Psychological Bulletin Copyright 1986 by the American Psychological Association, Inc.

1986, Vol. 99, No. 2, 263-281 0033-2909/86/$00.75

Sustained Attention Under Thermal Stress

P. A. Hancock
Department of Safety Science and Human Factors Department, Institute of Safety and
Systems Management, University of Southern California

This article reviews the effects of the thermal environment on vigilance.A reinterpretation of early
and contemporary studies contradicts the existingnotion that vigilanceis facilitatedin ambient tem-
peratures that exceed a comfortable level. Rather, performance is degraded as thermal homeostasis
of the observer is disturbed. Significantbreakdown in capability becomes manifest with measurable
perturbation to deep body temperature. This assertion requires that conditions induce a dynamic
change, as performance is unaffectedwith no variation in deep body temperature and is facilitated
when the observer is establishedin a static hyperthermic state. An attentional account of this phe-
nomenon is offered, which contrasts with previous arousal explanations.

A common strategy in experimental psychology is to study a (1950). Although he acknowledged a debt to Head (1926) in
particular phenomenon by examining response under the effects relation to the origin of the term, Mackworth's empirical ob-
of some perturbation or adverse condition. This tactic has the servations provided the fundamental foundation on which cur-
added benefit of also providing information about the action of rent research is built (see Davies & Parasuraman, 1982; Davies
some specific stressor or group of stressors on human capabilities. & Tune, 1970; Mackie, 1977; Warm, 1984). Among Mackworth's
Through this approach, a unitary theoretical account of stress initial experiments was a practical examination of the effect of
and performance has come to prominence, in that the single environmental temperature on watchkeeping ability. His findings
mechanism of arousal has been postulated to subsume the action suggested that performance was optimized at an ambient tem-
of many different stressors on a wide variety of abilities. Recent perature exceeding that of typical thermal comfort. This obser-
theoretical and empirical insights appear to belie such a simple vation was replicated by Pepler (1958), who affirmed an inverted
overall interpretation (e.g., Hancock, 1984b; Hockey & Hamilton, U-shaped relation between performance capability and the level
1983), although it should be noted that the disparate action of of thermal stress. Despite subsequent experimental work, current
differing stressors was first recognized over two decades ago assertions concerning sustenance of attention in heat and cold
(Broadbent, 1963). are essentially derived from these seminal investigations. Thus,
More thorough understanding of the action o f stress may be Davies and Tune (1970) indicated that capability in general was
derived from the examination of evidence concerning one par- vulnerable to thermal effects, whereas latterly, Davies and Para-
ticular type of performance and how it varies under the impact suraman (1982) concluded that mild heat facilitates performance,
of one specific stressor (e.g., Hancock, 1981, 1982). Using this and cold impairs monitoring capability.
tactic, I examine both early and contemporary evidence con- Grether (1973) was more specific in stating that optimal vig-
cerning one order of performance, that of sustained attention or ilance occurs at 80 ~ (26.7 ~ on the effective temperature
vigilance, and its variation under the impact of differing thermal (ET) scale, l which is an index of the thermal environment cal-
environments. Previous views concerning the change in vigilance culated from measures of dry-bulb temperature, relative hu-
capability under hot and cold conditions are not affirmed, and midity, and air velocity (Houghten & Yagloglou, 1923). Grether's
the preponderance of experimental evidence suggests a different value was derived from a mean of the two conditions noted by
perspective from which to view the action of thermal stress and, Mackworth (1950) and Pepler (1958) as those in which superior
potentially, a wide variety of other stressors. performance could be observed. The resultant inverted O-shaped
The genesis of coherent study of the phenomena of vigilance function has been used as support for an arousal account of
is attributed appropriately to the now classic work of Mackworth vigilance performance under thermal stress that remains the only
comprehensive explanatory construct of performance variations
noted (cf. Hancock, 1984b; Poulton, 1977). One requirement
The author gratefully acknowledgesthe support of H. H. Hurt, Jr., for a unitary account of vigilance under thermal stress is the
through the auspices of the Safety Helmet Council of America, and the ability to predict the change in performance as reported by var-
Head Protection Research Laboratory, Department of Safety Science, ious existing studies. The arousal position accomplishes this by
University of Southern California. The comments of K. M. NeweU,
identifying differing arousal states. These states are dependent
M. H. Chignell,B. M. House, and two anonymous reviewerson an earlier
version of this manuscript proved most relevant and their contribution on reaction to the level of the particular stressor at hand. Increase
is also gratefullyacknowledged. in performance level is induced by elevation of arousal, whereas
Correspondence concerning this article should be addressed to P. A.
Hancock, Department of Safety Science,Institute of Safety and Systems t Alltemperature valuesare expressedin both Farenheitand Centigrade
Management,Universityof Southern California,Los Angeles,California with the first figure being as in the original report referenced with its
90089-0021. scalar equivalentfollowingin parentheses.

263
264 P. A. HANCOCK

depression of arousal reduces performance efficiency (Poulton, individuals at the termination of the exposure. These observations
1977). suggest that the dynamic perturbation to the deep body tem-
The present approach distinguishes between the thermalstates perature of the subject may be an important element in perfor-
of the observer involved. In the following sections, three basic mance variation.
thermal states are presented. The first is a dynamic state in which, The second tendency that Mackworth noted was that the dec-
because of the environmental thermal load, the participant ex- rement in vigilance efficiency due to the stress was palliated by
periences a constant change in deep body temperature away from the experience level of the subject on watch (see also Hancock,
both a normative level and a steady state. The second is an el- in press). This trend was interactive with both increase in ambient
evated level of body temperature, or hyperthermic state, with temperature and time on watch, and indicated that the more
the distinction that the subject has been stabilized in such a experienced the subject at the task at hand, the less they were
condition. Parenthetically, there are currently no corresponding disturbed by the introduction of the thermal stressor. It is salutary
studies that report on performance under equivalent stabilized to note that this seminal work is, in many ways, still the most
hypothermic states. In the third condition the deep body tem- comprehensive empirical examination of sustained attention
perature of the observer does not vary despite the change in under thermal stress to date.
ambient environment. These three states are distinguished for A subsequent study by Pepler (1953) is usually projected as a
both hot and cold conditions. The synthesis of evidence that replication that provides direct support for the position advanced
follows is based on this tripartite differentiation and is followed by Mackworth. This is an oversimplification. Pepler (1953) con-
by a discussion that examines previous and current theoretical ducted two experiments on the identical clock test but with par-
proposals that might account for the results presented. ticipants who were naturally acclimatized to heat through hab-
itation in a tropical locality. Three ambient conditions, 67 ~
E x p e r i m e n t a l Evidence: Sustained A t t e n t i o n U n d e r (19.4 ~ 82 ~ (27.8 ~ and 92 ~ (33.3 ~ ET were used.
T h e r m a l Stress In an initial experiment, after eliminating the results from six
subjects whose data were confounded by artifactual contami-
Heat: Dynamic Change in Deep Body Temperature nation, Pepler found signal omission was greatest and perfor-
Degrades Performance mance poorest at the median 82 ~ (27.8 ~ ET condition.
In Mackworth's (1950) original experiment, subjects were re- These data were in direct disagreement with the initial findings
quired to monitor a hand on a clock face that made sequential of Mackworth some 7 years earlier. Pepler claimed that the ir-
clockwise jumps at 1-s intervals. A full revolution of the hand regularity of experimental exposures and the equivalency of the
consisted of 100 such jumps and the critical signal for response median condition to the ambient tropical climate in which the
was a temporally irregular double j u m p of the clock hand. Al- work was conducted were mainly responsible for the apparently
though immediate response was requested, subjects were also contradictory results.
instructed to respond if they recalled critical signals to which There is an alternate account of these data that is founded on
they had not responded one or more jumps before. A between- two assumptions. The first assumption is the existence of a prac-
subjects design was used to investigate the effect of four different tice effect on the vigilance task used. This appears reasonable,
heat conditions. Results for response latency, signal omission, given Pepler's observation of a significant Performance • Days
and a combined median response time performance metric in- effect in a second, more thorough, experiment although a null
dicated optimum efficiency at 79 ~ (26 ~ ET compared with effect for practice on the clock task has been reported by Car-
a lower, 70 ~ (21 ~ ET, and two higher, 87.5 ~ (31 ~ and penter (1946). The second assumption is that performance mo-
97 ~ (36 ~ ET, conditions. Performance capability declined tivation is reduced by testing over a period of days with irregular
during the second hour on watch and this propensity, commonly exposures. Again, Pepler provided subjects' evaluative responses,
termed the vigilancedecrementfunction, was exacerbated by in- which indicate that such is the case. These assumptions suggest
creasing the heat above the optimum 79 ~ ET level. that performance is at a maximum toward the middle of any
There are two important elements of this foundational work individual's testing regimen, when initial practice has improved
that have generally escaped notice. First, the optimal performance capability while motivation to perform has not been extinguished.
condition was the only one of the four investigated that did not Empirical data from Pepler's subjects indicate that more than
cause a dynamic change in the deep body temperatures of the 50% of the 18 subjects exhibited this particular trend. In this
participants. Mackworth indicated that his acclimatized subjects experiment, Pepler took care to counterbalance for this order
performed the sedentary task wearing only gym shorts. They effect by using four 6 X 6 Latin square designs for the 24 subjects
were exposed to an air movement of 100 ft/min and experienced tested. However, in eliminating the results for six subjects for
an average 1.0 ~ (0.6 ~ fall in deep body temperature during legitimate artifacts, that is, two subjects for sleeping on watch
the 2-hr exposure at 70 ~ (21 ~ ET. It has been indicated and four for guessing the algorithm for critical signal appearance,
that exposures above 85 ~ (29.4 ~ ET result in a noncom- Pepler inadvertently eliminated a disproportionate number of
pensable rise in deep body temperature, where absolute rate of occasions in which the intermediate 82 ~ (27.8 ~ ET condition
rise is dependent on severity of the heat and length of exposure occurred in the central testing position. In the 18 subjects ana-
(e.g., Grether, 1973; Hancock, 1982; Houghten & Yagloglou, lyzed subsequently, only 3, instead of 6, performed this condition
1923, Figure 5; Lind, 1963). The two highest heat stress condi- while in the middle phase of their testing regimen. This might
tions in Mackworth's experiment, that is, 31 ~ and 36 ~ ET, be responsible for the result of elevated signal omission at the
exceeded this threshold value. Mackworth affirmed body tem- median temperature. In consequence, Pepler's initial contrasting
perature increases by reporting elevated rectal temperatures for result may be due to a problem with subject elimination, rather
 SUSTAINED ATTENTION, TEMPERATURE 265

than equivalence of local thermal condition to one experimental

temperature, as he suggested.
This brief pr6cis of Pepler's first experiment does not exhaust
all the problems concerning the contradictory finding of perfor-
mance diminution at the middle test temperature. For example,
not all subjects exhibited the tendency toward optimum perfor-
mance in the middle of the testing regimen. Pepler achieved the
highest 92 ~ (33.3 ~ ET condition by two dry bulb-wet bulb
combinations with the subsequent addition of local air move-
ment. Consequently, each subject performed on six occasions
rather than three as the number of temperatures suggest. Also,
because of experimental procedures, subjects were not tested at
the same time of day for each temperature, a factor that has
been noted as influential in performance capability (e.g.,
Colquhoun, 1960; Craig, Wilkinson, & Colquhoun, 1981). In
sum, the problems outlined above appear to justify Pepler's de-
cision to reject the results of his initial experiment, which do
not appear subsequently in the final published report (cf. Pepler,
1953, 1958).
The second experiment reported by Pepler (1953) controlled
for many of the extraneous variables of the first experiment.
Eighteen participants were tested at the identical thermal con-
ditions used in the first experiment, but at the same time of day
for three consecutive days. A Latin square design, without elim-
ination of subjects, allowed for control of an order effect (but see
Poulton, 1973). The results for incidence of signal omission af-
firmed those of Mackworth (1950) by indicating optimal per-
formance at the median temperature value, although again the
tendency toward facilitation at the central condition might be Figure 1. Decrement in performance output versusenvironmental effective
due to a range effect (Poulton, 1973). temperature. Performance in the presence of a constant 72-dB noise
stressor. (Data from Viteles & Smith, 1946.)
Consistent with Mackworth's findings, the physiological data
showed that the two extremes of temperature also caused the
greatest perturbation of deep body temperature. Data from these
two foundational experiments allow two interpretations. The first, ~ (30.6 ~ ET condition exceeded the 85 ~ (29.4 ~ ET
which is based on an arousal position, emphasizes the increase threshold value at which complete bodily compensation to the
in vigilance in ambient temperatures that exceed a comfortable increased heat ceases to be tenable (Hancock, 1982). Viteles and
level (see Davies & Parasuraman, 1982; Grether, 1973). However, Smith (1946) confirmed this by reporting that oral temperature
a second interpretation is that vigilance performance is depressed exhibited a most marked increase in this condition, with an ele-
in conditions sufficient to induce a dynamic and noncompensable vation of 1.4 ~ (0.8 ~ on average for the participants tested.
change in the deep body temperature of the observer. There is Despite the potential confounding effect of noise introduction,
a considerable body of evidence suggesting that the latter prop- an effect that is not thought to be great (Hancock & Pierce,
osition describes, more veridicaUy, the limitation of human sus- 1985), these early data appear to support the notion that vigilance
tained attention in heat and cold. This evidence is explored in is reduced in conditions that perturb individual thermophysiol-
further detail below. ogical state.
Although Mackworth's investigation is identified as originating Although the work of Mackworth and Pepler is generally well
work on vigilance, a contemporary study by Viteles and Smith recognized, the experiments of Fraser (1957) have received little
(1946) examined a similar performance task in the presence of recognition. This is the case despite the fact that Fraser is probably
heat and noise. Subjects inspected pairs of numerical series and the first to introduce the notion of event rate as a potential factor
checked those pairs that contained identical digits. The task du- in vigilance efficiency. In his first, practically oriented investi-
ration was 30 min and it was embedded in an exposure totaling gation, the monitoring task was introduced after the establish-
6 hr. The task was performed at 73 ~ (22.8 ~ 80 ~ (26.7 ment of the overall experimental protocol, which was a large-
~ and 87 ~ (30.6 ~ ET, but with a constant noise back- scale investigation of the capability of mine personnel. As a result,
ground of 72 dB on the A-weighted scale. Performance for the Fraser's task took only 15 min to complete and was undertaken
entire group of subjects was scored on the basis of percentage prior to and immediately after heat exposure. Subjects monitored
output and percentage error. Although error remained essentially circles of light, displayed for 400 ms and presented at the rate
constant for this task, percentage output dropped with increasing of one every 2 s. A critical signal was constituted by the ap-
temperature. These data are presented in Figure 1. pearance of a larger ratio diameter circle in the ratio of 3:2 to
Significant decrement occurred at the highest ET condition nonsignal stimuli. There were 10 such signals during each session.
when compared with the two lower temperatures. The latter 87 This experiment had many problems, which cloud any simple
266 E A. HANCOCK

interpretations that might be drawn. First and foremost, the

length of exposure was inversely related to the severity of the
heat imposed. Individual exposures were halted if a preset phys-
iological criterion deep body temperature of 101.8 ~ (38.8 pc)
was exceeded, or if the subject experienced excessive discomfort.
Second, the task was interpolated with other performance mea-
sures and taken prior to and after, but not during, the heat stress
exposure. Third, different personnel collected the experimental
data, and problems were also noted with the test equipment. In
terms of incidence of missed signals, Fraser's data indicated a
progressive increase in the number of omitted responses up to
the middle temperature of the five investigated. However, above
this level, signal omission decreased, but these latter conditions
are those that Fraser specifically noted as being curtailed because
of violations of physiological and subjective tolerance.
In response to these limitations, a subsequent experiment was
conducted under laboratory rather than field conditions. In a
preexposure control session, only 2 of a possible 720 signals were
missed by the 24 participants who each undertook three 10-min
periods on watch. However, as the level of heat stress increased,
the number of missed signals ascended to 2.8% at approximately
100 ~ (37.8 ~ ET. Even using the brief 10-min periods on
watch after heat stress, there was a significant decrement when
performance was compared after the first hour with that after
the second hour of heat exposure. Unfortunately, the interactive
effect of heat against performance in a more comfortable thermal
condition was apparently not examined. Overall, Fraser's results
indicated that heat affects vigilance to brief but rapid stimulus Figure 2. Vigilance errors of omission versus effective temperature for
presentations, even after the subject has emerged from the stress- low- (e), intermediate- (It), and high-complexity(A) monitoring displays.
ful environment. The suggestion is that such decrement is ac- (Data from Carlson, 1961.)
companied by the dynamic body temperature change as reported
in the study (Fraser, 1957). In summary, the early work on vig-
ilance in hot atmospheric temperatures appears, superficially, to efficiency clearly depends on the number of stimulus sources to
support an arousal interpretation of performance results. How- be observed in this work by Carlson (1961).
ever, a more detailed analysis indicates that reduction in efficiency Subsequently, Teichner proposed that the thermoregulatory
is accompanied in each case by an uncontrolled, dynamic change mechanism acted as a general physiological arousal system, and
in the deep body temperature of the observers away from both thus detection efficiency should be related to thermal arousal
a normative and steady state. through the general inverted U-shaped function. Following Tei-
Following these early efforts, Carlson (1961), in work on stress chner's proposal, Arees (1963) reasoned that performance should
and performance, postulated that physiological and psychological be high when the thermal gradient between core and skin tem-
sources of input summated on a single structure or through a peratures was equivalent to that between skin and ambient air
common channel (see also Broadbent, 1971). One prediction temperatures, as in normal circumstances. When the former
from this postulate is that excessive stimulation, regardless of its gradient was small, that is, core and skin temperatures were close
origin, would result in information overload and consequently together as in heat stress conditions, detection efficiency would
in performance degradation. To test this hypothesis, Carlson used be poor. This would also be true of cold conditions when the
two levels of heat, 20 ~ (68 ~ and 33 ~ (91.4 ~ ET, in gradient between core and skin temperatures is higher than that
combination with a monitoring task with three levels of display in the comfortable heat balance condition.
complexity. Although nine subjects participated in the experi- Arees also noted that although variation in environmental
ment, only five completed the full series. The results, which are temperature was the typical way to manipulate thermal arousal,
shown in Figure 2, indicated that the introduction of heat had there are a number of alternate methods that achieve a similar
little effect at the lower and intermediate levels of complexity objective. One of these is to introduce a sudden, high-intensity
but resulted in a large decrement at the highest level of display noise, which causes rapid peripheral vasoconstriction and con-
complexity. sequently increases the gradient between core and skin temper-
Carlson (196 l) also provided evidence of an increase in ob- atures (cf. Dean & McGlothen, 1965). Whether the manipulation
server deep body temperature at the highest heat stress level. of environmental temperature and noise addresses a common
These data were taken as support for the initial suggestion con- arousal mechanism cannot be determined directly. A recent as-
cerning the summation of diverse stimulus inputs and also are sessment (Hancock & Pierce, 1985) has suggested that the actions
consistent with the present position that dynamic body temper- of heat and noise are relatively independent, particularly in re-
ature change degrades vigilance. However, this breakdown in lation to their impact on performance.
 SUSTAINED ATTENTION, TEMPERATURE 267

Arees (1963) investigated this potential interaction by testing

24 subjects on visual monitoring of a lights matrix under three
thermal conditions, 53.5 ~ 66.5 ~ and 89 ~ ET. Within these
temperature repetitions were embedded four groups of six sub-
jects in which a no-alerting noise control group was compared
with those hearing an early noise burst, a later noise burst, or
both early and late noise bursts. For the means of the subjects
tested, using only detection frequency as a dependent measure,
no significant results were found for thermal condition, noise
introduction, or interactions between these two factors. Initially,
this appears to contradict the assertion that vigilance is degraded
when the body suffers dynamic perturbation, particularly at the
89 ~ ET extreme. However, Arees (1963) reported only the
thermal gradient, that is, the difference between core and skin
temperatures. So a direct assessment of core temperature change
was not available. Also, the use of only detection frequency in
the absence of data on error rate and response latency makes a
simple interpretation of these results somewhat difficult. More
important, the use of mean data across groups appears to have
masked some systematic trends.
To examine these trends, Arees did test for a correlation by
individual subject between detection efficiency and thermal strain
or gradient value. He found that 10 oftbe 24 subjects exhibited
significant linear correlations and 5 of the 24 showed significant
curvilinear correlations. These 15 could be divided into three
distinct groups: (a) a group of 5 whose performance increased Figure 3. Latency of responses ( 9 and signal omission (e) for a visual
monotonically with thermal strain; (b) a group of 5 whose per- vigilance task and signal omission (m) for an auditory vigilance task, as
formance decreased monotonically with thermal strain; and (c) a function of performer oral temperature increase. (Data from C. R. Bell,
a group of 5 whose functions approximated the inverted U-shaped Provins, & Hioms, 1964.)
function originally postulated. He proposed that all subjects,
tentatively including those showing no correlational relation,
represented phases of the inverted U-shaped function in which and peripheral (skin) temperatures in influencing performance
either the whole curve or the ascending or descending arm of (see Allan & Gibson, 1979; Goodman, Hancock, Runnings, &
the function was present. This was one of the first observations Brown, 1984).
on the importance of both core and skin temperature in affecting Following this study, two reports appeared that examined the
performance, an approach that has been pursued in more con- effect of specifically increasing body temperature on sustained
temporary work by Allan and his colleagues (e.g., Allan & Gibson, attention. In experiments reported by C. R. Bell, Provins, and
1979; Allnutt & Allan, 1973). Hiorns (1964) the exposure time depended on the level of thermal
There were no recorded performance or physiological effects stress. As the heat level was increased across the different ex-
for the brief noise bursts. Arees suggested that the brevity of posures, the time that the participants could tolerate continuous
effects due to the vasoconstriction, which is on the order of sec- occupation decreased. Because of this variation of exposure time
onds, could not be detected in the performance of the task that by condition, it is difficult to interpret results for both the visual
required only periodic response. Indeed, it can be argued that and auditory monitoring tasks. However, vigilance efficiency
such an effect may not be observed even in continuous tasks plotted against performer's oral temperature, irrespective of the
such as tracking (Dean & McGlothen, 1965). Overall Arees con- time and/or temperature conditions that caused the increase,
eluded some tentative support for the inverted tJ notion relating exhibited certain consistent tendencies that are illustrated in
thermal arousal to performance. However, this interpretation is Figure 3, (cf. C. R. Bell, 1964).
derived from the post hoc analysis of data that indicated a non- Grether (1973) has criticized this study because increasingly
significant main effect for temperature condition on vigilance, high temperatures were undertaken on successive days, which
and was based on the trends of only 15 of the 24 subjects tested. allowed both a practice effect and a potential acclimation effect
These observations suggest that any conclusions drawn from this to confound results. More important however, was the intro-
work are at best tentative and should be treated with some cau- duction of varying noise between 85 and 95 dB, which alone has
tion. been demonstrated to exert some performance effect on this type
However, although Arees' (1963) data present insufficient of sensory vigilance task (Broadbent, 1954). Despite these ob-
physiological detail to affirm the current postulate, they do not jections, the data indicate progressive degradation in both visual
contradict the notion that it is dynamic change to deep body and auditory vigilance with increasing perturbation to the body
temperature that degrades performance. They do indicate some temperature of the observer, as shown in Figure 3. It is interesting
important variations in sensitivity across differing subjects and to note the forms that such functions take. For latency and signal
the eventual necessity to consider the role of both core (deep) omission, the function for visual sustained attention appears to
268 R A. HANCOCK

Table 1 from Colquhoun's work (Colquhoun, 1969, Figures l and 2), it

Response Time and Signal Omission as a Function of Ambient can be seen that the best performance was elicited under those
Temperature and Change in Deep Body Temperature conditions that altered body temperature the least. Thus signal
omission was 52%, 52%, and 56% and response latency 0.86s,
Effective temperature (~
0.80s, and 0.85s, respectively for the three ascending conditions
Subject responses 67 82 92 reproduced in Table 1. It should be recognized that these data
represent group mean performance, and individual conformation
Change in deep body to this principle requires a more detailed analysis of each indi-
temperature (~ -0.49 -0.30 +0.53
Response time (ms) 860 800 840 vidual subject's response (Arees, 1963).
Missed signals (%) 52.0 52.0 56.0 The importance of the effect of time on watch was demon-
strated by Mortagy (1971). In a practical investigation directed
Note. Data from Colquhoun (1969). toward problems in industry, the effects of three differing lengths
on watch were examined. Watches of 20, 40, or 60 min w e r e
undertaken by subjects in 74 ~ (23.3 ~ 82 ~ (27.8 ~ or
represent a positively accelerated exponential function. However, 90 ~ (32.2 ~ ET conditions. Results of this investigation are
for auditory vigilance, the dramatic increase in signal omission presented in Figure 4, and clearly illustrate that prolonging the
occurs at lower perturbations to deep body temperature. The period on watch has little effect at the lower room temperatures.
general function as expressed in the data for visual vigilance (see However, when the thermal surround exceeded the threshold
also Benor and Shvartz, 1971) has been taken as support for a value at which the individual can accomplish thermal equilibra-
position relating stress to the notion of attentional resources tion, that is, the 90 ~ ET exposure, then performance decreased
(Hancock & Chignell, 1985), which I explore later. (cf. Mortagy & Ramsey, 1973). This observation confirms the
Given the growing consensus of data from early studies con- suggestion that the value of the environmental temperature is
cerning body temperature change and impairment in vigilance not the key factor in dictating whether breakdown will occur.
efficiency, a study by Colquhoun (1969) provides an important
set of apparently contrasting data. He used three differing event
rates and three levels of ambient temperature: 67 ~ (19.4 ~ 15
82 ~ (27.8 ~ and 92 ~ (33.3 ~ ET. It should be recognized
that the latter condition exceeded the 85 ~ (29.4 ~ ET thresh-
old previously noted, and the length of temporal exposure was
not so brief as to curtail the increase in deep body temperature
as in some other experiments (e.g., Dean & McGlothen, 1965;
Poulton & Kerslake, 1965). Indeed, Colquhoun's subjects per-
formed for 2 hr on watch, and physiological recordings indicated
a slight drop in rectal temperature at the two lower conditions
and a rise at the highest heat stress condition (see Table 1). Despite 1
these physiological observations and clear main effects for both
event rate and successive sessions on watch, there were no dec-
rements due to the effect of ascending ET. Colquhoun, aware
that his results were in contrast to previous investigations, pos-
tulated six potential factors to account for such a discrepancy.
These were: differences in event rate, signal probability, accli-
matization status, type of subject, nature of the task, and pro-
longed exposure time.
|
However, from the present synthesis, two alternative factors
may be implicated. First, Hancock (in press) has indicated that
a key element in performance under stress is the level of ability
exhibited by the participant prior to the stress exposure. In the
case of Colquhoun's subjects, considerable practice was under-
taken prior to experimental sessions. This may have contributed
to the relatively small performance changes found, because ex-
perienced subjects appear to be better able to resist the detri-
mental effects of heat compared with their unskilled peers. Sec- 20 40 6'0
ond, and perhaps more important, the degree of perturbation
from normal body temperature must be considered in detail as Lengthof IndividualWork Periods
is suggested from observations in previous studies. Actual mean
changes in body temperature were -0.49 ~ (-0.27 ~ -0.30
~ (-0.17 ~ and +0.53 ~ (0.29 ~ for the 67 ~ (19.4 ~ Figure4. Signal omission versus length of period on watch at 74 ~ (A),
82 ~ (27.6 ~ and 92 ~ (33.3 ~ ET conditions, respectively. 82 ~ (Ill, and 90 ~ (O), effective temperature. (Data from Mortagy,
When mean data for overall latency and signal omission are taken 1971.)
 SUSTAINED ATTENTION, TEMPERATURE 269

Rather, it is the combination of temperature with exposure time,

sufficient to change deep body temperature, that affects perfor-
mance.
Several investigations of the effect of heat stress on vigilance
have been generated in response to the practical problems of
work in thermally stressful conditions. Among these is an em-
pirical analysis of augmented cooling reported by Benor and
Shvartz (1971). Seven healthy subjects performed an auditory
vigilance task in conditions up to 50 ~ (122 ~ dry-bulb tem-
perature while walking on a treadmill at a rate of 3.5 km/hr. In
their comprehensive study, Benor and Shvartz recorded detailed
physiological measures of both skin and rectal temperature. With
augmented cooling these physiological parameters did not vary
as environmental temperature increased, and under these con-
ditions neither detection rate nor false reactions changed across
environmental temperatures.
Without augmented cooling a different pattern of results
emerged. First, the no-cooling tests were terminated if subjects'
rectal temperatures exceeded 39.0 ~ (102.2 ~ or if subjects
reached exhaustion before the prescribed 2-hr period on watch
had elapsed. Due to this restriction, the time spent in any en-
vironment varied inversely with the severity of the thermal con-
dition. In order to compare performance across exposures, a
derived value of percentage missed signals per minute on watch
has been extracted from the data originally presented by Benor
and Shvartz ( 1971). Also, because of the problems of the inertial
interval, in which brief temporal exposures have relatively little
effect on core body temperature (e.g., Hancock, 1984a), a com-
bination of reported rectal and skin temperatures as recom-
mended by Burton and Edholm (1955) has been calculated as a Figure 5. Percentageof missed signalsper minute of exposure as a function
measure of total bodily temperature increase in this case. These of increase in mean body temperature. (Mean body temperature was
synthetic data are presented in Figure 5. The general pattern is derived from weighted values of core or rectal temperature [TR], and
consistent for both missed signals, as shown in Figure 5, and skin temperature was derived from several surface sites ['i's]. Data from
false reactions. Benor & Shvartz, 197l.)
These data are of particular interest as they not only affirm
the present postulate that sustained attention is degraded as body
temperature changes in response to the external thermal envi- decrease in the control condition was from 2.4 to 2.2. In terms
ronment, but they also suggest the form such degradation takes of absolute decrease, d' reduced from 2.49 in the comfortable
with progressive deviation from a normal bodily thermal state thermal condition to 2.12 in heat which represented a significant
(cf. C. R. Bell, Provins, & Hiorns, 1964). This function appears reduction (Poulton, 1977). However, the authors attributed this
to take an increasing exponential form and would be consistent result not to the effects of heat, but to an uncontrolled transfer
with a positive feedback system, which itself is the manner in effect favoring the control condition.
which body temperature increases under the driving force of a In the same series of experiments on stressor interactions,
noncompensable thermal condition. This is seen under the impact Poulton and his colleagues (Poulton & Edwards, 1974b; Poulton,
of heat disorders such as heat stroke. The potential isomorphism Edwards, & Colquhoun, 1974) examined the effect of increased
between breakdown in performance and physiological function- temperature on the Wilkinson (1969a) auditory vigilance test.
ing is explored briefly in a later section, however, a fuller dis- In the first experiment, Poulton and Edwards (1974b) examined
cussion may be found in Hancock and Chignell (1985). the performance of 12 subjects in a comfortable 19 ~ (66 ~
As with some previous experiments, the effect of heat on vig- ET and an elevated 34 ~ (93 ~ ET condition. Results indicated
ilance has been noted as an adjunct to a fuller investigation of that in heat, subjects responded more quickly but less accurately,
stressor interactions (e.g., Dean & McGlothen, 1965; Viteles & with a greater incidence of false detections. The latter experiment
Smith, 1946). Poulton and Edwards (1974a) reported the effect (Poulton et al., 1974) affirmed this pattern although at a slightly
of increasing heat level to 34 ~ (93 ~ ET on an analogue of lower (33 ~ [92 ~ ET. The same number of subjects were
the Wilkinson visual vigilance test (Wilkinson, 1969a). The tested although, because of equipment problems, results for only
comparable control was a 19 ~ (66 ~ ET condition. The 10 individuals were analysed. Comparative results for these ex-
ability to detect signals for the first versus the last 7.5-rain periods periments are given in Table 2. When performance on each mea-
of a total 30-rain exposure was compared. In the heat, the signal sure was broken down for sequential 7.5-rain periods on watch
detection metric d' decreased from approximately 2.2 to 1.9 from for the latter experiment, the pattern appeared as represented in
the first 7.5 min to the last 7.5 min, whereas the comparable Figure 6. This illustration is constructed so that poor performance
270 a . A . HANCOCK

is represented by ascending any of the three performance scales Table 2

presented. Sustained Attention Performance as a Function of
The results from the latter two experiments using auditory Environmental Effective Temperature
detection indicated a significant reduction in the percentage of
signals detected. Collectively, the overall results indicated the Effective Response Signal False
temperature time detection detection
effect of heat on vigilance was to reduce d', the index of the (~ (ms) (%) (%) d'
ability to detect signals. In none of the above experiments did
Poulton and his colleagues indicate the effect of heat on observer 19a 640 67.8 0.43 3.28
body temperature. Clearly, the 34.4 ~ ET exposure exceeded 33 610 51.3 0.68 2.63
19b 760 65.3 1.46 2.58
the 29.4 ~ ET threshold at which body temperature is disturbed. 34 700 55.3 1.61 2.27
Although the vigilance task lasted only 30 min, the time spent
in the heat marginally exceeded 90 min, of which the last one- Note. The values for d' are near approximations taken from illustrations
third was spent in vigilance performance. In consequence it may in the two cited articles.
be inferred that the subjects in these experiments experienced a Data from Poulton, Edwards, & Colquhoun (1974). b Data from Poulton
& Edwards (1974b).
an increase in body temperature (although see Poulton, 1977,
p. 436). This suggests that the overall change in performance
with perturbation in body temperature may be one of a d' re- impact needs to be clearly demonstrated across a variety of sen-
duction. sory systems that may be used for vigilance.
Poulton (1977) also noted that there is a sizeable difference In additional experimentation, Harminc (1976) examined the
in false detections of visual compared to auditory signals. This effect of dry-bulb temperatures ranging from 22 ~ (71.6 ~ to
may be due to the coarse-grained action of heat in that sweating 44 ~ (111.2 ~ on a vigilance task. Significant decrement for
is more likely to interfere with visual than the auditory assimi- signal omission with heat was reported. Inefficiency in vigilance
lation of both signal and nonsignal stimuli. Concerning inter- capability (e.g., Harminc, 1976, Figures 2 and 3) was related to
ference with vision, it is also possible that variation in visual change in deep body temperature (e.g., Harminc, 1976, Figure
acuity with change in core body temperature (Hohnsbein, Pie- 9) in a manner consistent with the present postulation.
karski, & Kampmann, 1983; Hohnsbein, Piekarski, Kampmann, The practicality of the problem of vigilance under thermal
& Noack, 1984) is influential in the reduction of vigilance. How- stress was elaborated by Mackie and O'Hanlon (1977) in a study
ever the impact of dynamically changing body temperature on of extended driver performance in heat. They indicated that
sustained attention does not appear to be specific to any particular driving a motor vehicle is one of the most prolonged attention-
sensory modality. Therefore, if change in sensory acuity with demanding tasks to which the average person is exposed (Brown,
body temperature is postulated as an important influence, its 1967). Vehicle operation under tedious conditions for extended
periods of time falls under the rubric of operational vigilance.
In a within-subject design, l0 men and l0 women were required
to drive a 600-km route once in conditions of 67 ~ (19.4 ~
5.0] 1.0. 100 on the Wet-Bulb, Globe Temperature (WBGT) heat stress scale
'90 and once in an elevated 90 ~ (32.2 ~ WBGT condition. 2 All
exposures took place during the day and the order of exposure
'80 was counterbalanced across subjects. Several physiological mea-
sures were recorded, including sweat rate, corticosteroid excre-
.70
tion, blood pressure, electroencephalogram, and electrocardio-
laot ..6. '60 gram measures. However, oral temperature was measured only
at the beginning and end of each driving segment and a general
.50 .~ trend rather than a time-related function was consequently ob-
"40 served (see Figure 7). Several measures of performance and sub-

ii"
jective reactions of the drivers were analysed. Measures of self-
alertness and fatigue indicated progressive decrement with time,
although some conflict appeared between what is reported in
1.o1 the text and the illustration for the problem of fatigue (cf. Mackie
& O'Hanlon, 1977, Figure 14).
0.:] .1o ", L ,.,.,
0
2 The Wet-Bulb, Globe Temperature Index, first proposed by Yaglou
O" 7.'5 13 22:5 30 and Minard (1957) synthesizes air temperature, relative humidity, air
movement, and radiant heat. It is a superior index for assessingheat load
MinutesFromStsrt with respect to performance capability, as noted in this article. However,
Figure6. Falsedetections (A, A), responsetime (O, O), and signalomission typically, radiant heat is not recorded in experiments examining such
(lq, O) for 19 ~ effective temperature (ET; solid line) and 33 ~ ET performance. Consequently, the more common but less appropriate ef-
(broken line) conditions by sequential 7.5-rain periods on watch. (Data fective temperature scale has been used, particularly where such values
from Poulton, Edwards, & Colquhoun, 1974.) appear in the original report.
 SUSTAINED ATTENTION, TEMPERATURE 271

also Hancock, in press). Consequently, they examined the ca-

pability on a military fire command task of naive individuals
who undertook prior standardized practice. Motivation to per-
form was sustained at a high level. Subjects were exposed for 7
hr to a 33.3 ~ ET condition, and in this practical investigation,
the effect of altitude up to 4,300 m was also analyzed. During
the 4th hr of the experiment a rest period was imposed, and in
the 1st, 3rd, 5th, and 7th hr, fire commands were communicated
at the rate of 30/hr. In order to simulate relatively inactive periods,
Fine and Kobrick reduced the rate of communication to only 5
fire commands/hr during the 2nd and 6th hr. Performance during
these hours best reflects the problem of sustained attention. Re-
suits indicated that whereas errors remained fairly constant at
approximately 10% in the control condition, errors increased
from 10% to 26% from the 2nd to the 6th hr in the heat.
Initially, the lack of increase in errors in heat at the 2nd hr
appears somewhat in contrast to the present assertion. Although
physiological data were not presented, the inference from pre-
vious indications is that body temperature would be elevated
after 2 hr in a 33.3 ~ ET environment. The reason for this
initial lack of decrement is probably related to the sustained
level of motivation of the individuals concerned. The precise
effect of this important factor is deserving of more thorough
Figure 7. Frequency of steering wheel movements greater than 10~
( 9 [3) in magnitude and increases in oral temperature (O, II) versus empirical analysis. The failure during the 6th hr is in accord
elapsed time of vehicle operation. (Circles = 67 ~ Wet Blub Globe Tem- with the present proposal, although again the ability to match
perature [WBGT] Index, squares = 90 ~ WBGT. Data from Mackie & such decrement with change in body temperature is not directly
O'Hanlon, 1977.) available from the results presented in this study.
A final study in which vigilance performance has been related
to body temperature change induced by heat stress has been
In the measures of performance that were recorded, the fre- reported by Epstein, Keren, Moisseiev, Gasko, and Yachin (1980).
quency of large-scale (> 10 ~ excursions of the steering wheel A sustained attention component of performance was incorpo-
movement, reflective of recoveries after momentary lapses of rated into an overall complex psychomotor task that was per-
attention, are somewhat analagous to the failure to detect a critical formed in ETs of 21 ~ (69.8 ~ 30 ~ (86 ~ and 35 ~
signal in a vigilance paradigm. These excursions decreased slightly (95 ~ Participants were able to stabilize body temperature
with the first half of travel in both the comfortable and heat with little increase in the lower two temperature conditions.
stress conditions. However, during the second half of travel, the However, the 35 ~ ET condition induced a dynamic and non-
frequency of such movements increased and this tendency was compensable increase in core body temperature (see Epstein et
magnified by the introduction of heat. This trend is illustrated at., 1980, Figure 2). In terms of performance, subjects traded
in Figure 7. speed for accuracy in the two lowest temperatures such that the
This pattern was confirmed for alternate measures of perfor- slow but accurate performance at 21 ~ was contrasted with
mance such as the variation in driving speed, the frequency of faster but less accurate performance at 30 ~ ET. At the highest
technical driving errors, and the rate of measurable lane drift. heat stress condition there was a general deterioration in overall
In summary, the authors concluded that drivers exposed to hot performance that was both slow and inaccurate. These data pro-
environments exhibited decreased precision in steering control vide substantive support for the assertion that performance
and an increased propensity toward error during the first 150 breakdown is observed when the individual manifests a dynamic
min of exposure, compared with performance when driving in and noncompensable change in deep body temperature.
thermally comfortable conditions. These differences continued
to increase with progressive temporal exposure. Consequently, Heat: No Change in Performance With No Change
these data support the current notion concerning body temper- in Body Temperature
ature and the sustenance of attention, but, because of limitations
of the experimental protocol, these results add relatively little to Whereas the above studies attest to the notion that perturbation
the detail concerning the overall function that such a relation results in vigilance decrement, there are several reports of no
might possess. performance change, and these are characterized by a lack of
Fine and Kobrick (1978) suggested that certain equivocal re- variation in core temperature. Among the first, Loeb and Jean-
suits that have been observed in performance under heat (see theau (1958) conducted a practical experiment on the effect of
Jones, 1970, for a review) might be due to a low level of subject three different stressors that were imposed either singly or in
motivation and the relative artificiality of laboratory-generated combination. On a task similar to Broadbent's (1953) monitoring
tasks. Fine and Kobrick were also most sensitive to the potential of 20 dials, the effects of heat, noise, and vibration stress were
effect of practice on the task prior to heat stress exposure (see evaluated. It is of particular interest that Loeb and Jeantheau
272 E A. HANCOCK

Condition interaction on performance (e.g., Mackworth, 1950),

the study does not allow inferences concerning the absolute level
of performance change. Second, the monitoring task was inter-
polated among other performance tasks of interest and lasted
only 20 min on each of the two occasions given. Finally, moni-
toring was of an auditory rather than visual task, and the modality
of attention is potentially an important variant with respect to
absolute performance level, level of decrement, and subsequent
recovery during and after exposure to an atypical thermal con-
dition (Poulton, 1977). In sum, Fine et al. (1960) questioned the
ubiquity of heat-induced vigilance decrement, but methodolog-
ical inadequacies, particularly the lack of data for body temper-
ature, clouds any interpretation that may be made.
Edholm (1963) reported a pr6cis of a field study in which the
effects of natural and artificial acclimatization were compared
with performance of a nonacclimatized group. Eighteen subjects
in each group experienced a 12-day desert exposure. Assessments
of both physiological and performance capabilities were recorded.
Among the tests was a vigilance task, although precise details of
the procedure were not reported. Edholm indicated that vigilance
did not vary across the three groups when control performance
was compared with ability under desert conditions. As precise
Figure 8. Environmental effectivetemperature versus omission error (X), details of decrement were not given, a thorough assessment of
commission error (O), and response time (O) in radar monitoring. (Solid the impact of heat is not directly available.
line: data for the presence of 70-dB noise stressor; broken line: 1l0 dB Stronger evidence for the dependency of change in performance
noise stressor.) on variation in body temperature comes from the work of Dean
and McGlothen (1965). In their work on the interaction of
noted no effect on performance with the addition of heat ranging stressors, they had occasion to observe vigilance under extremes
from 110 ~ (43.3 ~ to 125 ~ (51.7 ~ dry-bulb temperature of both heat and noise. Their data, which are reproduced in
and 4%-24% relative humidity. It has been observed (C. R. Bell Figures 8 and 9, indicated no effect for increasing ambient tem-
& Provins, 1962) that the comparison control condition was re- perature up to 93.0 ~ (33.9 ~ ET.
corded at night, and diurnal variation in performance has already
been noted as an important influence (Colquhoun, 1960). In
consequence, it is possible that nighttime control performance
was somewhat depressed with respect to comparative daytime
performance, a factor potentially responsible for the lack of main
effects. However, no variation in body temperature was recorded
in association with this lack of impact on vigilance. Despite these
observations. The overall study indicated the practical decrement
to be expected in a multiple stress condition, although the con-
clusion concerning heat alone remains somewhat equivocal.
The specific effects of introducing humidity into both mod-
erate and high-temperature environments were analyzed by Fine,
Cohen, and Crist (1960). They used an auditory discrimination
task in which three successive tones were presented. Subjects
were required to report which of the first two tones matched the
final tone, when the atypical tone represented one of four different
values along any of the three factors of intensity, frequency, and
duration. There were no significant effects for exposure at 93 ~
(33.9 ~ ET for the 6.5-hr exposure on the auditory discrimi-
nation task. However, there are several factors that may have
accounted for this lack of decrement. First, the auditory stimulus
characteristics were altered by the varying environmental con-
ditions, and hence the comparison of performance between dif-
fering temperatures was not a viable one. Rather, a discrepancy Figure 9. Environmental effectivetemperature versus commission error
score between performance after initial exposure and just prior ((3) and responsetime (e) in meter monitoring. Rectal temperature change
to exiting the environment was used. Although these data sug- for both tasks is also shown (lit).(Solid line: data for presence of 70-dB
gested comparable decrements across all four thermal conditions noise stressor; broken line: 110-dB noise stressor. From Dean & Mc-
investigated and belied earlier observations of a Time • Thermal Glothen, 1965.)
 SUSTAINED ATTENTION, TEMPERATURE 273

Initially this result appears in direct contradiction to the pre-

viously stated observations of reduced capability in heat. This
is particularly true as the highest temperature was well in excess
of the 85 ~ ET threshold as explored earlier. The key factor in
this study was the time of exposure: a maximum of 30 min, of
which only the last 20 min were used to assess performance.
This length of exposure was insufficient to induce a measurable
change in deep body temperature and has been described by
Hancock (1984a) as an inertial interval, in which the tempera-
ture, however high, has insufficient time to perturb deep body
temperature (cf. Dean & McGlothen, 1965, Figure 12) and Figure
9. This short exposure time is probably also responsible for the
plethora of null results reported in the study even in the face of
such stressful conditions as 110-dB noise in combination with
33.9 ~ ET. Therefore these data strongly support the suggestion
that it is not the actual ambient temperature that influences per-
formance, but instead it is the combination of exposure time
and temperature sufficient to perturb deep body temperature
that degrades vigilance.
There are a variety of studies that have incorporated vigilance-
type tasks into an overall performance battery to be used to
assess the effects of stress. Among these, Chiles, Iampietro, and
Higgins (1972) reported the effect of 34.4 ~ (94.0 ~ ET on
monitoring over a 75-min interval. They found no significant
difference in vigilance between the heated and control environ-
Figure 10. Response latency and signal recognition versus elevation in
ments. Although both the temperature and the exposure time observeroral temperature. Zero change = 36.5 ~ deep body temperature.
were sufficient to induce a change in deep body temperature, (Data from Wilkinson, Fox, Goldsmith, Hampton, & Lewis, 1964.)
the findings do not contradict previous observations of decre-
ment. This is because the monitoring task was always performed
in association with another task and often with more than one ature and 100% relative humidity. After the subjects had sustained
additional task. Also, performance periods of 15 min or less increases of 0.8 ~ (1.4 ~ 1.4 ~ (2.5 ~ and 2.0 ~ (3.6
were recorded, in which relatively few signals were presented. ~ in sublingual temperature, they were removed from the
These findings do follow their previous observations (Iampietro, heated room, and the increase was maintained by the use of an
Chiles, Higgins, & Gibbons, 1969; Iampietro et al., 1972), which impermeable suit through which heated air was circulated at
have been incorporated by Hancock (1982) into an overall picture periodic intervals. Performance at each hyperthermic state was
of performance in heat. Paradoxically, vigilance appears to be assessed for both auditory vigilance and a mental addition test.
disturbed earlier and more substantially by changes in deep body Results for auditory vigilance are reproduced in Figure 10.
temperature than is performance on multiple and complex tasks. These data show that reaction to a 0.8 ~ increase was slower
These latter observations provide one example of this interesting but more accurate, which reflects a more conservative strategy
phenomenon. caused by the hyperthermic level. Above this value, latency de-
creased while accuracy increased, indicating a true gain in per-
Heat: Static Hyperthermia Improves Performance formance efficiency. The key difference between this thermal
manipulation and previous studies (e.g., C. R. Bell et al., 1964)
As can be seen from the preceding argument, the term elevated is the establishment of a static hyperthermic state that lacked
body temperature is somewhat of a misnomer in that it gives no the stress associated with the constant body temperature increase
indication of the individual's dynamic thermal state. Thus an of other experiments. Although this is an important finding,
absolute temperature value does not designate whether an in- Wilkinson et al. (1964) also reported this somewhat incongruous
creasing, decreasing, or static condition exists. The arguments result for mental addition, which was adversely affected by hy-
presented above have suggested the importance of this dynamic perthermia. This appears to cast doubt on the generality of static
aspect in that it can help resolve the apparently contradictory hyperthermia as a beneficial state.
results that have included both improvement and decrement in In direct response to the anomalous findings of Wilkinson and
vigilance with increased body temperature. Improved perfor- his co-workers, and a previous failure to find any affect for tem-
mance has only been observed when the subject has been sta- perature on vigilance (i.e., Colquhoun, 1969), Colquhoun and
bilized in a static hyperthermic state. Details of these studies are Goldman (1972) conducted a study on performance under con-
presented below. trolled hyperthermia. Their subjects exercised in 103 ~ (39.4
Perhaps the first to report this seemingly anomalous perfor- ~ dry bulb and 93 ~ (33.9 ~ wet bulb. Each 10-rain exercise
mance increase, Wilkinson, Fox, Goldsmith, Hampton, and period caused a 0.4 ~ (0.2 ~ increase in deep body temper-
Lewis (1964), exposed 12 subjects for different periods of time ature. This gave a control condition and static elevations of 100.4
to an ambient condition of 43 ~ (109.4 ~ dry-bulb temper- ~ ~ 100.8 ~ (38.2 ~ 101.2 ~ (38.4 ~ and 101.5
274 P. A. HANCOCK

~ (38.6 ~ for the differing exercise periods. Using a visual

vigilance task, measures of detection frequency, latency, and false
responses were recorded. Unlike the facilitation reported by
Wilkinson et al., no effect was found for elevation of body tem-
perature. However, when the percentage of signals detected were
divided on the basis of subjective confidence in response, results
indicated some affirmation for the notion of improvement in the
category of highly confident response. It should be noted, however,
that in absolute terms the increase was only 5% between the two
physiological extremes (Figure 11).
As a result of this increase in "certain" responses with core
body temperature increase, Colquhoun and Goldman (1972)
suggested that ascending body temperature influenced decision
criterion. Although this is a reasonable conclusion, a number of
other factors may have influenced this finding. First, Colquhoun
and Goldman's subjects were highly familiar with the perfor-
mance task and this is an influence on performance under heat
stress (Hancock, in press). Second, a visual rather than auditory
vigilance task was used (cf. Wilkinson et al., 1964) and it seems
reasonable to suggest that modality of stimulus presentation may
have a considerable effect on performance in that vision may be
more affected by the coarse-grained actions of heat, such as
sweating, than is audition. Finally, performance in Colquhoun
and Goldman's study occurred in the presence of others, a factor Figure11. Percentage detection of "certain" signals versus observer rectal
that appears to influence performer efficiency (Warm, 1984). temperature. (Data from Colquhoun & Goldman, 1972.)
Although the above considerations merit some attention, the
overall conclusions of Colquhoun and Goldman appear sub-
stantiated in the present assessment. They indicated that it is based on body temperature, physiologicaladequacy, and attention
only when the body temperature increased that performance at constructs. For arousal it is sufficient to equate arousal state with
a vigilance task altered noticeably in hot conditions and, further, performance variation. However, this simple tautological view
that for highly trained subjects, such an increase must be con- disserves some of the more substantial theoretical contributions
siderable before such an alteration occurs (Hancock, in press). that a thorough understanding of the notion of behavioral arousal
The last of three studies reporting performance facilitation may supply. In sum, evidence supports the contention that static
used only a brief 18.5-min exposure. In this study, Poulton and hyperthermia benefits sustained attention.
Kerslake (1965) asked their subjects to perform a combination
of visual and auditory monitoring in parallel. As in certain pre- Cold: Dynamic Change in Deep Body Temperature
vious experimental investigations, the use of repetitive exposures Degrades Performance
resulted in asymmetric performance transfer that obscured the
results obtained (see Poulton, 1977). However, an overall facil- The number of studies reporting on vigilance in reduced am-
itating effect for initial entry into the 30 ~ (86 ~ ET envi- bient temperatures is particularly small compared with heat stress
ronment was reported. Physiological evidence indicated raised counterparts. One of the first studies of cold was conducted by
forehead skin temperature, but measures of deep body temper- Kissen, Reifler, and Thaler (1964), whose central concern was
ature were not taken. Poulton and Kerslake speculated than by the use of hypnosis in controlling thermoregulatory responses
the time a significant performance change was noticed at ap- to reduced environmental temperature. However, embedded
proximately 12 rain, rectal temperature would probably have within this investigation was an examination of visual vigilance
fallen. They buttressed this argument with reference to the work in 4 ~ (39.2 ~ air temperature for a 1-hr period. Participants
of Bazett (1949). However, this is not based on direct observation were required to distinguish randomly displayed matched pairs
and does not agree with physiological data recorded under com- of a visual pattern from unmatched pairs. These patterns were
parable conditions (Dean & McGlothen, 1965). These data do, presented at the background rate of approximately 1/s through-
however, indicate some facilitation for immediate entry into a out the experiment. Results suggested that the hypnotic state
heated environment, but whether this is due to the complex na- reduced the physiological stress of cold. However, of more rel-
ture of the performance required, the change in physiological evance to the current work are the data concerning performance,
state of the observer, or the impact of the heat per se cannot be as expressed in Figure 12.
directly determined. These data do indicate an interesting As Figure 12 shows, there is a progressive deterioration in
anomaly that needs further investigation, but do not contradict performance with increasing exposure to cold. This is reflected
the postulate that static hyperthermia is beneficial for vigilance in the number of correct identifications and correspondent signal
performance. omissions. It should be noted that the incidence of commission
Typically, reports of improved performance under thermal errors did not vary with exposure time. The overall performance
stress have proven difficult to encompass for theoretical positions decrement is accompanied by a progressive reduction in core
 SUSTAINED ATTENTION, TEMPERATURE 275

temperature that follows the characteristic transient increase on The most recent investigation of the effects of cold on vigilance
immediate exposure to cold (Figure 12). These data suggest, was conducted by Angus, Pearce, Buguet, and Olsen (1979). They
therefore, that sustained attention decrement is related to dy- used a Wilkinson visual vigilance task and compared baseline
namic deep body temperature change, although in the above data with that obtained during 16 days at an ambient temperature
case it is a reduction in core temperature level. This finding is ranging from 0 ~ to 5 ~ (32 ~ OF) in Arctic conditions.
consistent with information presented previously on performance The vigilance task was performed in 40-min sessions on altemate
under dynamic body temperature decrease as given in the reports days, with one 2-day break around Day 8. Results indicated an
of Mackworth (1950) and Pepler (1953, 1958). immediate effect on signal omission. This increased by 27% from
Following the work of Kissen and his colleagues, Poulton, the final day of baseline measurement to the first experimental
Hitchings, and Brooke (1965) conducted a practical evaluation day. The authors speculated that the deprivation of rapid eye
of lookout efficiency in Arctic 28 ~ (-2.2 ~ and a slightly movement (REM) sleep that accompanied initial cold exposure
more temperate 35 ~ (1.7 ~ climatic conditions. There were was related to the decrement observed. Although body temper-
considerable practical and methodological problems because of atures were measured during sleeping hours, there was no in-
the action of wind and rain during one watch period. These dication of their level during actual performance; therefore it is
problems were accompanied by uncontrolled performance difficult to assess the role of body temperature in this case, as
transfer between conditions for the two groups examined. Despite an indication of time-related variation in performance with ther-
these contaminants, the authors recorded a 1.2 ~ (2.16 ~ fall mal state was not reported. The data are interesting however, in
in oral temperature during the exposure to the lower air tem- that recovery of performance during sequential days was indi-
perature. They concluded that the reduction in performance ef- cated, which may be due to a task-practice effect. This progressive
ficiency of watchkeeping in the cold may be correlated with the recovery might also have been related to the sequential decrease
dynamic decrease in body temperature. Although this is not in the impact of the environment on thermal state, although this
strong evidence, it is in clear agreement with previous observa- remains a somewhat speculative suggestion. The authors' obser-
tions (e.g., Kissen et al., 1964; Mackworth, 1950; Pepler, 1958). vations on REM sleep loss also deserve consideration. Results
indicated total recovery of vigilance to baseline levels after 16
days, whereas REM recovery stood at 75% of preexposure levels.
The simple correlation between measures of REM and perfor-
mance was not significant. This indicated that REM loss cannot
be the sole factor involved. However, using a multiple correlation
technique, the authors calculated that over 90% of the variation
in the vigilance performance could be accounted for by three
factors: percentage of REM recovery, task days, and temperature.
Decline in vigilance efficiency with dynamic perturbation to deep
body temperature is suggested by the overall conclusions of the
study.

Cold: No Change in Performance With No Change in

Deep Body Temperature
As with the previous section, there are few studies in which
the effects of reduced ambient temperature have been examined.
In one exception, Teichner (1966) investigated prolonged visual
detection at 55 ~ (12.8 ~ and 80 ~ (26.7 ~ dry-bulb tem-
perature with a constant 40% relative humidity background. Two
groups were compared that varied in their response to cold con-
ditions as measured by cold-induced vasodilation or the so-called
Lewis wave. These air temperatures were insufficient to change
the deep body temperatures of either group in the experiment
(see Teichner, 1966, Figure 3). Although some intergroup dif-
ferences were observed in performance, means for percent de-
tection and response speed did not vary consistently across the
two temperature conditions. These results suggest using caution
in simple interindividual generalizations (see also Arees, 1963).
However, they support the present contention that consistent
vigilance performance decrement is observed only when deep
body temperature is perturbed dynamically rather than by the
mere manipulation of the thermal environment per se. This study
Figure 12. Errors of commission (0), errors of omission (O), correct is in accord with other reports noted in the comparable section
responses (11),and change in rectal temperature (• versus exposure time on heat, in which the elevation of the ambient temperature, which
at 4 ~ (Data from Kissen, Reifler, & Thaler 1964.) did not affect deep body temperature, failed to influence vigilance
276 E A. HANCOCK

efficiency (Dean & McGlothen, 1965; Edholm, 1963; Fine et al., variation in vigilance under the impact of thermal stress, there
1960; Loeb & Jeantheau, 1958). have been three basic proposals. These are physiological adequacy,
behavioral arousal, and attentional-resource capacity. The utility
Cold." Absence of Experimental Results Under of these three approaches is evaluated in light of the insights
Stabilized Hypothermia revealed by the preeeeding differentiation of performance based
on the thermophysiological state of the observer involved.
There are apparently no studies in which vigilance perfor-
mance has been specifically monitored while the subject was Physiological Adequacy
stabilized in a hypothermic state. Of the three options concerning
performance variation--no change, decrement, and improve- Perhaps the earliest notion concerning human capability in
ment--the least likely would seem to be no effect for such a heat is that of physiological adequacy, which emanates from initial
manipulation. However, judging the likelihood of the latter two descriptive and experimental observations (Blagden, 1775a,
possibilities appears to be a matter of speculation. Evidence from 1775b; Ellis, 1758; Sutton, 1909). According to this proposition,
other performance areas, such as time perception, appears to measurements of body temperature, sweat rate, and other ther-
favor the notion of a slowing of response and therefore decrement moregulatory actions are taken as indices of physiological state.
(Baddeley, 1966). However, even within this latter realm, indi- Tolerance is set in terms of those conditions in which the thermal
vidual differences exert an important effect (e.g., C. R. Bell, 1966; equilibrium of the body is first disturbed. Physiological adequacy
Hancock, 1983). In the absence of more substantial indications, implies that performance remains unaffected while regulatory
no firm assertion concerning vigilance performance under this mechanisms are effective and homeostasis prevails.
hypothermic condition is possible at this time. Although early experimental studies appeared to support the
In summary, the consensus of data available concerning the adequacy notion (see Connell, 1948), subsequent research failed
thermal environment, body temperature, and capability on sus- to confirm the generality of the construct (e.g., Bartlett & Gronow,
tained attention tasks does not support the notions advanced 1953; Weiner & Hutchinson, 1945). However, these early studies
by Grether (1973) and by Davies and his colleagues (e.g.,Davies failed to distinguish between the type of cognitive or psychomotor
& Parasuraman, 1982; Davies & Tune, 1970). These authors performance task undertaken. With a careful differentiation be-
have indicated a general decrement due to cold and facilitation tween such tasks, specific increases in deep body temperature
in thermal conditions exceeding subjective comfort, with capa- can be used to describe the limits of efficient performance in
bility not failing until imminent physiological collapse. The mental, psychomotor, and dual-task performance categories
present review, in contrast, indicates variation in sustained at- (Hancock, 1982). Although the concept of physiological adequacy
tention efficiency with a measurable change in the deep body was one of the first accounts of performance under thermal stress,
temperature of the observer (see Colquhoun & Goldman, 1972). its early rejection appears to have been based on insufficient
As indicated in the data of Benor and Shvartz (1971), this dec- recognition of the variety of tasks undertaken by the stress par-
remental function is continuous and appears to follow a positive ticipants. Consequently, it has languished, disused for a consid-
exponential-type function. This function, typical of a positive erable time. This may have been an inappropriate treatment of
feedback system, is mirrored in the morphology of the breakdown a useful construct.
in body temperature itself under the driving force of some non- From the foregoing evaluation of vigilance in heat and cold,
compensable thermal surround, such as desert conditions. Sig- where significant performance decrement is apparently induced
nificant decrement in vigilance appears to coincide with the point by dynamic change in deep body temperature, the construct of
at which deep body temperature change may first be observed. physiological adequacy appears a reasonable account of the data
This review has been restricted to those conditions in which reported. However, despite this recognition, it is important to
the effect of an external thermal stress has been imposed on the note that it is largely a descriptive rather than explanatory account
individual. This does not include many studies (e.g., Bonnet & of the phenomenon at hand. Although the appeal to physiological
Alter, 1982; Moses, Lubin, Naitoh, & Johnson, 1978; Poock, mechanisms implies some empirical predictions, no distinct
Tuck, & Tinsley, 1969; Taub & Berger, 1976) that have examined causal mechanism is identified by this proposal.
the effect on performance of naturally occurring fluctuations in
body temperature due to circadian rhythm. However, in accord Behavioral A rousal
with the observations on static hyperthermia, it is clear that the
natural increase in deep body temperature associated with cir- Perhaps the most influential theoretical account that has been
cadian rhythm is also accompanied by an increase in capability proposed concerning the overall effects of stress on human per-
formance is behavioral arousal. With respect to the effect of en-
on a variety of tasks (Kleitman, 1963), including vigilance.
vironmental temperature on sustained attention, this position
has been presented most cogently by Poulton (Poulton, 1976,
Discussion
1977). To account for the variations in performance, as outlined
There have been a number of accounts of the impact of en- in the previous section, Poulton proposed five different effects
vironmental stressors on vigilance (e.g., Hancock, 1984a; Poul- founded on changes in arousal state. Initially, and after the ex-
ton, 1977) and, similarly, several proposals concerning the effect perimental findings of Mackworth (1950) and Pepler (1958), he
of thermal stress on a wide variety of cognitive and neuromus- postulated that mild heat, sufficient to induce discomfort, facil-
cular abilities (see P. A. Bell, 1981; Kobrick & Fine, 1983). These itates performance through an increase in arousal. Such an in-
range from simple descriptive approaches to those postulating crease is also associated with the initial entry into a heated en-
some form of underlying causal mechanism. With respect to vironment (Poulton & Kerslake, 1965). Continuation in a mildly
 SUSTAINED ATTENTION, TEMPERATURE 277

uncomfortable ambient temperature is accompanied by both cessing, feature extraction, response choice, and motor adjust-
the sensation of lassitude and a slow increase in body temperature, ment. It is suggested that stress arises when the effort mechanism
which reduces arousal and the ability to sustain attention. Poulton is overloaded over a period of time, or fails completely in adjusting
(1977) suggested that, should conditions be sufficient to cause a necessary energetic requirements. This model provides some in-
more abrupt rise in deep body temperature, then such arousing teresting insights into the mechanism of stress and is superior
circumstances promptly improve efficiency (e.g., Colquhoun & to the simple arousal constructs that precede it.
Goldman, 1972; Wilkinson et al., 1964). As with all capabilities,
the sustenance of attention fails rapidly as the individual ap- Attentional Resource Capacity
proaches thermally intolerable conditions (e.g., Benor & Shvartz,
1971). From this post hoc perspective, the arousal construct ap- In contrast to a linear-stage model of information processing,
pears to account for the diversity of experimental results reported. the human may be characterized as possessing a limited pool of
However, the very flexibility of the construct that allows for such attentional resources (Knowles, 1963). These resources may be
encapsulation and lends supposed explanatory power is, in ac- distributed among tasks, or elements of tasks, in order to accom-
tuality, a liability. Thus, the arousal construct is unable to dif- plish successful performance. Kahneman (1973) viewed these
ferentiate a priori which conditions will prove arousing and which resources as residing in a single global pool to be allocated as
depressing. Consequently, arousal remains a post hoe construc- appropriate to demand. As resources themselves have a finite
tion based on the various data sets reported. limit, so the human has limitations on both single- and multiple-
In the arousal account, when performance is described solely task abilities. Wickens (1980) has elaborated on this position, to
in terms of arousal change, the argument is a simple tautology include structure-specific resource pools, thus integrating a re-
with no practical or predictive utility. However, the arousal po- source model with the linear-stage approach. Specific research
sition gains greater credence if some mediational construct is pools are differentiated by modality of stimulus assimilation,
interpolated between arousal state and performance level. Such decision making, and mode of response. Considerable support
a construct was introduced by Easterbrook (1959), who postu- for this position is gained from observations of difficulty insen-
lated the restriction of cue utilization with increasing arousal sitivity, which are occasions on which two tasks may be performed
level. As arousal increases from a minimal level, non-task-related simultaneously, or time-shared, with no appreciable reduction
stimuli are rejected, and in consequence, performance improves. in the performance efficiency of either task. There have been
With increasing arousal, task-relevant cues are filtered and per- some recent objections to the methodology used in experiments
formance is diminished. This form of construct results in the that support the multiple-resource model (Damos & Lyall, 1984),
classic inverted U-shaped function matching arousal with effi- and a general critique of the notion of attentional resources by
ciency. One important element of this account, not commonly Navon (1984), who was instrumental in introducing the idea of
recognized, is that the correct rejection of non-task-related cues, attentional resources originally. However, the full elaboration of
which gives rise to the ascending arm of the inverted U, must be these arguments is beyond the scope of the present work. (For
a somewhat different process from the subsequent elimination more complete discussions, see Wickens, 1984, and Parasuraman
of task-relevant cues. This is because the former requires some & Davies, 1984). Rather, my central concern in this article is
form of selection, whereas the latter process is some general form with the impact of stress on attentional resources, and particularly
of degradation. Consequently, a further construct is essentially with how stress might influence the performance of a task re-
needed to account for the two different processes illustrated. In quiring the long-term sustenance of attention.
previous work (Hancock & Dirkin, 1983; Hockey, 1970), it has The foregoing arguments suggest that stress acts to drain at-
been suggested that attention may be able to account for such tentional resources. Further, from evidence on vigilance it appears
actions. This proposal is explored in more detail in the next that stress sufficient to perturb efficient homeostasis, that is, to
section specifically devoted to attention. cause an uncontrolled rise or fall in deep body temperature,
In sum, although arousal has been a useful construct in ac- results in significant performance decrement. This position is
counting for performance variation under stress, and has a sub- somewhat reminiscent of that adopted by Cohen (1978, 1980),
stantive physiological basis, its ubiquitous application has been who reasoned that the threatening nature of stressors demand
questioned (Broadbent, 1963). Also, its composition as a unitary attentional resource capacity and that prolonged demand results
entity has been recently criticized (Hancock, 1984b). For a more in shrinkage or cognitive fatigue. However, Cohen suggested that
detailed examination of the current status of the arousal notion, stress both competes for and drains resources, without specifying
see Hockey and Hamilton (1983). In conclusion, the simple the level of stress at which such a draining process becomes crit-
arousal construct is failing as a descriptive account, whereas its ical. In the present work, such a critical point is identified as 85
use as a causal explanation has always been somewhat suspect ~ ET for heat stress, where an uncontrolled rise in deep body
in that on most occasions the appeal has been to some vague temperature may be observed.
physiological mechanism that often remains undefined. For vigilance, the notion of physiological adequacy is partic-
An interesting exception to the above statement is the recent ularly descriptive of the summary of data reported. However, it
work of Sanders in which he attempts to relate stress and arousal does not describe the increases in performance capability, and,
to a linear-stage model of human information processing (Sand- further, it requires the mediational attentional construct to pro-
ers, 1983). Briefly, he outlines a model that relates three energetic vide some element of causality. The use of attention is also par-
supply systems, as envisaged by Pribram and McGuiness (1975), ticularly appealing for two separate reasons. First, it has been
to specific cognitive processing mechanisms. Arousal, effort, and indicated that those tasks that demand greater attention for their
activation are linked to processing stages of stimulus prepro- successful completion are most vulnerable to the effects of per-
278 P. A. HANCOCK

turbation to deep body temperature (Hancock, 1982). Vigilance, In the review sections concerning dynamic change to deep
as reviewed in this work, is particularly vulnerable, such that body temperature and its functional relation with performance
any perturbation to deep body temperature is enough to affect efficiency, it was noted that above the threshold at which deep
efficiency (see also Hancock & Pierce, 1984). A second reason body temperature was perturbed, performance was degraded,
is that if stress acts to drain attention, then those individuals and this function took the form of an increasing exponential. It
who through some strategy need to devote less attention to suc- was also recognized that this same function describes physiolog-
cessfully perform the task should prove less vulnerable to stress ical breakdown under the driving force of an extreme thermal
effects. Hancock (in press) has indicated that individuals who condition. In other work (Hancock & Chignell, 1985) we have
are skilled on the task they are performing, and are able presum- attempted to explore the implications of this potential iso-
ably to use relatively attention-free automatic-type processes, morphism and the possibility that is may represent some more
clearly suffer less adverse stress effects than their unskilled peers ubiquitous stress effect.
(see also Schneider & Shiffrin, 1977). In summary, stress drains attention. With less attention the
The linkage of stressor effects and the resultant diminution in perceiver is constrained to adopt a tactic of narrowing onto cues
attentional resource capacity suggests the use of psychological of perceived task salience in order to sustain successful perfor-
measures (cognitive performance) as predictive of imminent mance for as long as is feasible. Individuals who are skilled on
physiological breakdown under extreme thermal stress. Hancock a task have the opportunity to use relatively attention-free au-
and Chignell (1985) proposed that, outside of a comfort zone, tomatic processes that leave them less vulnerable to the impact
there are regions of adaptability to stressors, with the zone of of the stressor. Parenthetically, those individuals who have fa-
psychological adaptability being enclosed within the zone of miliarity with the stressor and are acclimatized to the stressful
physiological adaptability (Hancock & Chignell, 1985, Figure conditions also suffer less (Wilkinson, 1969b). However, this may
1). In this view, breakdown of performance on psychological represent amelioration of the physiological impact of the stress
measures would be directly reflected in change of physiological rather than the influence of attentional capacity per se. Finally,
state. Thus performance on attention-demanding cognitive tasks unsupportable conditions drain all attention resources and this
may be used as a form of early warning signal for the loss of leads to the termination of purposive behavioral activity, or un-
regulatory homeostasis or physiological adaptability. The con- consciousness. The continued impact of the stress eventually su-
nection between thermal stress and attentional resource capacity persedes physiological capability and this represents the termi-
is a natural one given that the level of thermal stress induces nation of life-sustaining functions, or death.
instability in core temperature. Cognitive processes are liable to After exploring the detrimental effects of temperature, it is
be susceptible to fluctuations in core temperature and associated important to redress the balance somewhat by indicating those
change in central nervous system temperature (Goodman et al., occasions on which hyperthermia improves performance. This
1984; Hancock, 1983). This rationale has been used to justify occurs when the individual is established in a static hyperthermic
deep auditory canal temperature as a reflection of mental work- state (Colquhoun & Goldman, 1972; Wilkinson et al., 1964).
load (Hancock, Meshkati, & Robertson, 1985). These conditions are similar to those observed during the se-
The notion of narrowing, as seen in the work of Easterbrook quential increase in temperature associated with the circadian
on cue utilization, is also important in understanding the effects rhythm. Under these circumstances, in which the relative rate
of stress on attention. Clearly, if stress acts to drain attentional of change of body temperature is small, performance also im-
resources, then less resources are available to perform any par- proves with ascending temperature level. These indications sug-
ticular task. Under such conditions, the optimal performance gest that both absolute level and rate of change of deep body
strategy is to use what resources are left available to process task- temperature mediates the impact on attention.
relevant cues. This appears to be the tactic adopted by the per- In conclusion, it should be noted that not only the level and
ceiver, but with the caveat that cues that are processed are those rate of change affect performance, but also the concomitant
of perceived highest salience, which may or may not coincide change in peripheral or skin temperature. Although core tem-
with those of greatest task relevance (Cornsweet, 1969; Easter- perature has been implicated in the speed of performance (Allnutt
brook, 1959). This concept has been explored most thoroughly & Allan, 1973; Hancock, 1983; Hoagiand, 1933), the rate of
in experiments by Hockey using noise as a stressor (e.g., Hockey, error has been associated with fluctuation in skin temperature
1970). Recently, we have demonstrated that the narrowing effect (Allan & Gibson, 1979). With respect to vigilance, the relation
is an attentional rather than a visual process. This was accom- between these two measures and performance largely remains
plished by showing superior response to task-relevant and per- to be explored. A fuller picture based on these more compre-
ceived salient cues in the visual periphery, compared with less hensive physiological measures awaits further experimentation.
relevant but more centrally placed comparison signals (Dirkin
& Hancock, 1984). There are a variety of empirical studies that Summary
demonstrate the narrowing phenomenon. Among those using
heat stress, those by BursiU (1958) and Provins and Bell (1970) In this article I have proposed that the key factor in determining
are most notable. Results from their work using dual-task per- change in vigilance under thermal stress is the thermophysiologi-
formance imply that stress drains rather than competes for at- cal state of the observer involved. When conditions are insufficient
tentional resources. However, a clear and unequivocal demon- to change deep body temperature, performance remains essen-
stration of this phemomenon under the impact of both extremes tially unaffected. However, thermal stress exposures that perturb
of heat and cold awaits a more thorough experimental investi- deep body temperature away from both normal and steady-state
gation. conditions impair vigilance, while static hyperthermic states are
 SUSTAINED ATTENTION, TEMPERATURE 279

beneficial with respect to sustained attention. The insights derived to hot and humid conditions. Quarterly Journal of Experimental Psy-
from sustained attention performance under thermal stress are chology, 10, 113-129.
potentially useful for the construction of an overall theory of Burton, A. C., & Edholm, O. G. (1955). Man in a cold environment.
London: Arnold.
stress and attention. Whereas initial efforts have confirmed such
Carlson, L. D. (1961). Human performance under different thermal loads
utility (see Hancock & Chignell, 1985), the elaboration into a
(Tech. Rep. No. 61-43). Brooks AFB, TX: U.S.A.E Aerospace Medical
full theoretical perspective demands long-term efforts and an in- Center, School of Aviation Medicine.
tegration of understanding of the action of a variety of stressors Carpenter, A. (1946). The effect of room temperature on the performance
on human behavior that has yet to be achieved. of the resistance box test; A performance test of intelligence (R.N.P.
Rep. No. 46/318). Unpublished report to Royal Naval Personnel
References Committee of the Medical Research Council, Cambridge, England.
Chiles, W, D., Iampietro, P. E, & Higgins, E. A. (1972). Combined effects
Allan, J. R., & Gibson, T. M. (1979). Separation of the effects of raised of altitude and high temperature on complex performance. Human
skin and core temperature on performance of a pursuit rotor task. Factors, 14, 161-172.
Aviation, Space and Environmental Medicine, 50, 678-682. Cohen, S. (1978). Environmental load and the allocation of attention. In
Allnutt, M. E, & Allan, J. R. (1973). The effects of core temperature A. Baum, J. E. Singer, & S. Valins (Eds.), Advances in environmental
elevation and thermal sensation on performance. Ergonomics, 16, 189- psychology (pp. 1-29). Hillsdale, NJ: Erlbaum.
196. Cohen, S. (1980). Aftereffects of stress on human performance and social
Angus, R. G., Pearce, D. G., Buguet, A. G. C, & Olsen, L. (1979). Vig- behavior: A review of research and theory. PsychologicalBulletin, 88,
ilance performance of men sleeping under Artic conditions. Aviation, 82-108.
Space and Environmental Medicine, 50, 692-696. Colquhoun, W. P. (1960). Temperament, inspection efficiency and time
Arees, E. A. (1963). The effect of environmental temperature and alerting of day. Ergonomics, 3, 377-378.
stimuli on prolonged search. Unpublished doctoral dissertation, Uni- Colquhoun, W. P. (1969). Effects of raised ambient temperature and
versity of Massachusetts, Amherst. event rate on vigilance performance. Aerospace Medicine, 40, 413-
Baddeley, A. D. (1966). Time estimation at reduced body temperature. 417.
American Journal of Psychology, 79, 475-479. Colquhoun, W. P., & Goldman, R. E (1972). Vigilance under induced
Bartlett, D. J., & Gronow, D. G. C. (1953). The effects of heat stress on hyperthermia. Ergonomics, 15, 621-632.
mental performance (Rep. No. 846). England: Flying Personnel Re- Connell, L. (1948). The effect of heat upon the performance of men in
search Committee. high speed aircraft:A critical review (Pep. NO. 151-1-17). New York:
Bazett, H. C. (1949). The regulation of body temperatures. In L. M. Psychological Corporation.
Newburgh (Ed.), Physiologyof heat regulation and the scienceof clothing Cornsweet, D. M. (1969). Use of cues in the visual periphery under con-
(pp. 109-192). Philadelphia: Saunders. ditions of arousal. Journal of Experimental Psychology, 80, 14-18.
Bell, C. R. (1964). Climate, body temperature and vigilance performance. Craig, A., Wilkinson, R. T., & Colquhoun, W. P. (1981). Diurnal variation
Proceedings of the Second International Congress on Ergonomics, 2, in vigilance efficiency. Ergonomics, 24, 641-651.
169-172. Damos, D. L., & Lyall, E. A. (1984). The effect of asymmetric transfer
Bell, C. R. (1966). Control of time estimation by a chemical clock. Nature, on dual-task assessment of voice technology. Proceedings of the Human
210, 1189-1190. Factors Society, 28, 21-25.
Bell, C. R., & Provins, K. A. (1962). Effects of high temperature envi- Davies, D. R., & Parasuraman, R. (1982). The psychology of vigilance.
ronmental conditions on human performance. Journal of Occupational London: Academic Press.
Medicine, 4, 202-221. Davies, D. R., & Tune, G. S. (1970). Human vigilanceperformance. Lon-
Bell, C. R., Provins, K. A., & Hiorns, R. W. (1964). Visual and auditory don: Staples Press.
vigilance during exposure to hot and humid conditions. Ergonomics, Dean, R. D., & McGlothen, C. L. (1965). Effects of combined heat and
7, 279-282. noise on human performance, physiology and subjective estimates of
Bell, P. A. (1981). Physiological, comfort, performance, and social effects comfort and performance. Institute of Environmental Sciences Annual
of heat stress. Journal of Social Issues, 37, 71-94. Technical Meeting Proceedings, 55-64.
Benor, D., & Shvartz, E. (197i). Effect of body cooling on vigilance in Dirkin, G. R., & Hancock, P. A. (1984). Attentional narrowing to the
hot environments. Aerospace Medicine, 42, 727-730. visual periphery under temporal and acoustic stress. Aviation, Space
Blagden, C. (1775a). Experiments and observations in an heated room. and Environmental Medicine, 55, 454. (Abstract)
Philosophical Transactions, 65, 111-123. Easterbrook, J. A. (1959). The effect of emotion on cue utilization and
Blagden, C. (1775b). Further experiments and observations in an heated the organization of behavior. PsychologicalReview, 56, 183-201.
room. Philosophical Transactions, 65, 484-494. Edholm, O. G. (1963). Heat acclimatization studied in the laboratory
Bonnet, M. H., & Alter, J. (1982). Effects of irregular versus regular sleep and the field. A multi-disciplinary approach. Ergonomics, 6, 304-305.
schedules on performance, mood and body temperature. Biological Ellis, H. (1758). An account of the heat of the weather in Georgia. Phil-
Psychology, 14, 287-296. osophical Transactions, 50, 755.
Broadbent, D. E. (1953). Noise, paced performance and vigilance tasks. Epstein, Y., Keren, G., Moisseiev, J., Gasko, O., & Yachin, S. (1980).
British Journal of Psychology, 44, 295-303. Psychomotor deterioration during exposure to heat. Aviation, Space
Broadbent, D. E. (1954). Some effects of noise on visual performance. and Environmental Medicine, 51, 607-610.
Quarterly Journal of Experimental Psychology, 6, 1-5. Fine, B. J., Cohen, A., & Crist, B. (1960). Effect of exposure to high
Broadbent, D. E. (1963). Differences and interactions between stresses. humidity at high and moderate ambient temperature on anagram so-
Quarterly Journal of Experimental Psychology, 15, 205-211. lution and auditory discrimination. PsychologicalReports, 7, 171-181.
Broadbent, D. E. (1971). Decision and stress. London: Academic Press. Fine, B. J., & Kobrick, J. L. (1978). Effects of altitude and heat on complex
Brown, I. D. (1967). Measurement of control skills, vigilance and per- cognitive tasks. Human Factors, 20, 115-122.
formance of a subsidiary task during 12 hours of car driving. Ergo- Fraser, D. C. (1957). Some effects of heat stress on performance of a
nomics, 10, 665-673. vigilance task under speed stress. National Coal Board Medical Research
Bursill, A. E. (1958). The restriction of peripheral vision during exposure Memorandum, 1, 50-63. (London)
280 P. A. HANCOCK

Goodman, D., Hancock, P. A., Runnings, D. W., & Brown, S. L. (1984). Iampietro, P. E, Melton, C. E., Higgins, E. A., Vaughan, J. A., Hoffman,
Temperature-induced changes in neuromuscular function: Central and S. M., Funkhouser, G. E., & Saldivar, J. T. (1972). High temperature
peripheral mechanisms. Perceptual and Motor Skills, 59, 647-656. and performance in a flight task simulator. Aerospace Medicine, 43,
Grether, W. E (1973). Human performance at elevated environmental 1215-1218.
temperatures. Aerospace Medicine, 44, 747-755. Jones, R. D. (1970). Effects of thermal stress on human performance: A
Hancock, P. A. (1981). Heat stress impairment of mental performance: review and critique of existing methodology (Tech. Rep. No. 11-70).
A revision of tolerance limits. Aviation, Space and Environmental Aberdeen, MD: Human Engineering Laboratories, Aberdeen Research
Medicine, 52, 177-180. and Development Center.
Hancock, P. A. (1982). Task categorization and the limits of human Kahneman, D. (1973). Attention and effort, Englewood Cliffs, N J: Pren-
performance in extreme heat. Aviation, Space and Environmental tice-Hall.
Medicine, 53, 778-784. Kissen, A. T., Reifler, C. B., & Thaler, V. H. (1964). Modification of
Hancock, P. A. (1983). An endogenous metric for the perception of short thermoregulatory responses to cold by hypnosis. Journal of Applied
temporal intervals. Unpublished doctoral dissertation, University of Physiology, 19, 1043-1050.
Illinois, Urbana-Champaign. Kleitman, N. (1963). Sleep and wakefulness. Chicago: University of Chi-
Hancock, P. A. (1984a). Effect of environmental temperature on display cago Press.
monitoring performance: An overview with practical implications. Knowles, W. B. (1963). Operator loading tasks. Human Factors, 5, 155-
American Industrial Hygiene Association Journal, 45, 122-126. 162.
Hancock, P. A. (1984b). Environmental stressors. In J. S. Warm (Ed.), Kobrick, J. L., & Fine, B. J. (1983). Climate and human performance.
Sustained attention in human performance. London: Wiley. In D. J. Oborne & M. M. Gruneberg (Eds.), The physical environment
Hancock, P. A. (in press). The effect of skill on performance under an at work (pp. 69-107). London: Wiley.
environmental stressor. Aviation, Space and Environmental Medicine. Lind, A. R. (1963). A physiological criterion for setting thermal envi-
Hancock, P. A., & Chignell, M. H. (1985). The principle of maximal ronmental limits for everyday work. Journal of Applied Physiology, 18,
adaptability in setting stress tolerance standards. In R. Eberts & C. E. 51-56.
Eberts (Eds.), Trends in ergonomics/human factors H (pp. 117-125). Loeb, M., & Jeantheau, G. (1958). The influence of noxious environ-
Amsterdam: North-Holland. mental stimuli on vigilance. Journal of Applied Psychology, 42, 47-49.
Hancock, P. A., & Dirkin, G. R. (1983). Stressor induced attentional Mackie, R. R. (Ed.). (1977). Vigilance: Theory, operational performance
narrowing: Implications for design and operation of person-machine and physiological correlates. New York: Plenum Press.
systems. Proceedings of the Human Factors Association of Canada, Mackie, R. R., & O'Hanlon, J. E (1977). A study of the combined effects
16, 19-21. of extended driving and heat stress on driver arousal and performance.
Hancock, P. A., Meshkati, N., & Robertson, M. M. (1985). Physiological In R. R. Mackie (Ed.), Vigilance: Theory, operationalperformance and
reflections of mental workload. Aviation, Space and Environmental physiological correlates (pp. 537-558). New York: Plenum Press.
Medicine, 56, 1110-1114. Mackworth, N. H. (1950). Researches on the measurement of human
Hancock, P. A., & Pierce, J. O. (1984). Toward an attentional theory of performance. Medical Research Council Special Report Series 268.
performance under stress: Evidence from studies of vigilance in heat London: Her Majesty's Stationary Office.
and cold. In A. Mital (Ed.), Trends in ergonomics/human factors I Mortagy, A. K. (1971). Effects of work~restschedules on monitoring per-
(pp. 1-7). Amsterdam: North-Holland. formance in the heat. Unpublished doctoral dissertation, Texas Tech
Hancock, P. A., & Pierce, J. O. (1985). Combined effects of heat and University, Lubbock.
noise on human performance: A review. American Industrial Hygiene Mortagy, A. K., & Ramsey, J. D. (1973). Monitoring performance as a
Association Journal 46, 555-566. function of work/rest ratio schedule and thermal stress. American In-
Harminc, M. (1976). Einige psychische und physiologische Funktionen dustrial Hygiene Association Journal, 34, 474-480.
in der Beiziehung zum Mikroklima [Some psychological and physio- Moses, J., Lubin, A., Naitoh, P., & Johnson, L. C. (1978). Circadian
logical functions due to the state of the microclimate]. Studia Psy- variation in performance, subjective sleepiness, sleep and oral tem-
chologica, 18, 41-50. perature during an altered sleep--wake schedule. Biological Psychology,
Head, H. (1926). Aphasia. Cambridge, England: Cambridge University 6, 301-308.
Press. Navon, D. (1984). Resources: A theoretic soupstone. PsychologicalReview,
Hoagiand, H. (1933). The physiological control ofjudgements of duration: 91, 214-224.
Evidence for a chemical clock. Journal of General Psychology, 9, 267- Parasuraman, R., & Davies, D. R. (1984). Varieties of attention. New
287. York: Academic Press.
Hockey, G. R. J. (1970). Signal probability and spatial location as possible Pepler, R. D. (1953). The effect of climatic factors on the performance of
bases for increased selectivity in noise. Quarterly Journal of Experi- skilled tasks by young European men living in the tropics. 4: A task of
mental Psychology, 22, 37-42. prolonged visual vigilance. (Rep. No. 156/53). Cambridge, England:
Hockey, G. R. J., & Hamilton, P. (1983). The cognitive patterning of Medical Research Council, Royal Naval Research Committee, Applied
stress states. In G. R. J. Hockey (Ed.), Stress and fatigue in human Psychology Unit.
performance (pp. 331-362). London: Wiley. Pcpler,R. D. (1958). Warmth and performance: An investigationin the
Hohnsbein, J., Piekarski, C., & Kampmann, B. (1983). Influence of high tropics. Ergonomics, 2, 63-88.
ambient temperature and humidity on visual sensitivity. Ergonomics, Poock, G. K., Tuck, G. A., & Tinsley, J. H. (1969). Physiological correlates
26, 905-911. of visual monitoring. Perceptual and Motor Skills, 29, 334.
Hohnsbein, J., Piekarski, C., Kampmann, B., & Noack, T. (1984). Effects Poulton, E. C. (1973). Unwanted range effects from using within-subject
of heat on visual acuity. Ergonomics, 27, 1239-1246. experimental designs. PsychologicalBulletin, 80, 113-121.
Houghten, E C., & Yagloglou, C. P. (1923). Determining lines of equal Poulton, E. C. (1976). Arousing environmental stresses can improve per-
comfort. Transactions of the American Society of Heating and Venti- formance: Whatever people say. Aviation, Space and Environmental
lating Engineers, 29, 163-176. Medicine, 47, 1193-1204.
Iampietro, P. E, Chiles, W. D., Higgins, E. A., & Gibbons, H. L. (1969). Poulton, E. C. (1977). Arousing stresses increase vigilance. In R. R. Mackie
Complex performance during exposures to high temperatures. Aero- (Ed.), Vigilance: Theory, operational performance and physiological
space Medicine, 40, 1331- 1335. correlates (pp. 423-459). New York: Plenum Press.
 SUSTAINED ATTENTION, TEMPERATURE 281

Poulton, E. C., & Edwards, R. S. (1974a). Interactions and range effects Teichner, W. H. (1966). Individual thermal and behavioral factors in cold-
in experiments on pairs of stresses: Mild heat and low-frequency noise. induced vasodilation. Psychophysiology, 2, 295-304.
Journal of Experimental Psychology, 102, 621-628. Viteles, M. S., & Smith, K. R. (1946). An experimental investigation of
Poulton, E. C., & Edwards, R. S. (1974b). Interactions, range effects, and the effect of change in atmospheric conditions and noise upon perfor-
comparisons between tasks in experiments measuring performance mance. Heating, Piping and Air Conditioning, 18, 107-112.
with pairs of stresses: Mild heat and 1 nag of L-hyoscine hydrobromide. Warm, J. S. (Ed.). (1984). Sustained attention in human performance.
Aerospace Medicine,45, 735-741. London: Wiley.
Poulton, E. C., Edwards, R. S., & Colquhoun, W. E ( 1974). The interaction Weiner, J. S., & Hutchinson, J. C. D. (1945). Hot humid environment:
of a loss of a night's sleep with mild heat: Task variables. Ergonomics, Its effect on the performance of a motor coordination test. British
17, 59-73. Journal of lndustrial Medicine, 2, 154-157.
Poulton, E. C., Hitchings, N. B., & Brooke, R. B. (1965). Effect of cold Wickens, C. D. (1980). The structure of attentional resources. In R. S.
and rain upon the vigilance of lookouts. Ergonomics, 8, 163-168. Nickerson & R. Pew (Eds.), Attention and performance VIII (pp. 239-
Poulton, E. C., & Kerslake, D. (1965). Initial stimulating effect of warmth 257). Hillsdale, NJ: Edbaum.
upon perceptual efficiency. AerospaceMedicine, 36, 29-32. Wickens, C. D. (1984). Engineeringpsychology and human performance.
Pribram, K. H., & McGuinness, D. (1975). Arousal, activation and effort Columbus, OH: Merrill.
in the control of attention. PsychologicalReview, 82, 116-149. Wilkinson, R. T. (1969a). Sleep deprivation: Performance tests for partial
Provins, K. A., & Bell, C. R. (1970). Effects of heat stress on the perfor- and selective sleep deprivation. In L. A. Abt & B. E Reiss (Eds.), Progress
mance of two tasks running concurrently. Journal of Experimental in clinical psychology (pp. 28-43). New York: Grune & Stratton.
Psychology, 85, 40-44. Wilkinson, R. T. (1969b). Some factors influencing the effect of environ-
Sanders, A. E (1983). Toward a model of stress and human performance. mental stressors upon performance. Psychological Bulletin, 72, 260-
Acta Psychologica, 53, 61-97. 272.
Schneider, W., & Shiffrin, R. M. (1977). Controlled and automatic human Wilkinson, R. T., Fox, R. H., Goldsmith, R., Hampton, I. F. G., & Lewis,
information processing: I. Detection, search, and attention. Psycho- H. E. (1964). Psychological and physiological responses to raised body
logical Review, 84, 1-66. temperature. Journal of Applied Physiology, 19, 287-291.
Sutton, H. (1909). The influence of high temperatures on the human Yaglou, C. P., & Minard, D. (1957). Control of heat casualties at military
body, especially with regard to heat stroke. Journal of Pathology and training centers. Archives of lndustrial Heatlh, 16, 302-316.
Bacteriology, 13, 62-73.
Taub, J. M., & Berger, R. J. (1976). Effects of acute sleep pattern alteration Received July 1, 1985
depend upon sleep duration. Physiological Psychology, 4, 412-420. Revision received August 19, 1985 9

Hancock, P.A. (1986). Sustained attention under thermal stress. Psychological Bulletin, 99 (2), 263-281.