J. Sleep Res. (2006) 15, 23–29
Chronic partial sleep loss increases the facilitatory role
of a masked prime in a word recognition task
CLAIRE E. SWANN, GREG W. YELLAND, JENNIFER R. REDMAN
and S H A N T H A M . W . R A J A R A T N A M
School of Psychology, Psychiatry and Psychological Medicine, Monash University, Clayton, Victoria, Australia
Accepted in revised form 8 September 2005; received 28 June 2005
SUMMARY
Neurobehavioural performance deficits associated with sleep loss have been extensively
studied, in particular, the effects on psychomotor performance. However, there is no
consensus as to which, if any, cognitive functions are impaired by sleep loss. To
examine how sleep loss might affect cognition, the automatic processes supporting word
recognition were examined using the masked priming paradigm in participants who had
been exposed to two consecutive days of sleep restriction. Twelve healthy volunteers
(mean age 24.5 years) were recruited. Nocturnal sleep duration was restricted to 60% of
each participant’s habitual sleep duration for two consecutive nights by delaying
scheduled time of sleep onset and advancing time of awakening. In controlled
laboratory conditions, participants completed the Psychomotor Vigilance Task and
Karolinska Sleepiness Scale and a masked priming word recognition task. As expected,
significant increases in subjective sleepiness and impaired psychomotor performance
were observed after sleep loss. In contrast, response times and error rate on the masked
priming task were not significantly affected. However, the magnitude of the masked
priming effect, which can be taken as an index of automaticity of lexical processing,
increased following sleep loss. These findings suggest that while no evidence of
impairment to lexical access was observed after sleep loss, an increase in automatic
processing may occur as a consequence of compensatory mechanisms.
k e y w o r d s automatic processing, cognition, lexical processing, masked priming,
performance, sleep loss
INTRODUCTION
Sleep loss is increasingly recognised as a major cause of
transportation and industrial accidents (Dement, 1999; Horne
and Reyner, 1999; Smith and Kushida, 2000). Consequently,
the impact of sleep loss on human neurobehavioural function
is of great interest. This study examined the effect of chronic
partial sleep loss on two critical aspects of the automatic
processes supporting cognitive function: those underlying
visual perception and word recognition.
The most common cause of chronic partial sleep loss is a
reduction in the amount of time allocated for daily sleep
Correspondence: Shantha M.W. Rajaratnam, PhD, School of Psychology, Psychiatry and Psychological Medicine, Building 17, Monash
University, Vic. 3800, Australia. Tel.: +613 9905 3934; fax: +613 9905
3948; e-mail: [email protected]
2006 European Sleep Research Society
(Dement, 1999; Harrison and Horne, 1996; Liu et al., 2000;
Van Dongen et al., 2003b). Daily sleep requirement varies
among individuals and estimates of the average are generally
in the range of 7 to 8.5 h (Bonnet and Arand, 1995; Buysse and
Ganguli, 2002; Van Dongen et al., 2003b). However, estimates
of the average amount actually obtained vary from 6 to 7.5 h
(Bonnet and Arand, 1995; Harrison and Horne, 1996; Kripke
et al., 2002), supporting the contention that a substantial
proportion of the population in industrialised countries are
chronically sleep deprived (Dement, 1999; Liu et al., 2000).
A consistently observed effect of sleep loss is a deficit in
psychomotor performance, which manifests as delayed reaction
time and lapses in attention (Binks et al., 1999). Such deficits are
thought to contribute to sleepiness-related accidents. Although
there is no consensus as to which specific cognitive functions
are impaired by sleep loss, several studies have observed
23
24
C. E. Swann et al.
performance decrements on cognitive tasks following sleep loss
(e.g. Harrison and Horne, 1997, 1998; Harrison et al., 2000).
Discovery of the cognitive processes that may be affected by
sleep loss is complicated by several factors. First, tasks of less
than 10 min duration do not appear to show performance
impairment (Veasey et al., 2002), which has prompted the
suggestion that it is not the ability to perform the task per se that
is affected, but rather, the ability to maintain concentration.
Secondly, performance on many tasks involves multiple components of cognition. Consequently, when performance deficits
are observed, it is difficult to determine which functions are
affected. For example, the decline in performance on language
tasks following sleep loss observed by Harrison and colleagues
(Harrison and Horne, 1997, 1998; Harrison et al., 2000) may be
due to the slowing of lexical processing, loss of automaticity, a
deficit in the higher order mental functions or impairment in a
combination of these components.
To examine how sleep loss might affect cognitive processes,
it is necessary to isolate the components of a task that assess
cognition. Given that performance on many cognitive tasks is
dependent upon a range of processes, it is appropriate to start
with basic low-level processes such as the automatic processes
that support cognition. Thus, the automatic processes supporting word recognition were examined in the present study
using the masked priming paradigm (Forster, 1990; Forster,
1998; Forster and Davis, 1984).
The masked priming task measures the processes of lexical
access, the automatic processes that support word recognition
(Forster, 1990; Forster, 1998; Forster and Davis, 1984). The
masked priming paradigm has been used extensively as a means
of exploring the processes of lexical access, without contamination from extra-lexical cognitive processes (Badecker and
Allen, 2002; Bodner and Masson, 2001; Feldman and Soltano,
1999; Grainger et al., 1991; Longtin et al., 2003). As in any word
priming paradigm, the target word to which a response is to be
made is presented following a prime stimulus to which the target
is related or unrelated. Upon the presentation of target items,
which are a mix of words and orthographically legal non-words,
participants are required to make a timed word–non-word
(lexical) decision. The critical feature of the masked priming task
is that the prime is unavailable to conscious processing (Forster,
1990). This is achieved by presenting the prime briefly (50–
60 ms) in lowercase between a forward mask of a row of hash
marks for 500 ms and the target letter string that acts as a
backward mask. Thus, any subsequent facilitation of target
recognition can be attributed solely to the unconscious lexical
processing which begins automatically upon exposure to the
prime (Forster and Davis, 1984). The priming effect is the
difference between the response times to targets preceded by
their related primes relative to targets preceded by their
unrelated primes. The magnitude of the priming effect is an
index of the automaticity of lexical processing. When the lexical
processor is unimpaired, exposure to the prime activates the
lexical entry and facilitates subsequent recognition of the target.
However, when automatic processing slows down, such as with
normal ageing, the prime is only partially processed prior to the
target presentation. Therefore, the prime is of less benefit and
priming effects, and by implication the degree of automaticity of
lexical processing, would diminish. Hence, reduced masked
priming effects are potentially diagnostic of degradation of
lexical automaticity, which may occur following loss of sleep. If
such impairment does result from partial sleep loss it is likely to
be subtle and detection would therefore require a particularly
sensitive measure. The masked priming paradigm provides us
with such a measure.
The present study used two consecutive nights of restricted
sleep to mimic chronic partial sleep loss. Following sleep loss,
participants were exposed to a battery of tests in the laboratory.
The Karolinksa Sleepiness Scale (KSS) (Akerstedt and Gillberg,
1990) and the Psychomotor Vigilance Task (PVT) (Dinges et al.,
1987, 1997) have been shown to be sensitive to two nights of
restricted sleep (Dinges et al., 1997; Phipps-Nelson et al., 2003)
and were used to establish whether the sleep restriction protocol
was sufficient for cognitive impairment to manifest.
METHOD
Participants
Participants were 12 healthy adults (five females, seven males)
aged between 18 and 35 years (mean age 24.5 years), recruited
via poster advertisements displayed at Clayton and Caulfield
Campuses of Monash University and in public areas around
Melbourne, Australia. A further 13 people volunteered to take
part; however, 10 were rejected after inspection of completed
sleep diaries and screening questionnaires revealed that they
did not meet the inclusion criteria, two failed to adhere to the
sleep restriction protocol and were excluded, and one was
excluded for having an error rate exceeding 20% on the
masked priming task stimuli during the sleep loss condition
testing session. All participants were native speakers of
English. Participants were screened to exclude extreme circadian types using the Morningness–Eveningness Questionnaire
(Horne and Ostberg, 1976). Sleep diaries, general screening
and medical screening (including drug history) questionnaires
were used to identify and exclude any individuals with chronic
or recent acute medical conditions (other than mild asthma), a
history of drug or alcohol dependence, taking regular medication (other than the oral contraceptives or that used to treat
asthma), who on an average slept less than seven or more than
9 h nightly, smoked more than five cigarettes per day or
consumed more than 10 standard alcoholic beverages weekly
or 250 mg of caffeine (equivalent to approximately two 5 oz
servings of brewed coffee) each day.
Prior to the commencement of the study, participants were
given an explanatory statement detailing the aims and procedures of the research and written informed consent was
obtained. The research was approved by the Standing
Committee on Ethics in Research Involving Humans at
Monash University and was partly funded by a Monash Small
Grant. All participants who completed the procedure were
paid $50 (Australian).
2006 European Sleep Research Society, J. Sleep Res., 15, 23–29
Sleep loss affects automatic processing
Materials and design
The study had a repeated measures design. Participants
completed a battery of tests, once when fully rested (control
condition) and once following two consecutive nights of
restricted sleep (sleep loss condition). For each individual,
the sessions were held between 1 and 3 weeks apart. The
sequence of testing was fully counterbalanced: six participants
completed the control testing first and six completed the sleep
loss condition first.
The battery of tests included: three administrations of the
KSS to assess subjective sleepiness (Akerstedt and Gillberg,
1990), two PVT sessions (Dinges et al., 1987, 1997), one
session of a novel perceptual judgement task (included here as
part of another study)1, and the masked priming task to
measure lexical processing. The order of presentation of tests
within the battery remained the same.
Each session of the PVT was of 10 min duration and stimuli
were presented at random intervals of 2.0–10.0 s
(mean ¼ 6.0 s). There were 110–121 trials per 10-min testing
session (Dinges et al., 1997).
The masked priming task was administered using a Hewlett
Packard mini PC (Hewlett Packard, Melbourne, Australia)
attached to a Phillips 15 in flat screen running at 100 MHz with
a refresh rate of 13 ms and a screen resolution of 450 · 600 dpi.
The response device was an ÔEasy CatÕ external glide-point
mouse that measures response times to an accuracy of
±0.024 ms. DMDX software (Forster & Forster, University
of Arizona, Tucson, AZ, USA) was used to run the masked
priming task.
For the masked priming task, target items included 20
monomorphemic words, 10 of which were of high-frequency
occurrence and early age of acquisition (HE) and 10 of low
frequency and late age of acquisition (LL). Each target had
primes of two types: Identity (ID) and All Letters Different
(ALD). The ID prime was the same word as the target (e.g. blue–
BLUE), and the ALD prime differed at each position in the letter
string between prime and target (e.g. sand-BLUE). The ALD
prime serves as a baseline against which the magnitude of the
masked ID priming effect can be determined. The ALD primes
were matched to their targets (and by default, the ID prime) on
number of letters and syllables, word frequency and age of
acquisition. Twenty non-word targets, each with an ID and
ALD prime, were also included in the item set as distracters for
the lexical decision task. The non-word targets were selected to
match the 20 word targets in terms of number of letters and
syllables, and all were pronounceable and orthographically legal
(e.g. zilp-ZILP, serd-ZILP). In addition, eight words (four
words and four non-words) were included as practice items.
There were two counterbalanced versions of the masked
priming task (versions A and B), each consisting of 10 of the 20
1
The task forms part of another, the outcomes of which are subject to a
confidentiality agreement. It is mentioned here because it impacts on
the sequence of tests in the battery and the overall demand on a
participant.
2006 European Sleep Research Society, J. Sleep Res., 15, 23–29
25
word targets and 10 of the 20 non-word targets. In version A,
half the targets were preceded by their ID primes and half by
their ALD primes. In version B, this pairing was reversed, such
that the targets that were preceded by their ID primes in
version A were preceded by their ALD primes in version B and
vice versa. This counterbalancing procedure was used so that
each target appeared once with its ID prime and once with its
ALD prime.
The within subjects design of this study required each
participant to complete both versions so that appropriate
comparisons could be made across the sleep conditions (sleep
restriction versus control). One problem with repeating
versions is that the repeated target may attract an episodic
priming component. To reduce the potential impact of the
episodic priming component, both versions were completed
twice, in the sequence A–B–A–B, as pilot testing indicated that
by the second presentation, any facilitation of target recognition due to episodic memory was equal for both versions. The
A–B–A–B sequence was presented without interruption with
participants unaware that only responses from the latter part
of the task were being recorded. Data were collected only for
the second A–B sequence of each version; hence, the episodic
memory trace remaining from the initial exposure to the target
items was equal across all responses recorded. The authors
acknowledge that one consequence of this methodology is that
response times and priming effects may be somewhat smaller
when compared to targets that are not repeated.
Procedure
Baseline assessment
Participants used a sleep diary to record their sleep patterns
for a week. This required them to record their sleep/wake
schedule and subjectively assess the quality of each sleep
episode. The sleep diaries were used firstly to identify and
exclude any potential participants with irregular sleeping
habits and secondly, to establish the temporal placement and
average duration of their habitual sleep (baseline assessment).
Within 2 weeks of the baseline assessment, participants were
exposed to either the sleep restriction condition or control
condition.
Sleep restriction condition
For five consecutive nights immediately preceding the sleep
restriction phase, participants were instructed to adhere to their
usual pattern of sleep and record their sleep/wake schedule in a
sleep diary. For two nights of the sleep restriction phase,
participants were required to go to sleep and awaken at times
specified by the experimenter. Sleep time was calculated as 20%
later than their habitual sleep time and wakeup time was 20%
earlier than habitual wake time (as established from the baseline
assessment). The purpose of this method was to restrict sleep to
60% of habitual nocturnal sleep duration. To monitor compliance, participants were required to immediately telephone the
26
C. E. Swann et al.
experimenter prior to going to bed and upon waking. Those who
failed to do so were excluded from the study.
Target
Control condition
For the week prior to the control testing session, participants
were instructed to adhere to their usual sleep/wake schedule
and record it in a sleep diary.
Prime
500 ms
########
Performance testing
56 ms
Participants were instructed to abstain from consuming
products containing alcohol, caffeine or stimulants of any
kind for the 48 h preceding performance testing sessions.
Participants arrived at the Sleep Laboratory at Monash
University Caulfield Campus 1 h after their habitual wakeup time. Arrival time was the same for each participant in both
testing conditions. Upon arrival, they were given a standardised breakfast, consisting of apple juice, cereal and milk,
containing no added sugar. Following this was a 1.5 h
acclimatisation period. During the acclimatisation and testing
periods participants were seated in a comfortable armchair in
silence and isolation (other than necessary contact with the
experimenter). Reading was permitted during the acclimatisation period and during breaks between tasks, but electronic
devices such as mobile phones, televisions and laptops were
not. Room temperature was 21 C (mean ¼ 20.8 C,
SD ¼ 1.7 C) and ambient light measured at the horizontal
angle of gaze was 2.2 lux (±0.4 lux). The laboratory session
was conducted according to the schedule given in Table 1.
The performance battery commenced with the KSS, which
was followed by the PVT (see Table 1). Participants were given
the KSS and a pen or pencil and instructed to circle the
number corresponding to the label that best described how
sleepy they were feeling at that time. Prior to their first PVT
session, participants were given a demonstration of the PVT by
the experimenter, followed by a practice period before commencing the test session. On subsequent trials, the demonstration and practice trials were omitted.
Table 1 Schedule of laboratory session
Time into laboratory
session (mins)
0–30
30–120
120–130
130–140
140–155
155–165
165–175
175–185
185–195
Activity
Breakfast
Acclimatisation period
Start testing session
KSS#1
PVT#1
Break
Novel perceptual judgement task
Break
KSS#2
Masked priming task
Break
PVT#2
KSS#3
End testing session
500 ms
Figure 1. Presentation sequence for the masked priming task.
Each masked priming trial began with the presentation of
a forward mask consisting of a row of eight hash marks
(########) for 500 ms, followed by the lowercase prime for
56 ms, then the uppercase target for 500 ms (see Fig. 1).
Upon presentation of the target stimulus, participants were
required to make a lexical decision, classifying the target as
either a word or a non-word. Affirmative responses were
made by pressing a button, and if the target was a nonword, no response was required. Using this method, known
as the Ôgo/no-goÕ response, no data regarding the classification of non-words was collected. However, as non-words do
not have lexical representations, these responses do not
represent lexical processing and are irrelevant for the
purpose of this study. Furthermore, the go/no-go method
is associated with more accurate and less variable responses,
relative to tasks using the two-choice response method
(Perea et al., 2003). The task was of approximately 7 min
duration and was preceded by eight practice items, which
took approximately 30 s to complete. Practice items were
presented before test trials on each occasion.
Data analysis
Mean sleepiness level was calculated from the three administrations of the KSS and mean values were also computed for the
two PVT trials for each experimental condition. The measures
determined from the PVT were mean reaction time, total errors
and lapses (>500 ms), as these captured the main features of
psychomotor vigilance performance (i.e. reaction time, accuracy and attention). Hence only these measures are reported.
To minimise the influence of outlying response times, two
standard techniques were applied to the response time data of
each participant for the masked priming task. Only responses
within the ranges of 200–2000 ms were considered legitimate.
Responses outside this timeframe did not represent the cognitive processes under consideration, but rather, inattention or preemptive button pressing, and were therefore treated as errors
(Winer, 1971). Any further outliers were addressed by bringing
responses that were more than two standard deviations from
2006 European Sleep Research Society, J. Sleep Res., 15, 23–29
27
Sleep loss affects automatic processing
Sleep–wake patterns
Control
PVT measure
Mean
Sleep restriction
Standard
error
Mean
Mean reaction 251.15 9.70
time (ms)
Total errors
0.88 0.22
Lapses
0.71 0.17
(RT > 500s)
Standard
t-value P-value
error
285.29 12.10
0.79
2.42
600
Mean response time (ms)
RESULTS
Table 2 Means, standard errors and t-test results for Psychomotor
Vigilance Task (PVT) measures
0.23
0.64
)3.55
.005
0.41
)2.92
.689
.014
P = 0.016
580
60
50
560
ALD prime
40
540
30
520
ID prime
500
20
480
10
Priming effect (ms)
a participant’s mean response time back to the two standard
deviation mark. This technique, known as Windsorisation
(Winer, 1971), is standard in psycholinguistics research.
As response times for errors on the masked priming task are
discarded, an error rate of greater than 20% does not yield a
sufficient number of response times for means to be reliably
established for all conditions. One participant was excluded
from the study on this basis, and replaced by another
individual, to maintain a sample size of 12 participants.
For the masked priming task, response time and error rate
data were analysed by anova. Priming effects were calculated
by subtracting the response value (either response time or error
rate) for targets with their ID primes from the corresponding
value for the ALD primed targets. Only response times for
correct responses were included in the calculation of priming
effects.
460
0
During baseline assessment, average time of sleep onset ranged
across participants from 22:53 to 1:34 hours (M ¼ 12:20
hours, SD ¼ 52 min). Average time of awakening ranged
from 7:14 to 10:16 hours (M ¼ 8:23 hours, SD ¼ 54 min).
The mean amount of sleep obtained during this phase was
8.1 h (SD ¼ 36 min).
The mean amount of sleep permitted during the sleep
restriction phase was 4.8 h (SD ¼ 22 min). Designated time of
sleep onset ranged from 12:40 to 3:20 hours (M ¼ 1:57 hours,
SD ¼ 50 min) and wakeup time ranged from 5:40 to
8:30 hours (M ¼ 6:47 hours, SD ¼ 52 min).
During the week prior to the control testing session, when
participants were instructed to adhere as closely as possible to
their usual sleep/wake schedule, the mean amount of sleep
obtained was 7.90 h (SD ¼ 31 min). Average time of sleep
onset ranged from 23:05 to 1:45 hours (M ¼ 12:19 hours,
SD ¼ 45 min) and average time of awakening ranged from
7:21 to 10:30 hours (M ¼ 8:19 hours, SD ¼ 57 min).
KSS and PVT
Mean KSS ratings for the control and sleep loss conditions
were 4.1 (SD ¼ 1.0) and 7.1 (SD ¼ 1.0) respectively. A t-test
revealed that this difference was significant (t(11) ¼ )8.39,
P < 0.001). Mean reaction time, total errors and lapses
(>500 ms) are shown in Table 2. Mean reaction time and
lapses were significantly greater after sleep loss, but the
number of errors were not significantly affected.
Masked priming
Mean response times and standard errors were calculated for
the masked priming task. Priming effects were calculated by
subtracting response times for words preceded by their ID
primes from those of words preceded by their ALD primes (see
2006 European Sleep Research Society, J. Sleep Res., 15, 23–29
Control
Sleep
restriction
Control
Sleep
restriction
Figure 2. Mean response times (±standard error) for Identity (ID)
and All Letters Different (ALD) prime types for control and sleep
restriction conditions (left panel). The right panel shows mean priming
effects (±standard error), calculated by subtracting response times for
words preceded by their ID primes from those of words preceded by
their ALD primes for each subject separately for the control and sleep
restriction conditions.
Fig. 2). Two-way repeated measures anova (treatment condition, prime type) revealed a significant main effect of prime type
(i.e. priming effect) (F(1.11) ¼ 100.84, P < 0.001), showing
that targets were responded to faster when preceded by their ID
primes (M ¼ 508 ms) than ALD primes (M ¼ 544 ms). There
was no main effect of condition, with responses to targets in the
control condition (524 ms) being only 5 ms faster than during
the sleep loss condition (M ¼ 529 ms).
Importantly, there was an increase in the magnitude of the
masked priming effect following sleep loss, which is supported
by a significant treatment condition by prime type interaction
(F(1.11) ¼ 8.10, P ¼ 0.016). Posthoc analyses revealed that
there was no significant difference between control and sleep
restriction conditions on response times to targets with ALD
primes (t(11) ¼ 0.573, P ¼ 0.578) or with ID primes
(t(11) ¼ 0.219, P ¼ 0.830). The priming effect (ID versus
ALD) in the control condition was significant (t(11) ¼ 5.235,
P < 0.001) as was the priming effect in the sleep restriction
condition (t(11) ¼ 8.791, P < 0.001). The priming effect for
the sleep restriction condition (M ¼ 47 ms) was significantly
greater than the priming effect for the control condition
(M ¼ 26.0 ms; t(11) ¼ 2.847, P ¼ 0.016).
Error rate data showed that word recognition was not
impaired by restriction of sleep (see Table 3). The error rates
for the control and sleep loss conditions were identical and
negligible in magnitude. anova revealed no significant main
28
C. E. Swann et al.
Table 3 Group mean response times, standard errors and word category for control and sleep restriction conditions
Control
Sleep restriction
Prime
type
Mean error
rate (%)
Standard
error
Mean error
rate (%)
Standard
error
ID
ALD
Priming
0.42
0.83
0.41
0.4
0.8
0.42
0.83
0.41
0.4
0.8
effects or interactions in the error data. An absence of priming
effects for errors in the presence of significant priming effects
for response times is standard in masked priming research, and
indicates that there is no word-finding difficulty (loss of lexical
stock).
DISCUSSION
The aim of this study was to investigate the effects of chronic
partial sleep loss on automatic cognitive function. No evidence
to suggest impairment to either primary visual processing or
lexical access was found. Rather, word recognition performance remained stable following sleep loss and automatic
processing apparently became more important to word recognition following sleep loss.
The increased subjective sleepiness following loss of sleep is
consistent with the findings of other partial sleep loss studies
(Carskadon and Dement, 1981; Dinges et al., 1997; Van
Dongen et al., 2003a), as is the deterioration of psychomotor
performance (Dinges et al., 1997; Phipps-Nelson et al., 2003;
Van Dongen et al., 2003a). The significant increase in subjective sleepiness and impairment to psychomotor performance
following restriction of sleep indicates that participants were,
in fact, sleep deprived, and that compliance with the experimental protocol was adequate for the detrimental effects of
sleep loss to manifest.
Considerable research into the impact of sleep loss on
neurobehavioural function suggests that there is a positive
relationship between the level of sustained attention required
to perform a task, and the degree to which task performance is
impaired by sleep loss (Jennings et al., 2003). Accordingly,
performance on tasks that rely primarily on unconscious
cognitive resources would be less affected than performance on
tasks with a high requirement for sustained attention. The
contrast between performance deficits observed for the PVT,
which specifically measures sustained attention, and the lack of
impairment in performance on the task that targets automatic
processing supports this notion.
Results from the masked priming task produced no evidence
to suggest that automatic lexical processing was impaired by
sleep loss. Consistent with previous research (e.g. Bodner and
Masson, 2001; Forster and Davis, 1984; Pratarelli et al., 1994),
identity priming was found to occur in both conditions, but the
facilitatory role of the prime increased after loss of sleep, as
evidenced by the significant increase in masked priming effect
for response times in the sleep restriction condition. In
addition to eliminating the possibility of degradation of lexical
automaticity, these data indicate that unconscious processing
of the prime was occurring in the sleep loss condition, and the
increase in priming effect magnitude suggests that automatic
processing becomes more important to the process of word
recognition following partial sleep loss.
It is widely accepted that compensatory effort exerted by
research participants can temporarily ameliorate observable
performance decrement during a period of sleep loss, particularly on short tasks (Harrison and Horne, 2000; Jennings et al.,
2003). However, only since functional imaging technology
became available, has the possibility of neural compensation,
that occurs automatically, been explored. Drummond and
colleagues (Drummond et al., 2000, 2001) observed that levels of
cerebral activation in the prefrontal cortex and the parietal lobe
during verbal learning tasks were higher following 35 h of
continual wakefulness, relative to a rested state. The altered
brain responses were associated with mild impairment to recall
memory task but performance on a recognition task was
unaffected. Drummond et al. (2000) speculated that the outcome reflects a compensatory mechanism, either direct or
indirect, for the cerebral consequences of increased homeostatic
drive for sleep. Given that the underlying mechanisms of sleep
loss related performance decline are so poorly understood,
results should be interpreted with caution. However, it is
possible that the plasticity of brain function, tentatively
proposed by Drummond and colleagues, may explain the
increase in masked identity priming effects, in the absence of a
performance deficit observed in the present study. It might be
that the average of 4.8 h sleep allowed in the sleep loss condition
is sufficient to trigger recruitment of additional cognitive
resources, thus enhancing the role of automatic lexical access
and preventing an observable performance deficit. With more
severe sleep loss either a magnification of priming effects
combined with a global performance decrement may occur or
priming effects may diminish due to the impairment of automatic lexical processing (i.e. the compensatory mechanism may be
able to overcome mild impairment until it becomes impaired
itself). This issue could be explored in future research.
The restriction of sleep to a proportion of each participant’s usual amount was intended to reflect on the concept
of sleep debt, defined by Rogers et al. (2003) as Ôthe
cumulative hours of sleep lost with respect to a subjectspecific daily need for sleepÕ (p. 5). Although measures were
taken to guard against including sleep-deprived individuals
in the study cohort, such as excluding volunteers who
obtained on average less than 7 h sleep per night, it is
possible that some were still experiencing a deficit in their
actual sleep need even though their habitual sleep duration
is greater than 7 hours. Furthermore, estimation of participantsÕ habitual amount of sleep was based on self-report,
the accuracy of which cannot be verified. However, the
protocol appeared to achieve a degree of sleep loss that was
sufficient for the psychomotor impairment to manifest.
In conclusion, the sleep restriction protocol, which involved
reducing nocturnal sleep duration by delaying sleep onset and
2006 European Sleep Research Society, J. Sleep Res., 15, 23–29
Sleep loss affects automatic processing
advancing time of awakening by 20% of each participant’s usual
amount, resulted in significant increases in subjective sleepiness
and impairment to psychomotor performance. Response time
and accuracy on the test of automatic cognitive processing were
unaffected by sleep loss. The critical outcome was the significant
increase in magnitude of masked priming effect in the absence of
any global decrement to word recognition performance following restriction of sleep. As a tentative explanation for these
results, we suggest compensatory mechanisms which maintain
cognitive processing after sleep restriction.
ACKNOWLEDGEMENTS
The authors wish to express their gratitude to Ms Jo PhippsNelson for her assistance in this study. Parts of this research
were presented at the 19th Annual Meeting of the American
Professional Sleep Societies, Denver, USA (June, 2005). This
research was supported by a Monash Small Grant to JRR,
SWR and GWY.
REFERENCES
Akerstedt, T. and Gillberg, M. Subjective and objective sleepiness in
the active individual. Int. J. Neurosci., 1990, 52: 29–37.
Badecker, W. and Allen, M. Morphological parsing and the perception
of lexical identity: a masked priming study of stem homographs.
J. Mem. Lang., 2002, 47: 125–144.
Binks, P. G., Waters, W. F. and Hurry, M. Short-term total sleep
deprivation does not selectively impair higher cortical functioning.
Sleep, 1999, 22: 328–334.
Bodner, G. E. and Masson, M. E. J. Prime validity affects masked
repetition priming: Evidence for an episodic resource account of
priming. J. Mem. Lang., 2001, 45: 616–647.
Bonnet, M. H. and Arand, D. L. We are chronically sleep deprived.
Sleep, 1995, 18: 908–911.
Buysse, D. J. and Ganguli, M. Can sleep be bad for you? Can insomnia
be good? Arch. Gen. Psychiatry, 2002, 59: 137–138.
Carskadon, M. A. and Dement, W. C. Cumulative effects of sleep
restriction on daytime sleepiness. Psychophysiology, 1981, 18: 107–
113.
Dement, W. C. The Promise of Sleep. Delacorte Press, New York, NY,
1999.
Dinges, D. F., Orne, M. T., Whitehouse, W.G. and Orne, E.C.
Temporal placement of a nap for alertness: contributions of
circadian phase and prior wakefulness. Sleep, 1987, 10: 313–329.
Dinges, D. F., Pack, F., Williams, K., Gillen, K. A., Powell, J. W., Ott,
G. E., Aptowicz, C. and Pack, A. I. Cumulative sleepiness, mood
disturbance, and psychomotor vigilance performance decrements
during a week of sleep restricted to 4–5 hours per night. Sleep, 1997,
20: 267–277.
Drummond, S. P. A., Brown, G. G., Gillin, J. C., Stricker, J. L.,
Wong, E. C. and Buxton, R. B. Altered brain response to verbal
learning following sleep deprivation. Nature, 2000, 403: 655–657.
Drummond, S. P. A., Gillin, J. C. and Brown, G. G. Increased cerebral
response during a divided attention task following sleep deprivation.
J. Sleep Res., 2001, 10: 85–92.
Feldman, L. B. and Soltano, E. G. Morphological priming: the role of
prime duration, semantic transperancy, and affix position. Brain
Lang., 1999, 68: 33–39.
Forster, K. I. Lexical processing. In: D. N. Osheron and H. Lasnik
(Eds) Language: An invitation to cognitive science. Volume 1. The
MIT Press, Cambridge, MA, 1990: 96–128.
2006 European Sleep Research Society, J. Sleep Res., 15, 23–29
29
Forster, K. I. The pros and cons of masked priming. J. Psycholinguist.
Res., 1998, 27: 203–233.
Forster, K. I. and Davis, C. Repetition priming and frequency
attenuation in lexical access. J. Exp. Psychol. Learn. Mem. Cogn.,
1984, 10: 680–698.
Grainger, J., Cole, P. and Segui, J. Masked morphological priming in
visual word recognition. J. Mem. Lang., 1991, 30: 371–384.
Harrison, Y. and Horne, J. A. Long-term extension to sleep – are we
really chronically sleep deprived. Psychophysiology, 1996, 33: 22–30.
Harrison, Y. and Horne, J. A. Sleep deprivation affects speech. Sleep,
1997, 20: 871–877.
Harrison, Y. and Horne, J. A. Sleep loss impairs short and novel
language tasks having a prefrontal focus. J. Sleep Res., 1998, 7: 95–
100.
Harrison, Y. and Horne, J. A. The impact of sleep deprivation on
decision making: a review. J. Exp. Psychol. Appl., 2000, 6: 236–249.
Harrison, Y., Horne, J. A. and Rothwell, A. Prefrontal neuropsychological effects of sleep deprivation in young adults – a model for
healthy aging? Sleep, 2000, 23: 1067–1073.
Horne, J. A. and Ostberg, O. A self-assessment questionnaire to
determine morningness-eveningness in human circadian rhythms.
Int. J. Chronobiol., 1976, 4: 97–110.
Horne, J. A. and Reyner, L. Vehicle accidents related to sleep: a
review. Occup. Environ. Med., 1999, 56: 289–294.
Jennings, J. R., Monk, T. H. and van der Molen, M. W. Sleep
deprivation influences some but not all processes of supervisory
attention. Psychol. Sci., 2003, 14: 473–479.
Kripke, D. F., Garfinkel, L., Wingard, D. L., Klauber, M. R. and
Marler, M. R. Mortality associated with sleep duration and
insomnia. Arch. Gen. Psychiatry, 2002, 59: 131–138.
Liu, X., Uchiyama, M., Kim, K., Okawa, M., Shibui, K., Kudo, Y.,
Doi, Y., Minowa, M. and Ogihara, R. Sleep loss and daytime
sleepiness in the general adult population of Japan. Psychiatry Res.,
2000, 93: 1–11.
Longtin, C.-M., Segui, J. and Halle, P. A. Morphological priming
without morphological relationship. Lang. Cogn. Process., 2003, 18:
313–334.
Perea, M., Rosa, E. and Gomez, C. Influence of neighbourhood size
and exposure duration on visual-word recognition: evidence with the
yes/no and the go/no-go lexical decision tasks. Percept. Psychophys.,
2003, 65: 273–286.
Phipps-Nelson, J., Redman, J.R., Dijk, D.J. and Rajaratnam, S.M.
Daytime exposure to bright light, as compared to dim light,
decreases sleepiness and improves psychomotor vigilance performance. Sleep, 2003, 26: 695–700.
Pratarelli, M. E., Perry, K. E. and Galloway, A. M. Automatic lexical
access in children: new evidence from masked identity-priming.
J. Exp. Child Psychol., 1994, 58: 346–358.
Rogers, N. L., Dorrian, J. and Dinges, D. F. Sleep, waking and
neurobehavioural performance. Front. Biosci., 2003, 8: 1056–1067.
Smith, R. and Kushida, C. Risk of fatal occupational injury by time of
day. Sleep, 2000, 23: A110–A111.
Van Dongen, H. P. A., Maislin, G., Mullington, J. M. and Dinges, D.
F. The cumulative cost of additional wakefulness: dose–response
effects on neurobehavioural functions and sleep physiology from
chronic sleep restriction and total sleep deprivation. Sleep, 2003a,
26: 117–126.
Van Dongen, H. P. A., Rogers, N. L. and Dinges, D. F. Sleep debt:
theoretical and empirical issues. Sleep Biol. Rhythms, 2003b, 1: 813–
819.
Veasey, S., Rosen, R., Barzansky, B., Rosen, I. and Owens, J. Sleep
loss and fatigue in residency training: a reappraisal. JAMA, 2002,
288: 1116–1124.
Winer, B. J. Statistical Principles in Experimental Design, 2nd edn.
McGraw-Hill, New York, NY, 1971.