from a purely descriptive account. A designer can represent a proposed cockpit, vehicle interior, or control room as a set of concurrent tasks, code their resource demands, and use the model to estimate where overload will occur before any hardware is built (Wickens, 2008).

Try It Yourself

The interactive experiment below is a short dual-task study you run on yourself. You hold a memory load — a pattern of squares (spatial) or a set of letters (verbal) — while performing a second task that is also either spatial or verbal, and it measures how much each combination degrades your memory. Multiple resource theory predicts a crossover: a task interferes more with a memory load that shares its processing code. It takes about 12–15 minutes, generates all its own materials, and uploads nothing.

The SEEV Model and Visual Attention

A natural extension of the theory concerns not how two tasks interfere but how a person distributes limited visual attention among many sources of information. The SEEV model — named for its four determinants, salience, effort, expectancy, and value — predicts where attention will be directed across a visual workspace. Salience and effort are bottom-up and cost factors; expectancy (how often useful information appears at a location) and value (how important that information is) are the top-down drivers that dominate skilled performance (Wickens et al., 2003).

The model has performed strikingly well as a quantitative predictor of visual scanning. In a study of instrument-rated pilots flying a high-fidelity simulation, an expected-value model of selective attention accounted for 94% of the variance in scanning behavior, with 90% in a second validation (Wickens et al., 2003). Applied to driving, the SEEV model predicted roughly 95% of the variance in drivers' visual scanning across two experiments that manipulated the information bandwidth and priority of in-vehicle tasks (Horrey et al., 2006).

Empirical Support and Limitations

Across the evidence reviewed by Wickens, the four dimensions are not equally well established. The modality distinction has the strongest and most consistent support, and the code distinction is also well supported; the theory's account of interference has held up best where these dimensions are concerned (Wickens, 2008). Reviews also note that the resource dimensions are associated with distinct neural substrates, a point developed in the section on key structures below (Wickens, 2008).

The theory has nonetheless drawn substantive criticism, some of it from within the resource tradition itself. David Navon — a co-originator of the multiple-capacity view — later questioned whether "resources" had become a theoretical soup stone: a construct that can be invoked after the fact to explain almost any pattern of interference, but that is difficult to measure independently of the interference it is meant to explain (Navon, 1984). Navon argued that some dual-task interference attributed to shared resources may instead reflect outcome conflict or crosstalk — the output of one task disrupting the processing of the other — rather than competition for a common pool (Navon, 1984). The force of these critiques is methodological as much as theoretical: demonstrating that interference is genuinely due to shared resources requires careful control of both task difficulty and task priority, and the ruling-out of alternative mechanisms, before resource overlap can be credited as the cause.

Comparison With Other Theories of Attention

Multiple resource theory is best understood against the two main alternatives it was weighed against.

The single-resource (capacity) model holds that there is one undifferentiated pool of attentional capacity, flexibly allocated across tasks (Kahneman, 1973). Multiple resource theory retains the language of capacity but rejects the claim of a single pool, on the grounds that a unitary capacity predicts that any two tasks of equal combined difficulty should interfere equally — which they demonstrably do not (Wickens, 2002).

Bottleneck theories locate dual-task interference not in graded resource sharing but in a structural limit: a central stage, typically response selection, that can process only one task at a time, forcing the second to wait (Pashler, 1994). The two kinds of account are not strictly incompatible. Bottleneck effects are most evident in simple, speeded reaction-time tasks performed in close temporal succession, whereas resource effects are most evident in continuous, complex tasks performed concurrently; each framework captures part of the phenomenon of divided attention (Pashler, 1994; Wickens, 2008).

Applications in Human Factors

Multiple resource theory has had its largest impact outside the laboratory, where its predictions guide the design of tasks and displays in demanding, safety-critical settings (Wickens et al., 2022).

Aviation

In the cockpit, the theory and the SEEV model inform how flight information should be distributed across modalities and displays so that pilots can monitor instruments, scan for traffic, and communicate without overloading any single resource. Modeling of multitask pilot performance showed that auditory and visual presentations of the same information impose different costs, and that those costs and the resulting scanning patterns can be predicted by multiple-resource and attention-allocation models (Wickens et al., 2003).

Driving

The theory provides a principled explanation for why some secondary activities are far more dangerous behind the wheel than others. A meta-analysis of 23 studies found significant costs of cell-phone use on drivers' responses to external events, and — consistent with the modality and code dimensions — that hands-free phones did not eliminate or substantially reduce those costs, because the interference arises from shared cognitive and, for texting, visual-manual resources rather than from holding a handset (Horrey & Wickens, 2006). The SEEV model further predicts how drivers allocate their gaze between the road and in-vehicle tasks (Horrey et al., 2006).

Health Care and Automation

More recent work extends the theory to health care, human interaction with automation and artificial intelligence, and immersive technologies such as virtual and augmented reality, where the same logic — spreading demand across resources and designing for how attention is actually allocated — applies to clinicians, operators, and users of new display media (Wickens et al., 2022).

Key Researchers

The researchers below have shaped multiple resource theory and the resource-based account of attention it belongs to. The list begins with the theory's originator and the co-founders of the multiple-resource idea, continues with the contributors most responsible for its computational and applied development, and closes with the two figures whose competing theories define the contrasts against which it is usually assessed. Deceased researchers are listed with dates only.

• Christopher D. Wickens — Research Professor of Cognitive Psychology, Colorado State University, and Professor Emeritus of Psychology and Aviation, University of Illinois at Urbana-Champaign. Originated multiple resource theory and developed its four-dimensional structure, the computational conflict matrix that predicts dual-task interference from resource overlap, and the SEEV model of visual attention allocation (Wickens, 1984; Wickens, 2002; Wickens, 2008).
  Faculty
• David Navon — Professor Emeritus, Department of Psychology, University of Haifa. With Daniel Gopher, originated the multiple-capacity view that the processing system comprises several distinct resources rather than one pool, and introduced the performance operating characteristic for analyzing how time-shared tasks trade off — the conceptual groundwork the four-dimensional model later formalized. He also offered an influential critique of the resource construct itself (Navon & Gopher, 1979; Navon, 1984).
  Google Scholar · Faculty
• Daniel Gopher — Professor Emeritus of Cognitive Psychology and Human Factors Engineering, Technion – Israel Institute of Technology. Co-originated the multiple-capacity model and supplied early evidence for separable resources by showing that different difficulty manipulations trade off differently against task emphasis; he later became a leader in attention-strategy and skill training. His doctoral work was supervised by Daniel Kahneman — so the originator of the single-resource view trained one of the founders of the multiple-resource alternative (Navon & Gopher, 1979).
  Google Scholar · Faculty
• Jason S. McCarley — Professor, School of Psychological Science, Oregon State University. Co-developed the computational SEEV and visual-scanning modeling that underpins the modern, quantitative form of multiple resource theory, and co-authored the field's standard graduate text, Applied Attention Theory (Wickens et al., 2022).
  Google Scholar · Faculty
• William J. Horrey — Senior Research Scientist and Group Leader, AAA Foundation for Traffic Safety; Ph.D. in Engineering Psychology, University of Illinois at Urbana-Champaign. Applied multiple resource theory to driver distraction, including the meta-analytic test showing that visual-manual in-vehicle tasks interfere with driving far more than auditory-vocal conversation, and developed SEEV models of drivers' visual attention (Horrey & Wickens, 2006; Horrey et al., 2006).
  Google Scholar · Faculty
• Daniel Kahneman (1934–2024) — Was Eugene Higgins Professor of Psychology, Emeritus, Princeton University, and the 2002 Nobel laureate in Economic Sciences. In Attention and Effort (1973) he developed the most influential formulation of single-resource (capacity) theory, describing attention as a single, limited reservoir of mental effort, flexibly allocated across competing tasks according to an allocation policy shaped by arousal and intention — the unitary-resource model that multiple resource theory was built to refine and challenge (Kahneman, 1973).
• Harold Pashler — Distinguished Professor of Psychology, University of California, San Diego. Developed the central-bottleneck account of dual-task interference, attributing it to serial processing at a central response-selection stage rather than to graded resource sharing — the principal modern alternative to resource theories, and, with Kahneman's single-resource model, one of the two contrasts against which multiple resource theory is weighed (Pashler, 1994).
  Google Scholar · Faculty

Key Structures

Multiple resource theory was derived from behavioral patterns of dual-task interference rather than from neuroanatomy, but its resource dimensions have since been linked to distinct neural systems, and Wickens has argued that this neurophysiological plausibility lends the theory additional support (Wickens, 2008). The brain structures most closely associated with its dimensions are the following.

Cerebral hemispheres — the spatial and verbal codes. The distinction between verbal and spatial processing codes parallels one of the best-established findings in neuroscience: the left hemisphere is typically dominant for language and verbal processing, the right hemisphere for spatial processing. Functional imaging of tasks that differ in code — for instance, auditory sentence comprehension versus the mental rotation of three-dimensional objects — reveals activation in largely nonoverlapping cortical networks, consistent with the separability the code dimension predicts (Just et al., 2001).

Modality-specific sensory cortices — the auditory and visual modalities. Auditory and visual perception are served by anatomically distinct systems: the auditory cortex of the superior temporal lobe and the visual cortex of the occipital lobe. The separation of auditory from visual resources in the theory corresponds to this physical separation of the sensory pathways (Wickens, 2008).

Frontal and prefrontal cortex — central and response resources. Response selection, executive control, and the active maintenance of information in working memory recruit frontal and prefrontal regions. These areas figure centrally in the stage dimension's separation of perceptual and cognitive processing from response processing, and in the management of competing demands when two tasks must be performed at once (Just et al., 2001; Wickens, 2008).

A crucial qualification comes from the neuroimaging evidence itself. When two tasks that recruit largely nonoverlapping cortical systems — such as auditory comprehension and visual mental rotation — are carried out simultaneously, the activation in each system is reduced relative to performing that task alone. The nonoverlapping systems are therefore not fully independent; they appear to share some common capacity, which is the sense in which the cortical systems are interdependent (Just et al., 2001). This pattern both supports the theory's central premise — that different kinds of processing rest on separable neural resources — and tempers it, by showing that even anatomically distinct systems are subject to a shared limit.

These associations were identified largely through neuroimaging conducted after the theory was formulated, and are best understood as neural correlates of its functional dimensions rather than as its foundation.

References

1	Horrey, W. J., & Wickens, C. D. (2006). Examining the impact of cell phone conversations on driving using meta-analytic techniques. Human Factors, 48(1), 196–205. https://doi.org/10.1518/001872006776412135
2	Horrey, W. J., Wickens, C. D., & Consalus, K. P. (2006). Modeling drivers' visual attention allocation while interacting with in-vehicle technologies. Journal of Experimental Psychology: Applied, 12(2), 67–78. https://doi.org/10.1037/1076-898X.12.2.67
3	Just, M. A., Carpenter, P. A., Keller, T. A., Emery, L., Zajac, H., & Thulborn, K. R. (2001). Interdependence of nonoverlapping cortical systems in dual cognitive tasks. NeuroImage, 14(2), 417–426. https://doi.org/10.1006/nimg.2001.0826
4	Kahneman, D. (1973). Attention and effort. Prentice-Hall.
5	Navon, D. (1984). Resources—a theoretical soup stone? Psychological Review, 91(2), 216–234. https://doi.org/10.1037/0033-295X.91.2.216
6	Navon, D., & Gopher, D. (1979). On the economy of the human processing system. Psychological Review, 86(3), 214–255. https://doi.org/10.1037/0033-295X.86.3.214
7	Norman, D. A., & Bobrow, D. G. (1975). On data-limited and resource-limited processes. Cognitive Psychology, 7(1), 44–64. https://doi.org/10.1016/0010-0285(75)90004-3
8	Pashler, H. (1994). Dual-task interference in simple tasks: Data and theory. Psychological Bulletin, 116(2), 220–244. https://doi.org/10.1037/0033-2909.116.2.220
9	Wickens, C. D. (1980). The structure of attentional resources. In R. S. Nickerson (Ed.), Attention and performance VIII (pp. 239–257). Lawrence Erlbaum.
10	Wickens, C. D. (1984). Processing resources in attention. In R. Parasuraman & D. R. Davies (Eds.), Varieties of attention (pp. 63–101). Academic Press.
11	Wickens, C. D. (2002). Multiple resources and performance prediction. Theoretical Issues in Ergonomics Science, 3(2), 159–177. https://doi.org/10.1080/14639220210123806
12	Wickens, C. D. (2008). Multiple resources and mental workload. Human Factors, 50(3), 449–455. https://doi.org/10.1518/001872008X288394
13	Wickens, C. D., Goh, J., Helleberg, J., Horrey, W. J., & Talleur, D. A. (2003). Attentional models of multitask pilot performance using advanced display technology. Human Factors, 45(3), 360–380. https://doi.org/10.1518/hfes.45.3.360.27250
14	Wickens, C. D., McCarley, J. S., & Gutzwiller, R. S. (2022). Applied attention theory (2nd ed.). CRC Press. https://doi.org/10.1201/9781003081579