See discussions, stats, and author profiles for this publication at: [Link]

net/publication/261180964

Attention and multitasking

Chapter · August 2014

DOI: 10.1037/14528-017

CITATIONS READS
4 7,551

2 authors:

Roger Remington Shayne Loft

The University of Queensland The University of Western Australia
228 PUBLICATIONS 12,440 CITATIONS 146 PUBLICATIONS 2,977 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Shayne Loft on 16 July 2015.

The user has requested enhancement of the downloaded file.

 Chapter 17

Attention and Multitasking

Roger W. Remington and Shayne Loft

In his seminal work, The Principles of Psychology, attend suggests that failures of attention result from
William James (1918) famously wrote that “Everyone failures of motivation or a breakdown in mental
knows what attention is. It is the taking possession discipline. However, consider the case of Christo-
by the mind, in clear and vivid form, of one out of pher Hill, who on September 9, 2008, ran a red light
what seem several simultaneously possible objects or while texting a neighbor. He was going 45 mph
trains of thought” (pp. 403–404). His latter sentence when he collided with a vehicle driven by Linda
contains two critical observations. The first is that Hoyle, who was pronounced dead shortly thereafter.
attention is inherently a process of selection and loss: By all accounts, Hill was a conscientious person who
As knowledge about a small bit of the world is ampli- took pride in his clean driving record. When asked
fied, information about other unattended objects is by a policeman at the accident scene what color the
attenuated or lost. This selection process is necessi- light had been, Mr. Hill responded that he never saw
tated by the limited capacity of our cognitive system, it (Richtel, 2009). A number of similar reports high-
and the loss of information plays a key role in under- light the problems of using cell phones and texting
standing how failures of attention occur even for while driving. Yet, people continue with the prac-
experienced, conscientious operators in safety-critical tice, and accidents continue to occur. Despite using
environments. The second observation is that attend- a headset that leaves the hands free during the con-
ing is the act by which we become consciously aware versation, a person’s attention can still be taken
of events in the world that guide our behavior. Unat- away from driving. Problems with the management
tended events do not hold the same consequences for of attention are not confined to driving. A review of
our interpretation of the world. The practical signifi- air traffic control and flight crew errors shows a high
cance of this is that the proper management of atten- prevalence of failures of attention (Dismukes &
tion is central to situation awareness in operational Nowinski, 2007; Jones & Endsley, 1996; Shorrock,
environments (see Endsley, 1995). 2007). We can begin to understand the origin of
James also asserted that we have an intuitive these failures only when we have a clearer picture of
understanding of attention. To an extent this is true. how the task that we are doing influences the infor-
If admonished to “pay attention” or “keep alert,” we mation we take in from the world and why it is that
generally have little trouble understanding what is we do not attend to critical tasks when needed.
meant by this or of how to comply. The assumption In this chapter, we present a sketch of how atten-
that attending is simply the result of the intention to tion works within our cognitive system. By probing

This work was supported by University of Queensland Vice-Chancellor’s Senior Research Fellowship 2011001759 to R. W. Remington; Australian
Research Council Discovery Grant DP120103721 to R. W. Remington, S. Becker, C. Folk, and S. Yantis; and Australian Research Council Discovery
Grant DP120103111 to S. Loft and R. E. Smith.

[Link]
APA Handbook of Human Systems Integration, D. A. Boehm-Davis, F. T. Durso, and J. D. Lee (Editors-in-Chief)
261
Copyright © 2015 by the American Psychological Association. All rights reserved.
Remington and Loft

the mechanisms of attention, of which we are not situation awareness (e.g., Durso & Sethumadhavan,
by and large conscious, we illustrate how the inter- 2008; also see Chapter 19 of this handbook) and
action of attention with ongoing cognitive process- contribute significantly to our assessment of
ing predisposes us to specific types of errors under workload.
certain circumstances, even with the best of inten- When we use the term attention in everyday
tions. Our goal is not to provide guidelines on how speech, usually we are referring to the contents of
to “design for attention.” As much as we would like awareness, as when attending to the cell phone and
to do that, exact prescriptions will depend on the not the road. The study of attention, however, is
specific circumstances. Indeed, the current Depart- more concerned with how those contents came to be
ment of Defense design document, MIL-STD-1472 selected for awareness. Here, we consider selection
(U.S. Department of Defense, 1999), specifies goals with respect to three types of attention that influence
for attention management—for example, minimize the goal-directed behavior characteristic of human-
distraction by providing only essential information system interaction: input selection, which orients
(Section [Link].7) and provide salient cuing processing to specific sources of information in the
(Section [Link].8)—but does not specify how to world; executive control, which manages ongoing
achieve these goals. It also recommends using flash- tasks; and alerting, which interrupts ongoing tasks to
ing (Section [Link].2) or color (Section [Link].1) orient attention to new sources of information.
without specifying under what conditions flashing The mental operations associated with carrying
or color will be effective or considering conditions out goal-directed behavior in applied settings—
under which it will not. Answers to these questions whether one is driving, piloting, controlling air traf-
would require a detailed analysis of the tasks, proce- fic, or performing surgery—can be conceived of as a
dures, and consequences of errors or delays in cycle. Operators gather information from the exter-
action in the context of application. We feel that a nal world that accumulates to form a picture of the
broad understanding of the mechanisms of attention current situation. This picture is compared to an
will better assist the human systems engineer in intended or expected state of the world, and from
considering the implications of design decisions for this the operator selects actions that will reduce the
the management of attention. difference between actual and intended world states.
Figure 17.1 illustrates the role of attention in the
cycle of goal-directed processing characteristic of
ATTENTION AND HUMAN-SYSTEM
operational settings. The operator selects relevant
INTERACTION
information from the sensory representations of dis-
Attention is the name given to a collection of inter- played data and assesses this sensory input in work-
acting cognitive mechanisms that regulate mental ing memory with respect to ongoing tasks, goals,
activity to determine what we are aware of in the and expectations, updating the world representation
world around us. Although our sensory systems can in working memory as necessary. The operator then
process vast amounts of information in parallel, our elects to continue acquiring more information about
cognitive system is much more limited in its ability the current goal, switch goals, or make a response
to make use of this information. These processing that will change the state of the system. In Figure
limitations affect our ability to understand the impli- 17.1, the loop connecting these activities is enclosed
cations of events in the world with respect to our within the gray region to denote that they share the
current situation, as well as update our situational same limited-capacity central processing resources.
knowledge, select appropriate responses, and man- The world representation in working memory
age multiple concurrent tasks. These are functions sets input selection to sources of information whose
associated with the central processor, whose activi- properties match those of the relevant, active task
ties are closely associated with conscious awareness. goal. We refer to this as top-down control to denote
It is largely the allocation of central processing that the selective allocation of attention is driven by
resources to tasks and stimuli that determine our tasks and goals. By task we mean an operational

262
 Attention and Multitasking

(pre-attentive WORKING MEMORY

sensory input) • Goals & Tasks
(task/goal
(top-down set) • Expectations
selection)
• Recent History

EXECUTIVE
CONTROL
(bottom-up
activation)
INPUT Situation Assessment Action
Sensory Execution
SELECTION Action Planning
Processing
LIMITED-CAPACITY
CENTRAL PROCESSING

FIGURE 17.1. Schematic overview of the role of attention in regulating cognitive processing in
goal-oriented processing. Input selection filters sensory input according to relevant features specified
by the goals, tasks, expectations, and recent history stored in working memory. Information accumu-
lates on more active goals until the operator has assessed the situation and has selected a response.
Executive control manages tasks and goals, including suspending and switching from one task to
another. Situation assessment, action planning, and executive control all require limited-capacity
central processing resources. Some preattentive sensory information is available in working memory
and can function to capture attention and to switch the source of input selection.

objective, such as the procedure for the take-off or selection over time is evident in the pattern of eye
landing of an aircraft or the air traffic control proce- fixations on aircraft icons and associated data
dure for accepting an aircraft into a sector. By goal blocks. The aircraft data to be selected next by input
we mean a more specific, low-level step in a task, selection will be determined by recent updates of the
such as increasing thrust or reading the altitude of situation. Each fixation represents a momentary
an aircraft on a display. Input selection can also be allocation of attention to that aircraft symbol or data
driven by highly salient sensory data that capture block, to the exclusion of others. Moreover, atten-
attention, alerting the operator to potentially impor- tion must be tightly focused to read alphanumeric
tant new information. We refer to this as bottom-up strings or to make precise visual discriminations
control to denote that the allocation of attention is (Reichle, Pollatsek, Fisher, & Rayner, 1998). Each
driven by external events. of these individual momentary attentional episodes
The processing cycle in Figure 17.1 illustrates will last between a quarter and half a second, during
how our awareness of the state of the external world which period of time other goals demanding atten-
is codetermined by three interacting factors: top- tion will be postponed. Over the course of seconds
down orienting of input selection to goal-relevant or tens of seconds, goals and tasks will change so
information sources, policies at the level of execu- that information for several aircraft will have been
tive control that determine the active tasks and sampled and attention effectively shared between
goals, and bottom-up orienting of input selection in stimuli, each receiving a proportion of central pro-
alerting the operator to a source of new information. cessing resource. Aircraft symbols and data that do
Consider the case of an air traffic controller moni- not conform to goals and expectations will be sam-
toring the radar screen to assess the progress of air- pled less frequently.
craft with respect to flight plans and maintenance of Other tasks must be interleaved with the task of
safe separation. The spatial allocation of input monitoring the radar screen. Periodically, a signal

263
Remington and Loft

for a different air traffic control task will alert the information, a process referred to as spatial attention.
operator to a new task. A blinking aircraft icon The pattern of eye fixations when searching air traf-
would signal an aircraft ready for acceptance into fic control displays, for example, is a useful, but not
the sector, a symbol that turns red could signal a perfect, overt indication of the sequential allocation
potential conflict, or an aircraft could request a of spatial attention. In addition, input selection can
course or altitude change over the radio. The sud- be set for task-relevant visual features, such as the
den appearance of the new event will interrupt mon- color of an object or its orientation, to restrict visual
itoring, reorient attention to the new input, and search only to objects containing that feature; this is
mechanisms of executive control will establish new referred to as feature-based attention. Attention can
behavioral goals. If two or more new events occur also be directed at an object even when its spatial
close together in time—say, a blinking aircraft icon location overlaps that of other objects, which is
at the same time as the radio message—executive referred to as object attention. The same three bases
control will set priorities for the two tasks that will of selection—spatial location, features, and objects—
determine the proportion of limited capacity central also characterize attentional selection in audition.
processing to devote to each task over time. Central Auditory features include properties such as pitch or
processing will be needed to comprehend the speech timbre, with objects comprised of combinations of
message and choose an action, and also to assess the pitch, timbre, and time-varying patterns (e.g., police
acceptance request. These central demands will con- sirens). The distinctions between spatial attention,
flict and cause delays in completing one or both feature-based attention, and object attention need to
tasks. Importantly, executive control also demands be kept in mind, but for brevity we collectively refer
central processing resources, meaning that deciding to the selection of sensory information as input
how to prioritize the two tasks will conflict with selection, except where this would lead to confusion
performing central operations on the two tasks. The or inaccuracy.
practical implication of the limitations on central Attention can also be tuned broadly or narrowly
processing capacity is to force a serial allocation of with respect to sensory input. The air traffic control-
attention to tasks, with the result that the manage- ler, for example, can set spatial attention to cover a
ment of task scheduling will increase workload and large portion of the radar display or tune it narrowly
constrain situation awareness. In the remainder of on a specific flight path corridor. Tuning broadly
the chapter, we explore the cycle in Figure 17.1 in gives the operator information about the overall
more detail as it relates to the nature of goal-directed structure of the information at the expense of detail.
input selection, alerting, and executive control In fact, when attention is broadly tuned, even brief
strategies for dealing with concurrent tasks. glimpses can yield reasonably accurate judgments of
overall scene properties (i.e., the gist of a complex
Input Selection scene; Potter, 2012). Radiologists are better than
A high proportion of operational errors are associ- chance at discriminating randomly chosen normal
ated with failures to attend to relevant information mammograms from those with a tumor after only a
(Dismukes & Nowinski, 2007). Consistent with this, 200-ms presentation, even though they are unable to
errors frequently occur in the first of the three stages accurately locate the tumor (Drew, Evans, Vo, Jacob-
of situation awareness, that associated with informa- son, & Wolfe, 2013). Tuning attention narrowly to a
tion acquisition (Jones & Endsley, 1996). We have single object selects that object for further processing,
introduced the term input selection to refer to the making it possible to extract semantic properties.
process of attending to external events. Studies of
attention, however, recognize varieties of sensory The efficiency of input selection. One of the
selection tied both to sensory modality (e.g., audi- most important jobs of the human systems engi-
tory, visual) and the basis of selection (e.g., spatial, neer is to determine which information should be
feature, object). In vision, selection is done primarily explicitly displayed and which of several possible
by attending to the spatial location of task-relevant formats to use. The choice of formats and the types

264
 Attention and Multitasking

of coexisting formats define a structure for the discriminated without focused attention. Tones
information display that will strongly influence the within a critical band of each other will cause inter-
time and effort it takes to attend to the right piece of ference and require focused attention to identify.
information. Several factors are known to contribute Even when edges fall outside the critical distance,
to the ease of information access. the overall number of visible objects can interfere
Spatial proximity of stimuli to each other has a with attention. Beck et al. (2010) found that global
strong effect on attention and perception. The ability clutter was in fact a better predictor of performance
to identify the shapes of objects plays a critical role than local clutter, even though local clutter would
in tasks such as reading or symbol recognition. Iden- be expected to cause more crowding. Consistent
tification relies on the accurate assignment of edges with the effects of global clutter, studies of perceptual
to objects, that is, the differentiation of edges that are load (Lavie & Tsal, 1994) have shown that the num-
connected to form one object from those that belong ber of visual items on a display affects the efficiency
to other nearby objects. When the edges of adjacent of input selection. As the number of simultaneously
objects fall within a critical distance of each other presented visual objects increases from two to six,
they give rise to a phenomenon referred to as visual with six or more constituting high perceptual load,
crowding, which interferes with the ability to bind attention becomes narrowly focused on individual
edges to objects (Pelli & Tillman, 2008). Visual items, requiring an item-by-item search for selection
crowding is a function of the distance between the (Yantis & Johnston, 1990). The narrowing of atten-
edges of objects, which varies as a function of the tion also occurs under high arousal, as with fear or
distance of the objects from the fovea of the eye. stress (Cain, Dunsmoor, LaBar, & Mitroff, 2011), or
Crowding will occur when the edges of adjacent as a primary task becomes more difficult (Ikeda &
objects fall within a critical distance of each other, Takeuchi, 1975). This narrowing of attention
which is approximately half the angular distance of reduces the amount of information obtained from
the object from central fixation along a radial path, unattended scene elements.
as shown in Figure 17.2. One effect of display clutter Attention also will become finely tuned to the
(Beck, Lohrenz, & Trafton, 2010) is to increase the regularity in display structure and organization. By
likelihood that the edges of separate symbols fall structure we mean the overall visual or auditory
within the critical distance, thus giving rise to placement and spacing of information sources. Reg-
crowding. Crowding does not affect judgments ularities in structure can be achieved through the
about properties such as color, as these do not entail well-understood practice of presenting information
assignment of edges to objects. Proximity effects also in fixed spatial locations. Interestingly, the visual
occur in audition, where the critical band describes system appears to be quite sensitive to structural
a separation relationship between frequencies of regularities in a display, even when we are con-
simultaneously presented sounds similar to the criti- sciously unaware of it. When people are shown ran-
cal distance in visual space. For pure tones above domly generated visual displays, half of which are
500 Hz, the critical band is a 3% window around the repeated several times, and half of which are newly
frequency of the tone. Tones separated from each generated on each trial, target identification times
other by more than a critical band can be easily become faster for repeated displays even though
posttests show no ability to discriminate old from
new displays (Chun & Jiang, 1998). By organization
A. + WRT we mean the way information is grouped. It is good
practice to provide a regular and consistent group-
B. + W R T ing of information (e.g., U.S. Department of
ϕ°
Defense, 1999, Section [Link].2), but it can be
Critical distance ≈ 0.5 ϕ° complicated to determine just how to organize the
information. The time to access information has
FIGURE 17.2. An example of visual crowding. been shown to be reduced by placing information

265
Remington and Loft

Zero Point Normal Range

Aligned to Zero Point

Aligned to “Normal”

Normal Abnormal

FIGURE 17.3. Emergent properties in information displays.

from the same subsystem, or information routinely In visual displays, eye fixations are directed selec-
accessed together, in close spatial proximity tively to scene elements that provide information
(Wickens & Carswell, 1995). Yet, whether to group the observers are trying to obtain (Yarbus, 1967).
by subsystem or frequency of coaccess depends crit- Patterns of eye fixation differ between novice and
ically on the nature of the information being sought experienced drivers (Mourant & Rockwell, 1972)
and the overall set of tasks to perform. In Figure and pilots (Bellenkes, Wickens, & Kramer, 1997),
17.3, we illustrate two ways to present information reflecting the effects of knowledge on information
from several dials. In the top row, the dials are seeking. These top-down guided shifts of attention
aligned to the zero point, and in the bottom row, are not necessarily deliberate. In well-learned tasks,
they are aligned by the limit (red zone). If the oper- such as reading (Reichle et al., 1998), reaching
ator wants to know whether all the dials read nor- and grabbing (Epelboim, 1998), or typing (Shaffer,
mal, then the bottom is preferred, as any abnormal 1975), attention shifts automatically as a component
reading will stand out as unique in the orientation of a larger behavioral routine.
of the needle. However, if the operator wants to
Not seeing or hearing what we do not expect to see
know about position relative to the zero point, then
or hear. Top-down guidance of attention predis-
aligning by zero would be better.
poses us to miss information that does not conform
Top-down control: Modulation of attention by task to our goals or expectations. Directing attention to
goals, expectancies, and state. Attention is guided one source of information in the world impairs our
not just by the sensory structure of displays but also ability to process other unattended information.
by the relevance of information sources to active When people attend to one of two different audi-
task goals. The modulation of information access by tory messages played concurrently in different ears
task relevance is indicated by the top-down pathway (i.e., dichotic presentation), they generally are unable
linking working memory to input selection in Figure to report the contents of the unattended messages,
17.1 (Folk, Remington, & Johnston, 1992). Top- detect when the language switches, or determine
down control biases us to selectively sample infor- whether the two messages are the same (see Pashler,
mation sources consistent with our current goals 1999, pp. 39–67, for a comprehensive review). Even
and expectations. For example, in simulated driving a highly salient stimulus in the unattended channel,
experiments, drivers asked to follow road signs of a such as one’s name, is only reported about a third
specific color, while ignoring signs in other colors, of the time (Moray, 1959), a value that is further
brake sooner to avoid colliding with pedestrians reduced when care is taken to ensure attention is
when the color of a pedestrian’s clothes matches directed to the target channel when the name is
the color they are primed for (Most & Astur, 2007). presented (Wood & Cowan, 1995). Only the most

266
 Attention and Multitasking

rudimentary sensory information—for example, HUD, however, the instrument data are superim-
the presence or absence of input, the gender of the posed over the scene. To select instrument data,
speaker (i.e., pitch), or the replacement of a speech attention must actively suppress scene information,
stream with a pure tone—can be reliably reported. because the spatial locations of the scene elements
Likewise, observers cannot report the properties of and instruments overlap. As the aircraft maneuvers,
unattended visual objects even when they appear the location of scene information changes with
in close spatial proximity. Neisser and Becklen respect to the fixed data on the HUD, further facili-
(1975) had participants attend to one of two videos tating the grouping of the HUD as foreground and
superimposed over each other. When participants the external scene as background. Indeed, studies
attended to events in one video, they could not have shown an advantage for state data when they
report any meaningful content from the unattended are linked to scene elements (scene-linked displays)
video, even salient or unusual events. Such limits versus when they are placed on a separate virtual
on awareness of events in unattended channels have plane (Fadden et al., 2001).
consequences for the workplace. Operators often Cases of inattentional blindness show that
monitor multiple simultaneous conversations, track when people are actively engaged in a task that
sounds while engaged in conversation, or deal with requires attending to a stream of ongoing activity,
multiple simultaneous visual objects. We are able they miss conspicuous objects in their line of sight
to follow multiple auditory streams or keep track of whose properties are irrelevant to the ongoing task
multiple objects only by constantly shifting atten- (Simons & Chabris, 1999). Studies of experienced
tion between streams. radiologists have found that more than 80% failed
Failures to note unattended stimuli are not sim- to detect a gorilla 48 times larger than the average
ply due to the low salience of the sensory signals. tumor that was superimposed on an X-ray, even
Highly salient objects, or unusual events, are rou- when their eyes fixated it (Drew, Vo, & Wolfe,
tinely missed when they are irrelevant to current 2013), and that 60% failed to notice the absence of
task goals, a phenomenon best illustrated perhaps in the clavicle in a chest X-ray (Potchen, 2006). In an
studies of inattentional blindness. One of the first analysis of air traffic control reports, Shorrock
demonstrations of blindness induced by task goals (2007) found that one third of errors, such as failing
came from a simulation study of a head-up display to detect possible mid-air collisions, arose during
(HUD). Haines (1991) tested the efficacy of a HUD circumstances when controllers failed to perceive
in commercial aircraft by having airline pilots moni- the relevance of crucial aircraft flight data informa-
tor a simulated landing. He found several cases in tion despite focusing directly on the location of rele-
which experienced airline captains failed to notice vant visual inputs. Similarly, people can be quite
the image of a Boeing 747 jumbo jet straddling the insensitive to significant changes in the visual world,
runway in the direct path of the aircraft. Although a phenomenon referred to as change blindness
having many benefits, HUD displays appear to pro- (Simons & Rensink, 2005).
mote blindness to scene information. A meta- In part, inattentional and change blindness are
analysis of reported cases found a statistically ele- surprising because we implicitly assume that we
vated miss rate in the detection of unexpected attend to the objects our eyes fixate. However,
events when pilots were attending to a HUD (Wick- attending requires that we actively select the items at
ens, Hooey, Gore, Sebok, & Koenicke, 2009). It is fixation. Selection is affected by what we expect to be
interesting to speculate on why HUD displays are there, which, in turn, is affected by our past experi-
associated with increased misses of scene informa- ence. Consequently, errors of attention are sensitive
tion (Fadden, Ververs, & Wickens, 2001). In a typi- to the prevalence of events, resulting in increased
cal head-down display, the pilot elects to fixate the miss rates for rare events. Our conscious expecta-
cockpit instruments rather than the external scene. tions do not play as important a role in prevalence as
The time away from the scene causes delays, and do our active encounters with events. Rare events are
perhaps misses, in perceiving scene events. With a frequently missed even in such domains as baggage

267
Remington and Loft

screening and radiology where events are expected The adaptation of attention to the frequency of
to be rare. A study of almost 2 million chest X-rays occurrence is consistent with mathematically opti-
in Los Angeles from 1950 to 1955 found the preva- mal sampling. Optimal sampling also is affected by
lence of cancerous tumors to be low (0.9/1,000), and the relevance and importance of specific informa-
tumors were missed in the X-rays of approximately tion. Patterns of pilot eye fixations in monitoring
70% (662 of 955) of patients who died from cancer instruments during flight simulation are well
(Kundel, 2000). A recent study simulating baggage approximated by an optimal model that includes
screening found that 7% of weapons were missed both the likelihood and relevance (value to the task)
when they occurred on 50% of trials, compared to of information from various displays (Wickens &
30% when the frequency was reduced to 1% (Wolfe, McCarley, 2007). Indeed, it also has been argued
Horowitz, & Kenner, 2005). Studies of vigilance that attributions of carelessness, complacency, or
have repeatedly shown that over time observers neglect in the allocation of attention cannot be made
become increasingly prone to missing rare target with confidence without a careful consideration of
events (MacKworth, 1968). The effect of event likeli- what optimal monitoring would yield (Moray &
hood may seem surprising, as it is well known that Inagaki, 2000). This does not mean that human
we are poor at weighing the frequency or probability sampling is optimal in the strict sense that no sys-
of occurrence of an event against other factors in tem could do better, but rather that with practice,
making conscious decisions (Kahneman & Tversky, human expert behavior is sensitive to the same fac-
1996). Yet, the above examples show that perceptual tors that determine mathematical optimality. Know-
and attentional mechanisms are quite sensitive to the ing that human operators behave rationally allows
frequency with which we experience an event and one to use direct system measurements (e.g., event
the probability of its current occurrence. rate, relevance) while still in the design stage to
The difficulty in detecting rare events seems to anticipate patterns of human performance.
reflect the natural adaptation of our decision system
to event likelihood rather than gross negligence. Alerting
Adaptation to event likelihood may have evolution- We have emphasized the extent to which top-
ary roots, as it appears also in the foraging behavior down control biases us to attend to information con-
of animals and in people’s search for information sistent with task goals, and we have discussed how
on the Internet or in libraries, as described in infor- little content other than coarse sensory data is avail-
mation foraging theory (Pirolli & Card, 1999). The able on unattended channels. As a result, to capture
detection and recognition of target events can be an operator’s attention and alert him or her to criti-
seen as a process of collecting evidence until it is cal new information, the alert signal must be con-
sufficient to cross a threshold determining whether spicuous from other events in sensory properties
a target is present. When events are rare, our sys- that can be detected during preattentive processing
tem naturally sets a high criterion as protection (e.g., Woods, 1995). Conspicuousness is not simply
against excessive false alarms, which leads to a function of the physical magnitude of the object
increases in missed targets. Adaptation to base-rate (e.g., louder, brighter) but rather of its salience or
occurrence also may underlie the phenomenon of physical distinctiveness from other objects. It is easy
satisfaction of search, which refers to a significant to find a target if it is unique from background
reduction in the detection of second or third targets objects in one specific sensory dimension—color,
in cases when an unknown number of multiple tar- pitch, orientation, or timbre. The display in the bot-
gets are present. Single targets generally occur more tom of Figure 17.2 takes advantage of salience to
frequently than multiple targets, making the second draw attention to the unique orientation of one of
target a relatively rare event. When second-target the needles. In the air traffic control example pre-
likelihood is approximately that of the first, detec- sented earlier, the blinking of the aircraft icon will
tion rates for first and second targets are the same be highly salient if all other icons are static. Its
(Cain, Vul, Clark, & Mitroff, 2012). salience diminishes as the number of blinking

268
 Attention and Multitasking

objects increases. Salience is also low for signals To ensure the salience of displayed information, it
defined as the conjunction of two sensory proper- is better to make the critical information greater in
ties. A red blinking icon cannot be easily detected in magnitude along some sensory dimension than the
a background of white blinking icons and red static irrelevant objects, or to add a distinctive feature
icons (Treisman & Gelade, 1980). Attention cannot rather than take one away.
be set for red + blinking but rather only for red or It is important to distinguish the properties of an
only for blinking. alert intended to capture the operator’s attention
It is also necessary to distinguish between an from information in the alerting signal intended to
alert that draws the operator’s attention to a frequent facilitate subsequent fault diagnosis. It is standard
event, whose occurrence can be said to be part of practice to modify the alerting or alarm signal to
ongoing contextual expectations, and an alarm that contain information about the severity of the alert,
calls the operator’s attention to an event outside con- the subsystem associated with it, or the likelihood
textual expectations. The blinking icon that alerts that the alarm truly indicates a system failure rather
the operator to the aircraft ready for acceptance into than a false alarm (a “likelihood alarm display”;
the airspace is part of the task expectancies and will e.g., Sorkin & Woods, 1985). MIL-STD-1472F (U.S.
readily capture attention so long as there is sufficient Department of Defense, 1999, p. 23, Table II), for
perceptual salience (Folk et al., 1992; Steelman, example, recommends combining several pieces of
McCarley, & Wickens, 2011). In vision, the lumi- information into a single indicator light, such as by
nance transients found in motion, flicker, and the conjunction of color (red, yellow, green, white),
abrupt-onset stimuli are very effective in capturing flashing (flashing, steady), and size (large, small), to
attention (Remington, Johnston, & Yantis, 1992; indicate the severity of the fault and the system
Yantis & Jonides, 1984). Auditory signals, however, state. In this coding scheme, a large, steady red light
have an inherently greater alerting ability than do would indicate one state, whereas a small, steady
visual signals (Posner, Nissen, & Klein, 1976); the yellow light would indicate another separate state.
effectiveness of auditory alerting signals is enhanced Although these conjoined properties may facilitate
when they include information about the spatial the understanding of system state, the ability of a
location of the fault system that can guide visual stimulus to capture attention depends on its salience
attention (Ho & Spence, 2005). Thus, to alert an with respect to a single feature dimension, not the
operator to an emergency situation that falls outside conjunction of several. Consider the case of a traffic
the current expectations, as when a system error light that has a turn arrow in addition to the three
occurs, the alerting signal will be more effective in circular lights (red, yellow, green). Motorists wait-
capturing the operator’s attention if it combines both ing for a green turn arrow frequently start to go
a salient auditory and visual feature. when the main light turns green even when the
There is an asymmetry in the combination of arrow remains red. Going only on a green arrow
stimulus magnitude and salience. Attention is more requires a judgment of the conjunction of color
easily directed to an object of greater magnitude (green) and shape (arrow). This impulse to acceler-
among lesser magnitude objects than it is to a unique ate occurs because attention can be set to green
item of lesser magnitude among those of greater alone but not to the conjunction of two features
magnitude (Steelman et al., 2011; Treisman & (e.g., green and arrow). Attention needs to be
Gelade, 1980; Wickens et al., 2009). Attention is focused on the green object and further cognitive
more readily drawn to a bright light among dim processing done to bind the color with the shape
lights, or a large circle among small circles, than the (Treisman & Gelade, 1980). In the MIL-STD-1472F
converse. The absence of a feature also does not example, whether attention is captured by one of the
strongly capture attention (Wolfe, 2001). Thus, it is colors or by size depends on the relative salience
more difficult to detect the disappearance of an air- along the color or size dimensions; the conjunction
craft icon from a controller’s radar screen than it is to does not participate in drawing the operator’s atten-
detect a flashing icon or a change to a unique color. tion to the light. Decoding the state information

269
Remington and Loft

embedded in the conjunction of properties requires entering the sector. There is strong evidence that
further cognitive processing after attention has been over brief intervals of approximately a quarter of a
focused on the indicator. second the central processor acts as an all-or-none
Alerting signals capable of automatically captur- bottleneck, a concept referred to as central bottleneck
ing attention have been shown to pose particular theory; during this time, only one central processor
problems for operators (for a detailed discussion, action can be executed at any given moment
see Woods, 1995). Because the signals are effective (Pashler & Johnston, 1989). This places a strict
in capturing attention, false alarms are particularly limit on the ability to process two tasks at once.
disruptive as they distract operators for no purpose. According to central bottleneck theory, however,
Operators have been known to disconnect alerts the structure of information processing (see Figure
prone to excessive false alarms, thereby endangering 17.1) allows operators to overlap sensory processing
system safety (Parasuraman & Riley, 1997; Sorkin, and action execution for two or more tasks; the con-
Kantowitz, & Kantowitz, 1988). Evidence suggests troller might fixate the blinking icon of the entering
that the false alarm problem could be mitigated to a aircraft while issuing a spoken clearance to another.
degree if the magnitude of the alarm was graded in Moreover, the sensory processing of auditory and
proportion to the likelihood of a true fault, rather visual information can be overlapped (Pashler,
than simply being an all-or-none binary response 1984; Pashler & Johnston, 1989); the controller
(Sorkin et al., 1988). Alarms also can be problematic might fixate the blinking icon of the entering aircraft
because a fault in one system can trigger errors from while comprehending the radio message from
other coupled systems, resulting in a cascade of another. Evidence for this overlap can be seen in
alarms that distract operators from diagnosing and patterns of eye fixation when executing tasks in suc-
responding to the original fault. This has been a cession (Remington, Wu, & Pashler, 2011) and
problem in surgery, prompting calls for more when reading (Reichle et al., 1998), where it has
sophisticated alarm systems (Edworthy, 2005). One been shown that the eyes move to the next item
interesting possibility is to take advantage of the role prior to responding to the previous one. This over-
of top-down expectancies and display critical infor- lap must be considered when estimating the time it
mation in a way that matches those expectations. takes to complete procedures, such as mouse-and-
New alert displays being tested in simulated hospital click actions in checklist completion or other
surgeries have shown that changes to a regular computer interaction (Vera, John, Remington,
rhythmic sound are quickly detected when those Matessa, & Freed, 2005), in typing (Card, Moran, &
sounds correspond to the internal model the sur- Newell, 1983), and in performing a succession of
geon has of the physiological process being depicted grasping actions (Epelboim et al., 1994). Moreover,
(Watson & Sanderson, 2004). The advantage of this central bottleneck approaches have been shown to
technique is that the sound becomes part of the rou- provide precise timing estimates for the execution
tine representation of the context and contributes of keystrokes, mouse move-and-click, and other
throughout to the assessment of patient state. This detailed behaviors in more complex applied settings
suggests that it may be possible to improve alarms (Gray, John, & Atwood, 1993). Overall, the assump-
by carefully considering the ways in which operators tion of parallel processing of sensory and motor
derive meaning from a context. operations combined with the assumption of limita-
tions in central processing fit human data well,
Executive Control and Task Scheduling which suggests that interference between simultane-
in Multitasking ous tasks is due in large part to central processing
It is common for operators to have several concur- demands. It is important to emphasize that to show
rently active tasks. The air traffic controller may parallel processing, laboratory studies must take
have course changes to give to one aircraft at the care to eliminate sources of sensory and motor con-
same time another is requesting an altitude change, flict. Interference can also arise from perceptual pro-
while there also is a blinking icon for an aircraft cessing (e.g., if two tasks require the eyes to fixate

270
 Attention and Multitasking

two separate locations in parallel) or from response tasks (e.g., monitoring for threats, assessing threat
conflicts (e.g., if tasks require two messages to be status of new targets). Each step in a task can be
spoken simultaneously). assumed to require a proportion of sensory, central,
Central bottleneck theory also has been shown to and response resources over some time interval, an
predict patterns of interference in common dual- assumption that, in turn, can be used to examine
task situations (Pashler, 1984; Pashler & Johnston, the conflicts between tasks over some interval of
1989). Driving simulations have shown delays in time (see Chapter 18, this handbook). Estimates can
braking following the onset of tail lights by drivers be adjusted to the scale required and can be made
trying to comprehend simultaneous auditory input easily and quickly by querying experienced opera-
(Levy, Pashler, & Boer, 2006). This illustrates that tors, in keeping with the demands of system
central processing operations extend to the compre- engineering.
hension of spoken messages, explaining why hands- We noted previously that interference between
free cell phones continue to produce some two tasks does not always mean that central process-
interference with vehicle maneuvering. ing conflicts have occurred, because interference can
Because central bottleneck theory requires con- also arise from competing demands on sensory and
sideration of events on the order of tens of millisec- motor systems. Interference can also occur when
onds, it can be difficult to apply to processing that two tasks rely on shared internal representations
extends over intervals on the order of tens of sec- that are similar enough to cause crosstalk even
onds or more, a time scale often of interest in system though neither task is associated with central pro-
design. It is often more tractable to consider central cessing. It is easier, for example, to listen to music
processing as a pool of fixed resources that can be while speaking than to listen to music and try and
allocated proportionally to two or more tasks, an hum a different tune, or to listen to one conversa-
idea known as capacity-sharing accounts (Navon & tion while engaging in another. Likewise, conflicts
Gopher, 1979). Capacity-sharing accounts assume occur if one is imagining the spatial layout of a
that tasks compete for resources and that to a point scene while attempting to perform another spatial
task performance is a function of the resources task. In all the above cases, interference occurs
allocated to it. Allocating more resources to one because inputs and outputs share a common repre-
task (A) will reduce the allocation to the other (B), sentation. In his multiple-resource theory, Wickens
increasing the rate at which A is processed and (2008) referred to this as code interference to distin-
decreasing that of B. Performance on A will improve guish it from the more general competition for cen-
at the expense of B to the extent that completion tral processing resources. Multiple-resource theory
time is in fact a function of the resource allocation. posits that separate resources are associated with
If B is very easy, devoting more resources to it verbal and spatial representations, sensory modali-
likely will not improve its performance and in fact ties (vision, audition), and motor systems. Items for
could be detrimental; taking resources away from a two tasks interfere with one another when they
difficult auditory discrimination and applying them compete for the same representation but can be exe-
to the detection of a bright light likely would result cuted in parallel if they do not.
in poorer performance on the auditory task with lit-
tle benefit to the visual task. Likewise, devoting Strategies of task selection. When multiple tasks
additional resources to a difficult task will not compete for processing resources, mechanisms of
improve performance when the source of the diffi- executive control determine which task should have
culty is poor or noisy data (see Chapter 18, this access at any given time and also provide specialized
handbook). cognitive operations that swap out an old task and
Capacity-sharing models can be readily applied insert a new one. Switching tasks imposes a cost on
to staffing decisions, as in Case Study 3 (see Chapter performance that must be considered in designing
4, this handbook), by estimating the workload procedures. Consider, for example, a common labora-
involved in completing a sequence of high-level tory task in which Task 1 (T1) requires a response

271
Remington and Loft

to indicate whether a single digit is odd or even, and heuristics. Raby and Wickens (1994) had pilots
Task 2 (T2) requires a response to indicate whether make simulated landings under low, medium, and
the digit is less than or greater than 5. The time to high workload conditions. When workload was
complete a list of trials that alternate between T1 high, pilots spent proportionally more time on
and T2 is slower than the time to complete a list “must do” tasks, to the exclusion of less important
containing only T1 or T2 (Rogers & Monsell, 1995). tasks. They were still sensitive to task importance,
Switching costs have been observed in HUD when but they tended to overengage in important tasks,
switching from the display to the out-the-window becoming less sensitive to the ensemble of consider-
scene (Fadden et al., 2001). Switching costs and errors ations that characterizes optimal scheduling. Under
resulting from switches can be mitigated by percep- time pressure, operators appear to adopt proactive
tual support, in the form of spatial separation, color, strategies that minimize future information acquisi-
or some other salient sensory feature, that provides tion to maintain acceptable levels of workload
operators a means to more easily detect which task (Loft, Sanderson, Neal, & Mooij, 2007). For exam-
operations to apply (Dismukes & Nowinski, 2007). ple, Loft, Bolland, Humphreys, and Neal (2009)
Studies of executive control have focused on found that controllers approximate the future rela-
costs in switching, with far less research devoted to tive positions of aircraft to ensure aircraft separation
how we select which task to pursue—that is, the and minimize future monitoring (i.e., “set and for-
problem of task scheduling. We previously get”). Nonetheless, some inefficiency in task sched-
described how with practice operators come to uling is expected with increasing task load, because
approximate optimal sampling of information. That scheduling itself uses the same limited-capacity cen-
is, there is a statistical underpinning to task moni- tral resources as task execution. Time spent devising
toring that leads to a rational scheduling of or updating a schedule is time not spent on the tasks
resources to each task. A rational schedule would themselves. This inefficiency could spiral, as increas-
consider the rate at which information arrives, the ing the number of tasks increases demands for both
magnitude of the changes in value, the likelihood scheduling and task processing.
that a response would be required, the importance
of the task, and the cost of switching. If the changes Interruption and distraction. Interruption and
are frequent and large, then the operator should distraction represent cases of involuntary capture
sample that system often, as frequent corrective of attention similar in many respects to alerts, dis-
actions will need to be taken. If the changes are fre- cussed earlier. They involve a signal (e.g., phone
quent but small, corrective actions will be required ringing, alarm, colleague knocking on your door)
less frequently, and the task needs to be monitored that reorients spatial attention coupled with a shift
only frequently enough to avoid the cumulative of task set to the interrupting event. Interruptions
effect of a number of small deviations. With prac- need not have an external trigger—they can also
tice, experts incorporate these statistical properties occur when our thoughts stray from the imme-
into their monitoring strategies. In many cases, the diate task (Boehm-Davis & Remington, 2009).
monitoring behavior of trained operators is well fit Interruption and distraction have been implicated as
by optimal models that incorporate these and simi- causal factors in accidents in aviation (Loukopoulos,
lar assumptions (Steelman et al., 2011). Dismukes, & Barshi, 2001), medicine (Beyea, 2007),
Under conditions of light to moderate task and motor vehicle operation (Horberry, Anderson,
demands, human task scheduling also appears sensi- Regan, Triggs, & Brown, 2006).
tive to the combination of the importance of tasks, A major problem with interruption is the failure
the urgency of tasks, the frequency with which the to resume the original task, or the failure to resume
task state must be updated, and the time available the same point of progress, leading to a failure to
for the task, as is an optimal model (Dessouky, complete all steps of the original task. There are sev-
Moray, & Kijowski, 1995). As demands increase, eral reasons for this. Increases in the time spent on
however, people tend to revert to simpler scheduling the interrupting task increases the likelihood that

272
 Attention and Multitasking

state information is lost from memory, including References

knowledge of which parts of the original task have Altmann, E. M., & Gray, W. D. (2008). An inte-
been completed (Monk, Trafton, & Boehm-Davis, grated model of cognitive control in task
2008). It is important to keep in mind that the switching. Psychological Review, 115, 602–639.
doi:10.1037/0033-295X.115.3.602
switch back to the original task will be a central
resource-demanding voluntary switch. Another con- Beck, M. R., Lohrenz, M. C., & Trafton, J. G. (2010).
Measuring search efficiency in complex visual
sequence of interruption is that the original task may search tasks: Global and local clutter. Journal of
have to be suspended at a suboptimal point of com- Experimental Psychology: Applied, 16, 238–250.
pletion. Evidence suggests that people voluntarily doi:10.1037/a0019633
suspend one task to attend to another at the comple- Bellenkes, A. H., Wickens, C. D., & Kramer, A. F. (1997).
tions of well-defined subgoals within the original Visual scanning and pilot expertise: The role of
attentional flexibility and mental model develop-
task to facilitate the process of resuming the original ment. Aviation, Space, and Environmental Medicine,
task (Altmann & Gray, 2008). 68, 569–579.
It has been difficult to devise effective aids for Beyea, S. C. (2007). Distractions, interruptions, and
the problem of interruption. Operators have been patient safety. AORN Journal, 86, 109–110, 112.
observed to use notes to remind themselves of doi:10.1016/[Link].2007.06.015
uncompleted tasks (John, Remington, & Steier, Boehm-Davis, D. A., & Remington, R. (2009). Reducing
1991), an indication that operators are aware of the the disruptive effects of interruption: A cognitive
framework for analysing the costs and benefits of
difficulty of keeping track of multiple tasks. Others intervention strategies. Accident Analysis & Prevention,
have advocated briefing operators on their vulnera- 41, 1124–1129. doi:10.1016/[Link].2009.06.029
bility to forgetting after being interrupted and how Cades, D. M., Boehm-Davis, D. A., Trafton, J. G., &
to recognize warning signs (Beyea, 2007; Louko- Monk, C. A. (2011). Mitigating disruptive effects
poulos et al., 2001; Loukopoulos, Dismukes, & of interruptions through training: What needs to
be practiced? Journal of Experimental Psychology:
Barshi, 2012). However, both of these approaches Applied, 17, 97–109. doi:10.1037/a0023497
impose an extra task that requires central process-
Cain, M. S., Dunsmoor, J. E., LaBar, K. S., & Mitroff, S. R.
ing resources and time to execute, interfering with (2011). Anticipatory anxiety hinders detection of
other tasks. Laboratory studies have shown a second target in dual-target search. Psychological
improved performance following interruption after Science, 22, 866–871. doi:10.1177/0956797611412393
practice on interruption with a fixed ensemble of Cain, M. S., Vul, E., Clark, K., & Mitroff, S. R. (2012).
tasks (Cades, Boehm-Davis, Trafton, & Monk, A Bayesian optimal foraging model of human
visual search. Psychological Science, 23, 1047–1054.
2011). Whether this result will scale to complex doi:10.1177/0956797612440460
operations remains an open question.
Card, S. K., Moran, T. P., & Newell, A. (1983). The psy-
chology of human–computer interaction. Hillsdale, NJ:
Erlbaum.
SUMMARY
Chun, M. M., & Jiang, Y. (1998). Contextual cueing:
This overview showed how the allocation of limited- Implicit learning and memory of visual context
capacity resources is determined by both bottom-up guides spatial attention. Cognitive Psychology, 36,
28–71. doi:10.1006/cogp.1998.0681
stimulus processing and top-down, goal-directed
Dessouky, M. I., Moray, N., & Kijowski, B. (1995).
task processing. Errors of attention often occur as
Taxonomy of scheduling systems as a basis for the
emergent properties of the interaction of these bottom- study of strategic behavior. Human Factors, 37,
up and top-down mechanisms, regardless of the 443–472. doi:10.1518/001872095779049282
intentions or motivations of the operator. An Dismukes, R. K., & Nowinski, J. L. (2007). Prospective
understanding of how incoming stimuli are selected memory, concurrent task management, and task error.
and tasks are managed can aid in determining how In A. F. Kramer, D. A. Wiegmann, & A. Kirlik (Eds.),
Attention: From theory to practice (pp. 225–239). New
many tasks are feasible and how information York, NY: Oxford University Press.
should be structured to facilitate efficiency and to Drew, T., Evans, K., Vo, M. L.-H., Jacobson, F. L., & Wolfe,
mitigate error. J. M. (2013). Informatics in radiology: What can you

273
Remington and Loft

see in a single glance and how might this guide visual Ikeda, M., & Takeuchi, T. (1975). Influence of foveal
search in medical images? Radiographics, 33, 263–274. load on the functional visual field. Perception &
doi:10.1148/rg.331125023 Psychophysics, 18, 255–260. doi:10.3758/
Drew, T., Vo, M. L.-H., & Wolfe, J. M. (2013, July 17). BF03199371
The invisible gorilla strikes again: Sustained James, W. (1918). The principles of psychology. New York,
inattentional blindness in expert observers. NY: Henry Holt.
Psychological Science. Advance online publication. John, B. E., Remington, R. W., & Steier, D. (1991).
doi:10.1177/0956797613479386 An analysis of space shuttle countdown activities:
Durso, F. T., & Sethumadhavan, A. (2008). Preliminaries to a computational model of the NASA
Situation awareness: Understanding dynamic test director (Technical Report No. CMU-CS-91-138).
environments. Human Factors, 50, 442–448. Pittsburgh, PA: Carnegie Mellon University.
doi:10.1518/001872008X288448 Jones, D. G., & Endsley, M. R. (1996). Sources of situa-
Edworthy, J. (2005). Fewer but better auditory alarms tion awareness errors in aviation. Aviation, Space, and
will improve patient safety. Quality & Safety in Health Environmental Medicine, 67, 507–512.
Care, 14, 212–215. doi:10.1136/qshc.2004.013052 Kahneman, D., & Tversky, A. (1996). On the reality
Endsley, M. R. (1995). Toward a theory of situation of cognitive illusions. Psychological Review, 103,
awareness in dynamic systems. Human Factors, 37, 582–591. doi:10.1037/0033-295X.103.3.582
32–64. doi:10.1518/001872095779049543 Kundel, H. L. (2000, April 14). Disease prevalence and
Epelboim, J. (1998). Gaze and retinal-image-stability in the index of detectability: A survey of studies of
two kinds of sequential looking tasks. Vision Research, lung cancer detection by chest radiography. In E. A.
38, 3773–3784. doi:10.1016/S0042-6989(97)00450-1 Krupinski (Ed.), SPIE Proceedings: Medical Imaging
2000: Image Perception and Performance (Vol. 3981,
Epelboim, J., Collewijn, H., Edwards, M. A., Erkelens, pp. 135–144). doi:10.1117/12.383100
C. J., Kowler, E., Pizlo, Z., & Steinman, R. M.
(1994). Coordinated movements of the arm and Lavie, N., & Tsal, Y. (1994). Perceptual load as a major
head increase gaze-shift velocity. Investigative determinant of the locus of selection in visual-
Ophthalmology & Visual Science, 35, 1550. attention. Perception & Psychophysics, 56, 183–197.
doi:10.3758/BF03213897
Fadden, S., Ververs, P. M., & Wickens, C. D. (2001).
Pathway HUDs: Are they viable? Human Factors, 43, Levy, J., Pashler, H., & Boer, E. (2006). Central interfer-
173–193. doi:10.1518/001872001775900841 ence in driving: Is there any stopping the psycho-
logical refractory period? Psychological Science, 17,
Folk, C. L., Remington, R. W., & Johnston, J. C. (1992). 228–235. doi:10.1111/j.1467-9280.2006.01690.x
Involuntary covert orienting is contingent on
attentional control settings. Journal of Experimental Loft, S., Bolland, S., Humphreys, M. S., & Neal, A.
Psychology: Human Perception and Performance, 18, (2009). A theory and model of conflict detection in
1030–1044. doi:10.1037/0096-1523.18.4.1030 air traffic control: Incorporating environmental con-
straints. Journal of Experimental Psychology: Applied,
Gray, W. D., John, B. E., & Atwood, M. E. (1993). 15, 106–124. doi:10.1037/a0016118
Project Ernestine: Validating a GOMS analysis for
predicting and explaining real-world task perfor- Loft, S., Sanderson, P., Neal, A., & Mooij, M. (2007).
mance. Human–Computer Interaction, 8, 237–309. Modeling and predicting mental workload in
doi:10.1207/s15327051hci0803_3 en route air traffic control: Critical review and
broader implications. Human Factors, 49, 376–399.
Haines, R. F. (1991). A breakdown in simultaneous doi:10.1518/001872007X197017
information processing. In G. Obrecht & L. W. Stark
(Eds.), Presbyopia research: From molecular biology to Loukopoulos, L. D., Dismukes, R. K., & Barshi, I. (2001).
visual adaptation (pp. 171–175). doi:10.1007/978-1- Cockpit interruptions and distractions: A line obser-
4757-2131-7_17 vation study. In R. Jensen (Ed.), Proceedings of the
11th International Symposium on Aviation Psychology.
Ho, C., & Spence, C. (2005). Assessing the effectiveness Columbus: Ohio State University.
of various auditory cues in capturing a driver’s visual
attention. Journal of Experimental Psychology: Applied, Loukopoulos, L. D., Dismukes, R. K., & Barshi, I. (2012).
11, 157–174. doi:10.1037/1076-898X.11.3.157 The multitasking myth: Handling complexity in real-
world operations. Surrey, England: Ashgate Publishing.
Horberry, T., Anderson, J., Regan, M. A., Triggs, T. J., &
Brown, J. (2006). Driver distraction: The effects of MacKworth, J. F. (1968). Vigilance, arousal, and habitua-
concurrent in-vehicle tasks, road environment com- tion. Psychological Review, 75, 308–322. doi:10.1037/
plexity and age on driving performance. Accident h0025896
Analysis & Prevention, 38, 185–191. doi:10.1016/ Monk, C. A., Trafton, J. G., & Boehm-Davis, D. A.
[Link].2005.09.007 (2008). The effect of interruption duration and

274
 Attention and Multitasking

demand on resuming suspended goals. Journal of Potter, M. C. (2012). Recognition and memory for
Experimental Psychology: Applied, 14, 299–313. briefly presented scenes. Frontiers in Psychology, 3,
doi:10.1037/a0014402 32.1–32.9. doi:10.3389/fpsyg.2012.00032
Moray, N. (1959). Attention in dichotic listening: Raby, M., & Wickens, C. D. (1994). Strategic work-
Affective cues and the influence of instructions. The load management and decision biases in aviation.
Quarterly Journal of Experimental Psychology, 11, The International Journal of Aviation Psychology, 4,
56–60. doi:10.1080/17470215908416289 211–240. doi:10.1207/s15327108ijap0403_2
Moray, N., & Inagaki, T. (2000). Attention and compla- Reichle, E. D., Pollatsek, A., Fisher, D. L., & Rayner, K.
cency. Theoretical Issues in Ergonomics Science, 1, (1998). Toward a model of eye movement control
354–365. doi:10.1080/14639220052399159 in reading. Psychological Review, 105, 125–157.
Most, S. B., & Astur, R. S. (2007). Feature-based attentional doi:10.1037/0033-295X.105.1.125
set as a cause of traffic accidents. Visual Cognition, 15, Remington, R. W., Johnston, J. C., & Yantis, S.
125–132. doi:10.1080/13506280600959316 (1992). Involuntary attentional capture by abrupt
onsets. Perception & Psychophysics, 51, 279–290.
Mourant, R. R., & Rockwell, T. H. (1972). Strategies
doi:10.3758/BF03212254
of visual search by novice and experienced
drivers. Human Factors, 14, 325–335. Remington, R. W., Wu, S.-C., & Pashler, H. (2011).
doi:10.1177/001872087201400405 What determines saccade timing in sequences of
coordinated eye and hand movements? Psychonomic
Navon, D., & Gopher, D. (1979). Economy of the
Bulletin & Review, 18, 538–543. doi:10.3758/s13423-
human-processing system. Psychological Review, 86,
011-0066-0
214–255. doi:10.1037/0033-295X.86.3.214
Richtel, M. (2009, July 19). Drivers and legislators
Neisser, U., & Becklen, R. (1975). Selective looking:
dismiss cellphone risks. The New York Times.
Attending to visually specified events. Cognitive
Retrieved from [Link]
Psychology, 7, 480–494. doi:10.1016/0010-
technology/[Link]
0285(75)90019-5
Rogers, R. D., & Monsell, S. (1995). Costs of a predict-
Parasuraman, R., & Riley, V. (1997). Humans and auto-
able switch between simple cognitive tasks. Journal
mation: Use, misuse, disuse, abuse. Human Factors,
of Experimental Psychology: General, 124, 207–231.
39, 230–253. doi:10.1518/001872097778543886
doi:10.1037/0096-3445.124.2.207
Pashler, H. (1984). Processing stages in overlapping
Shaffer, L. H. (1975). Control processes in typing. The
tasks: Evidence for a central bottleneck. Journal
Quarterly Journal of Experimental Psychology, 27,
of Experimental Psychology: Human Perception and
419–432. doi:10.1080/14640747508400502
Performance, 10, 358–377. doi:10.1037/0096-
1523.10.3.358 Shorrock, S. T. (2007). Errors of perception in air traffic
control. Safety Science, 45, 890–904. doi:10.1016/j.
Pashler, H. (1999). The psychology of attention. Cambridge,
ssci.2006.08.018
MA: MIT Press.
Simons, D. J., & Chabris, C. F. (1999). Gorillas in our
Pashler, H., & Johnston, J. C. (1989). Chronometric evi-
midst: Sustained inattentional blindness for dynamic
dence for central postponement in temporally over-
events. Perception, 28, 1059–1074. doi:10.1068/p2952
lapping tasks. The Quarterly Journal of Experimental
Psychology A: Human Experimental Psychology, 41, Simons, D. J., & Rensink, R. A. (2005). Change blind-
19–45. ness: Past, present, and future. Trends in Cognitive
Sciences, 9, 16–20. doi:10.1016/[Link].2004.11.006
Pelli, D. G., & Tillman, K. A. (2008). The uncrowded
window of object recognition. Nature Neuroscience, Sorkin, R. D., Kantowitz, B. H., & Kantowitz, S. C. (1988).
11, 1129–1135. doi:10.1038/nn.2187 Likelihood alarm displays. Human Factors, 30, 445–459.
Pirolli, P., & Card, S. (1999). Information foraging. Sorkin, R. D., & Woods, D. D. (1985). Systems with
Psychological Review, 106, 643–675. doi:10.1037/0033- human monitors: A signal detection analysis.
295X.106.4.643 Human–Computer Interaction, 1, 49–75. doi:10.1207/
s15327051hci0101_2
Posner, M. I., Nissen, M. J., & Klein, R. M. (1976). Visual
dominance: An information-processing account of Steelman, K. S., McCarley, J. S., & Wickens, C. D.
its origins and significance. Psychological Review, 83, (2011). Modeling the control of attention in
157–171. doi:10.1037/0033-295X.83.2.157 visual workspaces. Human Factors, 53, 142–153.
doi:10.1177/0018720811404026
Potchen, E. J. (2006). Measuring observer performance
in chest radiology: Some experiences. Journal of Treisman, A. M., & Gelade, G. (1980). A feature-integration
the American College of Radiology, 3, 423–432. theory of attention. Cognitive Psychology, 12, 97–136.
doi:10.1016/[Link].2006.02.020 doi:10.1016/0010-0285(80)90005-5

275
 Remington and Loft

U.S. Department of Defense. (1999). Department of Wolfe, J. M. (2001). Asymmetries in visual search:
Defense design criteria standard: Human engineering An introduction. Perception & Psychophysics, 63,
(Document No. MIL-STD-1472F). Retrieved from 381–389. doi:10.3758/BF03194406
[Link]
Wolfe, J. M., Horowitz, T. S., & Kenner, N. M. (2005,
Management%20Documents/Appendix%20B%20
May 26). Cognitive psychology: Rare items often
Reference%20Documents/[Link]
missed in visual searches. Nature, 435, 439–440.
Vera, A. H., John, B. E., Remington, R., Matessa, M., & doi:10.1038/435439a
Freed, M. A. (2005). Automating human-
Wood, N. L., & Cowan, N. (1995). The cocktail party
performance modeling at the millisecond level.
phenomenon revisited: Attention and memory in
Human–Computer Interaction, 20, 225–265.
the classic selective listening procedure of Cherry
doi:10.1207/s15327051hci2003_1
(1953). Journal of Experimental Psychology: General,
Watson, M., & Sanderson, P. (2004). Sonification sup- 124, 243–262. doi:10.1037/0096-3445.124.3.243
ports eyes-free respiratory monitoring and task time-
Woods, D. D. (1995). The alarm problem and directed
sharing. Human Factors, 46, 497–517. doi:10.1518/
attention in dynamic fault management. Ergonomics,
hfes.46.3.497.50401
38, 2371–2393. doi:10.1080/00140139508925274
Wickens, C. D. (2008). Multiple resources and men-
Yantis, S., & Johnston, J. C. (1990). On the locus of
tal workload. Human Factors, 50, 449–455.
visual selection: Evidence from focused atten-
doi:10.1518/001872008X288394
tion tasks. Journal of Experimental Psychology:
Wickens, C. D., & Carswell, C. M. (1995). The proximity Human Perception and Performance, 16, 135–149.
compatibility principle: Its psychological foundation doi:10.1037/0096-1523.16.1.135
and relevance to display design. Human Factors, 37,
Yantis, S., & Jonides, J. (1984). Abrupt visual onsets and
473–494. doi:10.1518/001872095779049408
selective attention: Evidence from visual-search.
Wickens, C. D., Hooey, B. L., Gore, B. F., Sebok, A., & Journal of Experimental Psychology: Human Perception
Koenicke, C. S. (2009). Identifying black swans and Performance, 10, 601–621. doi:10.1037/0096-
in NextGen: Predicting human performance in off- 1523.10.5.601
nominal conditions. Human Factors, 51, 638–651.
Yarbus, A. (1967). Eye movements and vision.
doi:10.1177/0018720809349709
doi:10.1007/978-1-4899-5379-7
Wickens, C. D., & McCarley, J. S. (2007). Applied atten-
tion theory. doi:10.1201/9781420063363

276

View publication stats