Sleep Deprivation - Fatigue - Perception - Heuristics - Effort-Mental - Decision Making
- Motivation
INTRODUCTION
Elevated fatigue in critical decision making is associated with costly real-world
outcomes. In a sample of 204 primary care physicians, the prescribing of antibiotics
meeting the “sometimes indicated” and “never indicated” criteria increased progressively
over three-hour work sessions[1]. A study of 4,000 health care workers found an 8% decline in the frequency of hand
washing over the course of the work shift[2]. In a study of parole verdicts, judges made progressively fewer favorable verdicts
(which are more demanding than unfavorable verdicts) from 65% favorable at the start
of the session to 0% favorable at the end[3]. Fatigued individuals respond with racial bias by producing shorter reaction times
in a shooting simulation of Black armed suspects, depictions which support stereotypes[4].
Fatigue, the subjective feeling of tiredness and exhaustion, accompanies experimentally
controlled sleep deprivation[5]
,
[6], naturally occurring poor quality sleep[7], and insufficient sleep[8], and seems to be associated with behavioral reductions in effort. Though loss of
sleep typically leads to greater fatigue, there is considerable variability in subjective
feelings of fatigue following sleep loss[9]. Total sleep deprivation in the experimental setting refers to twenty-four hours
or more of extended wakefulness not induced by naturalistic causes, such as illness.
Such experimenter-controlled total sleep deprivation results in objective increases
in simpler behaviors, including fewer attempts at problems[10], the choice of low-effort/low-monetary reward tasks over high-effort/high-reward
tasks[11], the selection of easier math problems[12], and increased number of unsolved math sequences when informed of task difficulty[13].
Sensations that accompany sleep deprivation, such as fatigue, may reflect physical
or cognitive limitations and signal reductions in capacity that directly affect behavior.
Reductions in task engagement may result from a heightened perception of task difficulty
caused by the absence of sleep and the fatigue that ensues. For example, fatigued
individuals evaluated a hill as steeper and physical distances as greater[14], though participants did not actually measure hill steepness or physical distance.
In a sample of ice skaters, reports of less sleep and more frequent awakenings were
associated with perception of greater difficulty of certain skating maneuvers, and
poorer sleep quality predicted choice of easier maneuvers[15]. In studies of college students, math tasks perceived to be less difficult were
chosen following sleep deprivation compared to those chosen following full sleep[12]. As perception precedes behavior[16], increased perceptions of task difficulty could help explain alterations in behavior
following sleep deprivation.
Heuristics, as defined by Simon[17], are cognitive strategies in decision making which are used to obtain adequate solutions
while minimizing systematic processing. Heuristics involve simpler strategies to arrive
at solutions, such as examining fewer cues or integrating less information[18]. Thus, decisions made through heuristic processing rely on less effortful strategies
than those made through systematic processing[19]. In the one total sleep deprivation study that assessed heuristics, an increased
use of the local-representativeness heuristic was observed; participants were unable
to suppress pre-potent biases following sleep deprivation[20]. An understanding of the specific heuristics used following sleep deprivation in
the laboratory can help to identify similar applied situations in which heuristics
may be used.
The present study examined the impact of sleep deprivation on the use of three types
of heuristics: the “what-is-beautiful-is-good” heuristic[21] in which a stimulus perceived to be physically attractive (“what-is-beautiful”)
is judged to be inherently more valuable (“good”) than an unattractive stimulus, the
“greedy algorithm” heuristic[22], in which minimal time is used to reach a solution rather than a systematic evaluation
of information, and the “speed-accuracy trade-off”[23], in which effort is conserved to end task engagement.
Use of these heuristics can negatively impact productivity and quality of life. Using
the what-is-beautiful-is-good heuristic can lead to discrimination in the workplace;
physically attractive individuals receive higher starting salaries[24] and better performance evaluations[25] than less attractive individuals. Use of the greedy algorithm heuristic may result
in the worst possible solution[26], and by choosing the more expedient option with the speed-accuracy trade-off, tasks
completed more quickly may contain more errors[27].
The aim of this study was to determine whether total sleep deprivation produces greater
perception of task difficulty and use of heuristics compared to naturally-experienced
sleep. We hypothesized that in comparison to those with naturally-experienced sleep
at home, sleep-deprived participants would (1) rate certain tasks as more difficult,
(2) perceive greater task difficulty of task-specific elements, and (3) use the what-is-beautiful-is-good,
greedy algorithm, and speed-accuracy trade-off heuristics rather than complex mental
processes. We further hypothesized that (4) increased levels of reported fatigue would
predict greater perception of difficulty and heuristic use, regardless of experimental
condition.
METHODS
Participants
Thirty-nine undergraduate students (17 female) from the ages of 18 to 29 (M=19.18, SD=2.67) were assigned through block-random assignment to Naturally-Experienced Sleep
(NES;n=19) or Total Sleep Deprivation (TSD; n=20). The sample was 49% Asian (19), 18% Latino (7), 15% African American/Black (6),
15% Caucasian/White (6), and 3% West Indian (1). Participants were enrolled in Introductory
Psychology, Social Psychology, or Introductory Management courses and received credit
toward their course research requirement.
Those in good physical and mental health as assessed by the Patient Health Questionnaire-9
(PHQ-9)[28] were included in the study. Exclusion criteria were assessed during screening and
included reports of sleep problems (e.g. insomnia or hypersomnia), circadian rhythm
disorder as measured through questionnaire, use of sleeping pills, sedatives, or stimulant
medications, pregnancy or possible pregnancy, dependence on nicotine or caffeine,
and use of recreational drugs (e.g., marijuana, psychedelics, heroin). The protocol
was approved by the Human Research Protection Program, Baruch College, City University
of New York. Participants provided written informed consent prior to the medical screening
and the morning assessments.
MATERIALS
Actigraph Watches
NES participants wore an actigraph watch (Micro Motionlogger Sleep Watch; www.ambulatory-monitoring.com)
to assess sleep the night before the Final Assessments.
Profile of Mood States (POMS-Short Form)
The Profile of Mood States-Short Form (POMS-SF)29 measures participants’ self-reported mood and includes a five-item fatigue subscale
used in the current study.
Morningness-Eveningness Questionnaire (MEQ)
This questionnaire[30] was used to assess chronotype, or sleep timing preference. The total scores range
from 16 (definitely evening type) to 86 (definitely morning type).
Sleep Questionnaire
Prior to the perception of task difficulty and heuristics assessments, all potential
participants were asked to report their typical nighttime sleep quality, including
total sleep time, amount of sleep needed to feel refreshed, number of nightly awakenings,
and insomnia symptoms (Insomnia Severity Index [ISI])[31].
Perception of Difficulty Assessment: Article Task and Puzzle Task
The purpose of this assessment was to determine whether total sleep deprivation affected
the perception of task difficulty in comparison to those who had naturally-experienced
sleep. Participants were asked for their estimates of task qualities. Such estimates
could reflect a perception of greater task challenge. Participants were not asked
to complete these tasks.
In the Article Task, a paper copy of an article is examined for 30 seconds. With the
article in front of them, participants estimate: 1) the time they would need to read
the article, 2) the number of pages the article contains (actual number = 35), 3)
the number of words presented on the first page (actual number = 277), 4) how difficult
it would be to read the article, and 5) how difficult it would be to write a summary
of the article.
In the Puzzle Task, an unassembled jigsaw puzzle is examined for 30 seconds. With
the puzzle pieces in front of them, participants estimate: 1) the time it would take
them to complete the puzzle, 2) the number of pieces they believe the puzzle contains
(actual number = 100), and 3) how difficult it would be to complete the puzzle.
Heuristics Assessment: Quality Judgment Task (What-Is-Beautiful-Is-Good)
In this task, participants examine two separate images of refrigerators (previously
assessed for attractiveness), each paired with a different consumer review. A photo
of an attractive refrigerator is paired with an unfavorable review and a photo of
an unattractive refrigerator is paired with a favorable review. Participants evaluate
the refrigerator quality and report their likelihood of purchasing each refrigerator.
Heuristics Assessment: Following Instructions Task (Greedy Algorithm)
This task has an instruction section, a reading passage, and four subsequent questions.
If participants read the instructions, they know to answer only Question 4. Those
who answer all four questions will have skipped the instructions.
Heuristics Assessment: Math Difficulty-Time Choice (Speed-Accuracy Trade-Off)
In this assessment, participants expect to work on arithmetic problems for 20 minutes.
After seven minutes, they are asked whether they would like to continue solving similar
problems or work on more difficult problems for less time. Participants who choose
the more difficult problems receive a follow-up question asking if they chose this
option because they wanted to finish the task more quickly or because they desired
a challenge.
Procedure
Recruitment
Prior to Fall 2015 (Recruitment 1), screened participants selected one of two assessment
dates which were randomly determined to be NES or TSD. Students were informed of the
condition after they selected a date. In Fall 2015 (Recruitment 2), immediately after
screening, eligible participants were randomly assigned to NES or TSD and began their
study involvement that night. During the screening, potential participants were informed
of the study itinerary and continued with procedures knowing they could be randomly
assigned to either sleeping at home or staying awake overnight in the lab space.
Actigraph Pickup and Overnight Monitoring Session
NES participants began wearing the actigraph watch the day prior to the Final Assessments
and slept at home that night. TSD participants arrived the night before the Final
Assessments and stayed overnight at the college, monitored by research assistants.
The period of extended wakefulness in TSD ranged from 22.5 to 29.5 hours, depending
on individual wake time on the day prior to the Final Assessments. Neither NES nor
TSD participants were permitted to consume caffeine or nicotine after 14:00 the day
of the overnight session until the end of the Final Assessments, and TSD were not
permitted to use electronics after 24:00 until the beginning of the assessments. The
light emitted from these devices mimics sunlight, which decreases melatonin levels.
Since melatonin is involved in promoting sleep, exposure to the frequency of blue
light emitted by electronic devices can increase wakefulness[32].
Breakfast and Final Assessments
At 08:30 the following morning, the groups ate the same breakfast items; at 09:00,
they began the Final Assessments (both separately).
Design
The design of this study was between-groups with eligible participants block-randomly
assigned to either the NES or TSD group. Statistical analyses were conducted comparing
answers on the Final Assessments between the two groups.
Statistical Analyses
Outcome variables with non-normally distributed data were transformed using non-linear
transformations to meet the assumptions of parametric tests. Specifically, between-group
effects were examined through independent groups t tests for continuous outcomes and chi-square analysis for binary outcomes. Outcome
variables which did not meet the standard of normality, nor could be transformed to
become normal, were analyzed with non-parametric Mann-Whitney U tests. POMS-SF fatigue served as a predictor of all outcome variables through linear
regression for continuous outcomes and logistic regression for binary outcomes.
RESULTS
Preliminary Analyses
Pre-Study Sleep Characteristics
None of the assessed pre-study sleep variables differed significantly between groups.
According to independent-groups t tests, the average nightly sleep duration for the experimental group (M=7.19 hrs, SD=0.84) was not significantly different from the control group (M=6.99 hrs,SD=1.15), p=.546,t(37) = 0.61, suggesting both groups had similar habitual sleep duration prior to the
study. Other assessed sleep characteristics for each group may be viewed in [Table 1].
Table 1
Self-reported typical sleep quality indicators in Naturally-Experienced Sleep and
Total Sleep Deprivation groups.
|
|
Naturally-Experienced Sleep
|
Total Sleep Deprivation
|
Total
|
|
|
M (SD)
|
M (SD)
|
M (SD)
|
|
Nightly total sleep time (mins)
|
419.40 (50.40)
|
431.40 (69.00)
|
425.40 (60.00)
|
|
Amount of sleep needed to feel refreshed (mins)
|
439.80 (106.80)
|
398.40 (153.60)
|
418.20 (133.20)
|
|
Insomnia Severity Index (ISI)
|
5.12
a
(3.15)
|
3.91
a
(2.99)
|
4.50
a
(3.09)
|
|
Number of nightly awakenings
|
0.50 (0.53)
|
0.50 (0.71)
|
0.50 (0.61)
|
Note. All data were self-reported on pre-study screening questionnaires. No significant
differences were found between groups according to independent-groups t tests (not
shown).
aCorresponds to no clinical insomnia.
Objective Measure of Sleep Quality in Naturally-Experienced Sleep Group
The total sleep time of NES ranged from 202 to 461 minutes (3.37 to 7.68 hours). The
average total sleep time in NES was 5.91 hours, constituting a mild sleep deficit33,34. Participants slept for a significantly shorter period of time (M=354.74,SD=72.84) than they reported needing to feel rested (M=462.63, SD=84.45),t(18)=4.03, p=.001,d=0.93, according to paired-samples t tests. See [Table 2] for the NES actigraph data.
Table 2
Objective sleep quality indicators from actigraph data in Naturally-Experienced Sleep
group.
|
|
M (SD)
|
Min
|
Max
|
|
Total sleep time (mins)
|
354.74 (72.84)
|
202.00
|
461.00
|
|
Number of awakenings per hour sleep
|
1.08 (0.79)
|
0
|
3.11
|
|
Duration of awakenings per hour sleep (mins)
|
3.62 (3.04)
|
0
|
9.56
|
|
Mean length of awakenings (entire sleep)
|
3.49 (2.03)
|
1.00
|
7.00
|
|
Sleep efficiency (%)
|
89.65% (10.56%)
|
63.75%
|
100%
|
|
Sleep onset latency (mins)
|
13.74 (7.40)
|
4.00
|
30.00
|
|
Sleep deficit
a
|
107.89 (116.44)
|
-52.00
|
349.00
|
|
|
Median
|
Earliest
|
Latest
|
|
Time fell asleep
|
00:17
|
21:59
|
02:35
|
|
Time woke up
|
06:58
|
05:21
|
09:38
b
|
aCalculated as self-reported amount of sleep needed in order to feel rested minus
the total sleep time recorded by the actigraph device. Negative value indicates sleep
surplus; positive value indicates sleep deficit.
bOne participant overslept past the designated arrival time (08:30). This participant
completed the Final Assessments at 11:00 instead of 09:00 as originally intended.
Effects of Chronotype on Outcome Variables
The MEQ scores for NES ranged from 28 (definite evening type) to 63 (moderate morning
type); the scores for TSD ranged from 38 (moderate evening type) to 62 (moderate morning
type). According to independent-groups t tests, there were no differences between NES (M=51.32, SD=7.70) and TSD (M=49.00, SD=7.06) in chronotype as measured by the MEQ, t(37)=0.98, p=.334,d=0.32. On average, neither group could be classified as morning or evening type. Based
on the assessment of morningness-eveningness, no participant had a circadian rhythm
dysfunction. MEQ score did not significantly predict fatigue, perception of task difficulty,
or use of heuristics (all p>.05) according to regression analyses.
Effects of Sleep Deprivation on Reported Fatigue
TSD reported significantly greater fatigue (M=18.60,SD=5.49) than did NES (M=9.37,SD=4.46), t(37)=5.92,p<.001, d=1.95, on the POMS-SF according to linear regression analyses.
Effects of Sleep Deprivation on Perception of Task Difficulty and Use of Heuristics
Perception of Difficulty Assessment: Article Task and Puzzle Task
According to independent-groups t tests, TSD participants estimated significantly more time would be needed to read
the article (M=129.25, SD=106.16) and rated the article as significantly more difficult (M=4.10,SD=0.97) than did NES participants (estimated time: M=77.11, SD=44.64,t[37]=2.31, p=.026,d=0.76; difficulty rating: M=3.16,SD=1.21, t[37]=2.69,p=.011, d=0.88). No differences in groups were found in estimated number of pages (p=.737), number of words on the first page (p=.741), or difficulty rating for writing an article summary (p=.213).
No significant differences were found when NES and TSD groups were compared on estimated
time to complete the puzzle (p=.511), estimated number of puzzle pieces (p=.142), or difficulty rating for the puzzle (p=.531) according to independent-groups t tests. See [Figure 1]
,
[Figure 2]
, and
[Table 3].
Figure 1 Differences between Naturally-Experienced Sleep (NES; light gray bars) and Total
Sleep Deprivation (TSD; dark gray bars) on estimated time in minutes to read the article
(left) and complete the puzzle (right). Error bars ± standard error of the mean. *p < .05, two-tailed. n.s. = not significant.
Figure 2 Differences between Naturally-Experienced Sleep (NES; light gray bars) and Total
Sleep Deprivation (TSD; dark gray bars) on subjective task difficulty ratings (1 =
Very easy, 5 = Very difficult) for reading the article (left) and completing the puzzle
(right). Error bars ± standard error of the mean. *p < .05, two-tailed. n.s. = not significant.
Table 3
Differences between Naturally-Experienced Sleep and Total Sleep Deprivation in Perception
of Difficulty Assessment.
|
|
Naturally-Experienced Sleep
|
Total Sleep Deprivation
|
t (df)
|
p
|
d
|
|
|
M (SD)
|
M (SD)
|
|
Article Task
|
|
|
|
|
|
|
Time (minutes)
§
|
77.11 (44.64)
|
129.25 (106.16)
|
2.31 (37)
|
.026*
|
0.76
|
|
Number of pages
a
§
|
38.11 (21.26)
|
46.70 (43.72)
|
0.34 (37)
|
.737
|
0.11
|
|
Number of words on first page
b
§
|
208.68 (86.38)
|
309.20 (374.89)
|
0.33 (27.12)
c
|
.741
|
0.13
|
|
Difficulty rating (reading)
|
3.16 (1.21)
|
4.10 (0.97)
|
2.69 (37)
|
.011*
|
0.88
|
|
Difficulty rating (summary)
|
3.53 (1.07)
|
3.95 (0.97)
|
1.27 (36)
|
.213
|
0.42
|
|
Puzzle Task
|
|
|
|
|
|
|
Time (minutes)
§
|
36.58 (18.19)
|
45.26 (34.05)
|
0.66 (36)
|
.511
|
0.22
|
|
Number of pieces
d
§
|
70.21 (26.81)
|
133.16 (213.76)
|
1.50 (36)
|
.142
|
0.50
|
|
Difficulty rating
§
|
2.68 (1.11)
|
2.78 (0.88)
|
0.63 (35)
|
.531
|
0.21
|
Note. All values are estimated by the participants.
aActual number of pages = 35.
bActual number of words = 277.
cLevene's test for homogeneity of variance was significant (p < .05); t test statistic
corrected through degrees of freedom was used to determine significance.
dActual number of pieces = 100.
§Variable has been transformed to attain normality.
*
p < .05, two-tailed.
Heuristics Assessment: Quality Judgment Task (What-Is-Beautiful-Is-Good)
No significant differences between NES and TSD participants were found in reported
quality rating (p=.163) or purchase likelihood (p=.223) for the attractive refrigerator according to Mann-Whitney U tests.
According to independent-groups t tests, TSD participants rated the unattractive refrigerator with the favorable review
as significantly lower in quality (M=3.37,SD=0.83) than NES participants (M=4.21,SD=0.79), t(22.45)=-3.75,p=.001, d=-1.58. TSD participants also reported being significantly less likely to purchase
this refrigerator (M=3.42, SD=1.02) than NES participants (M=4.00, SD=0.75),t(25.46)=-2.23, p=.035,d=-0.88. See [Figure 3].
Figure 3 Differences between Naturally-Experienced Sleep (NES; light gray bars) and Total
Sleep Deprivation (TSD; dark gray bars) in subjective quality ratings (1 = Low quality,
5 = High quality) for image of attractive refrigerator with unfavorable review (left)
and unattractive refrigerator with favorable review (right). Error bars ± standard
error of the mean. *p < .05, two-tailed. n.s. = not significant.
Heuristics Assessment: Following Instructions Task (Greedy Algorithm)
Fifty-eight percent of NES skipped instructions compared to 90% of TSD. According
to chi-square analysis, TSD participants answered all four questions significantly
more than NES participants, X2 (1)=4.89,p=.027. Based on the odds ratio, the odds of participants skipping the instructions
were 6.18 times higher for the TSD group than for the NES group.
Heuristics Assessment: Math Difficulty-Time Choice (Speed-Accuracy Trade-Off)
There were no significant differences between NES and TSD groups in choice to complete
more difficult math problems for less time (p=1.00) according to chi-square analysis. Of those who completed the challenging problems
for a shorter period of time (n=20, 10 from each condition), a greater proportion of TSD participants chose this
option to finish the task quickly (70% of TSD) compared to NES participants (30% of
NES), X2 (1)=3.20,p=.074, in findings trending towards significance. TSD participants who chose the difficult
math problems had 5.44 times higher odds of reporting the desire to finish the task
quickly (rather than wanting a challenge) as compared with NES participants who chose
the more difficult math problems. See [Figure 4] and [Table 4].
Figure 4 Percentage of Naturally-Experienced Sleep (NES; light gray bars) and Total Sleep
Deprivation (TSD; dark gray bars) using the greedy algorithm heuristic (skipped instructions)
in Following Instructions Task (left); of those choosing the more difficult math problems
(n = 20; 10 each from NES and TSD), percentage of NES and TSD using the speed-accuracy
trade-off (chose more difficult problems to conserve time) in Math Difficulty-Time
Choice (right). ┼ p < .10, two-tailed. *p < .05, two-tailed.
Table 4
Differences between Naturally-Experienced Sleep and Total Sleep Deprivation in Heuristics
Assessment.
|
|
Naturally-Experienced Sleep
|
Total Sleep
Deprivation
|
U
|
p
|
r
|
|
|
Median
|
Median
|
|
Quality Judgment Task Attractive/ unfavorable fridge
|
|
|
|
|
|
|
Quality rating
a
|
1.00
|
1.00
|
132.50
|
.163
|
.28
|
|
Purchase likelihood
a
|
1.00
|
1.00
|
138.50
|
.223
|
.26
|
|
|
M (SD)
|
M (SD)
|
t(df)
|
p
|
d
|
|
Unattractive/favorable fridge
|
|
|
|
|
|
|
Quality rating
§
|
4.21 (0.79)
|
3.37 (0.83)
|
3.75 (22.45)
b
|
.001**
|
-1.58
|
|
Purchase likelihood
§
|
4.00 (0.75)
|
3.42 (1.02)
|
2.23 (25.46)
b
|
.035*
|
-0.88
|
|
|
%
|
%
|
X2
|
p
|
|
% Skipping instructions (greedy algorithm)
|
58%
|
90%
|
4.89
|
.027*
|
|
% Choosing easier math (speed-accuracy)
|
47%
|
47%
|
0
|
1.000
|
|
% Choosing difficult math to complete quickly
c
|
30%
|
70%
|
3.20
|
.074
†
|
aAnalyzed with nonparametric Mann-Whitney U test.
bLevene's test for homogeneity of variance was significant (p < .05); t test statistic
corrected through degrees of freedom was used to determine significance.
cPercentage of group reporting reasoning for choosing difficult path problems (n =
20) in order to complete task more quickly.
§Variable has been transformed to attain normality.
†
p < .10, two-tailed.
*
p < .05, two-tailed.
**
p < .01, two-tailed.
Reported Fatigue (POMS-SF) as Predictor of Perception of Task Difficulty and Use of
Heuristics Across Conditions
Perception of Difficulty Assessment: Article Task and Puzzle Task
According to linear regression analyses, greater fatigue significantly predicted a
higher difficulty rating for reading the article, ß=0.52,t(37)=3.66, p=.001,R2
=.27, and greater estimated time to read the article, ß=0.55, t(37)=3.96,p<.001, R2
=.30. In findings approaching significance, greater fatigue predicted a higher difficulty
rating for writing a summary, ß=0.30,t(36)=1.91, p=.064,R2
=.09. There was no association between fatigue and estimated number of pages (p=.127) or estimated number of words on first page (p=.188).
Greater fatigue significantly predicted greater estimated time to complete the puzzle,
ß=0.36, t(36)=2.32,p=.026, R2
=.13, according to linear regression analysis. In findings approaching significance,
greater fatigue predicted estimation of a greater number of puzzle pieces, ß=0.30,
t(36)=1.87, p=.070,R2
=.09. Fatigue levels were not significantly associated with the difficulty rating
for the puzzle (p=.164).
Heuristics Assessment: Quality Judgment Task (What-Is-Beautiful-Is-Good)
According to linear regression analyses, greater fatigue significantly predicted a
higher quality rating for the attractive refrigerator with the unfavorable review,
ß=0.45, t(36)=3.06,p=.004, R=.21, and predicted a greater likelihood of purchasing this refrigerator in findings
approaching significance, ß=0.30, t(36)=1.92,p=.063, R2
=.09.
Greater fatigue significantly predicted a rating of lower quality for the unattractive
refrigerator with the favorable review, ß=-0.34,t(36)=-2.16, p=.038,R2
=.12, according to linear regression analysis; however, there was no association
between reported fatigue and purchase likelihood for this refrigerator, p=.205.
Heuristics Assessment: Following Instructions Task (Greedy Algorithm)
No significant association between reported fatigue and skipping the instructions
for the Following Instructions Task was found (p=.204) according to logistic regression analysis.
Heuristics Assessment: Math Difficulty-Time Choice (Speed-Accuracy Trade-Off)
A logistic regression showed greater fatigue predicted choice of easier math problems
in findings approaching significance, X2 (1)=3.29,p=.070, RN
2 = .11. Among those choosing the more difficult math problems, no relationship between
reported fatigue and reasoning for choosing these problems was found (p=.122).
DISCUSSION
Overview and Implications of Findings
Sleep Deprivation and Perception of Task Difficulty
Sleep-deprived participants expected that reading the article would be more difficult
and that more time would be necessary to complete the task. The assessment of the
specific, countable aspects of the task, including number of pages and number of words,
was unaffected by sleep deprivation. These findings suggest that while perception
of objective task elements is unchanged after total sleep deprivation, sleep loss
results in expected performance limitations and a decrease in estimation of one›s
own ability.
If individuals perceive tasks as more difficult following sleep deprivation or insufficient
sleep, they are reflecting the impaired status of the system, and they may be less
motivated to expend effort because the task appears to be-and perhaps is-less feasible.
The increased perception of difficulty for the Article Task may result in the reduction
in motivation to complete such a task. One study, for example, found that self-reported
motivation decreased progressively throughout completion of a task perceived as difficult[35]. It was hypothesized that this reduced engagement was due, in part, to progressively
decreasing expectations of successful task completion.
According to Bandura›s Self-Efficacy Theory[36], an individual’s level of perceived self-efficacy determines the extent of effort
they will expend in the face of adversity and the length of time they will persist
at obstacles. Expectations of greater self-efficacy will lead to more intense efforts
when facing difficult tasks. Thus, individuals who experience total sleep deprivation
and perceive tasks to be more difficult may also perceive themselves to be less able
to perform these tasks, creating a self-fulfilling prophecy. These findings have implications
in settings such as school and the workplace, where individuals’ assessment of their
own ability to perform tasks is likely to impact initial engagement and performance
outcomes.
Sleep Deprivation and the Use of Heuristics
The tendency to perceive physically attractive stimuli as possessing favorable traits
is known as the what-is-beautiful-is-good heuristic[21]. Sleep-deprived participants gave a lower quality rating and were less likely to
purchase the unattractive refrigerator than the participants who had slept. Those
who had not slept therefore used the what-is-beautiful-is-good heuristic to a greater
extent than did the controls. Greater reported fatigue predicted a higher quality
rating for the attractive refrigerator and a rating of lower quality for the unattractive
refrigerator. The use of the what-is-beautiful-is-good heuristic in the sleep-deprived
participants may be explained by the elevated fatigue levels, or the use of this heuristic
may reflect a common underlying physiological response to sleep deprivation.
The what-is-beautiful-is-good stereotype has been explored in a study of cognitive
load[37]. When cognitive load was high, consumer products which were unattractive but paired
with superior consumer reviews were deemed as low quality (i.e., judged through their
negative physical appearance). When cognitive load was low, participants judged these
products as higher quality (i.e., judged based on the favorable consumer review).
Similarly, the participants in the present study likely experienced a limitation on
cognitive resources and greater cognitive load due to sleep deprivation[38] and judged the unattractive refrigerator with the favorable consumer review by its
negative appearance rather than its favorable review. These findings are of considerable
importance since sleep deprivation may result in heuristic processing and judgment
based on appearance rather than the systematic, effortful processing of the important
details of a stimulus. Such limitations in processing may influence judgment in critical
situations in the workplace and personal settings. Future research may determine whether
sleep deprivation affects stereotypic judgment of gender, age, and ethnicity.
Sleep-deprived participants used the greedy algorithm heuristic[22] during the Following Instructions Task. Instead of reading the instructions, the
sleep-deprived participants completed more questions on the task than needed and thus
spent more time than required. Total sleep deprivation seems to limit the thorough
examination of stimuli and instead promotes decision making which relies on automatic
behavior. Such skipping of instructions saves energy and time in the short term but
can lead to errors, especially when instructions provide unique information.
In assessment of the speed-accuracy trade-off[23], among participants who chose the more difficult math problems (10 participants
in each condition), sleep-deprived participants had over five times the odds of choosing
these problems in order to complete the task more quickly when compared with participants
who slept at home. These findings trended toward significance, likely due to the smaller
sample of participants (n=20) choosing the more difficult problems who answered the follow-up question. Such
findings corroborate that total sleep deprivation imposes limitations on effort. In
comparison to the control group, sleep-deprived participants preferred to limit the
time engaged on the task, utilizing a time-conservation strategy. This strategy appears
to be an attempt to exert effort for less time, indicating they “traded” short-term
cognitive resources for escape from the task.
Though heuristics are used frequently in everyday life and can often help individuals,
they can also have deleterious consequences. In medical residents, who often suffer
from a lack of sleep, those with more experience were found to use the availability
bias heuristic; they made decisions in new cases based on previous cases rather than
using analytical reasoning[39]. Similarly, overconfident venture capitalists used heuristics by making decisions
based on past successes without taking time to process new information that would
improve their accuracy, resulting in more incorrect decisions[40]. Overall, the findings from the current study indicate the importance of sleep for
engagement in systematic mental processes.
The Role of Fatigue in Perception of Task Difficulty and Use of Heuristics
In the current study, sleep deprivation induced greater fatigue and predicted a higher
difficulty rating for reading the article, a greater estimated amount of time to read
the article, greater estimated time to complete the puzzle, a higher quality rating
for the attractive refrigerator, and a rating of lower quality for the unattractive
refrigerator. Though sleep-deprived participants, compared to those who slept, were
less likely to purchase the unattractive refrigerator with the favorable review, skipped
instructions, and trended toward choosing difficult math problems to save time, fatigue
was not associated with any of these outcomes. The considerable variability in sensitivity
to the effects of subjective fatigue on cognitive performance may explain these findings[41]. That is, sleep may lead to a limitation in cognitive resources[38] which is accompanied by greater subjective fatigue in some individuals, but not
in others. Future studies may identify predictors of sensitivity to subjective fatigue
following sleep deprivation, and which factors predict inter-individual differences
in the association between fatigue and use of cognitive heuristics.
The findings from the current study indicate that fatigue induced by sleep deprivation
may influence critical decision making outside of the lab environment. Judges who
are sleep deprived and use the what-is-beautiful-is-good heuristic may make more favorable
rulings for attractive people. Sleep-deprived healthcare professionals may ignore
hand-washing instructions, and physicians who have not obtained sufficient sleep may
wish to complete tasks quickly, such as prescribing antibiotics rather than discussing
alternative treatment plans. Sleep-deprived law enforcement may be more likely to
judge a criminal suspect as a threat based on the suspect›s fulfillment of stereotypes,
critically impacting the officer›s decision to act with force against the perceived
threat.
Sleep loss increases fatigue and affects millions. Approximately 29% of adults in
the United States report getting less sleep than they need each night, with 27% of
respondents reporting being unable to work efficiently because they are too sleepy[42]. Given the findings from the current study, millions of American adults may be vulnerable
to perception of greater difficulty and use of heuristics and, consequently, errors
in judgment and decision making. This is particularly concerning given the necessity
of making decisions based on the careful estimation of the alternatives and logic
rather than expedience. The present findings therefore emphasize the importance of
sleep in reducing perceptions of task difficulty, successful completion of such tasks
through utilization of effortful mental processes, and preservation of decision-making
skills.
Future Directions
Patterns of adenosinergic activity in the nucleus accumbens (NAcc) may constitute
the physiological substrates of behavioral effort reduction induced by sleep loss.
Neural activity during wake coincides with elevated metabolism and increased concentration
of extracellular adenosine in the central nervous system[43]. During sleep, cortical interstitial space increases dramatically, allowing for
the removal of toxins[44] including adenosine, higher levels of which are correlated with the subjective experience
of fatigue[45]. Sleep deprivation results in the up-regulation of adenosine receptors[46] in or close to the NAcc shell[47]. Adenosine, acting on A2A receptors in opposition to the dopamine (DA) D2 receptor system, modulates the activity of GABAergic neurons within the NAcc, reducing
arousal and initiating sleep via multiple inhibitory projections throughout the arousal
system[47]. It is the A2A receptors, for example, that are uniquely receptive to the arousal effects of caffeine[48].
Separate from sleep investigations, researchers examining effort-related decision
making have found that the receptors responsible for arousal inhibition and sleep
promotion also regulate behavioral effort. Adenosine regulates effort-related processes
through a selective interaction between adenosine A2A receptors and antagonists of DA D2 receptors[49]. For example, A2A antagonists can reverse the behavioral effects of DA antagonists on effort-related
choice behavior[50]. These researchers suggest that stress on this system may be responsible for fatigue
and psychomotor slowing[51]. Sleep deprivation may, indeed, be a stressor affecting the interaction of A2A and D2 receptor systems. Sleep is promoted through adenosinergic activity via the A2A receptors and effortful behavior is inhibited at the same synapses. Future studies
examining the production of adenosine during wakefulness, stimulated up-regulation
of adenosine receptors in the NAcc after sleep deprivation, and adenosinergic projections
that inhibit arousal may help clarify the mechanisms in the cascade responsible for
reduced behavioral effort due to sleep loss.
Limitations
This was a naturalistic study; the purpose was to compare individuals maintaining
their typical sleep patterns with those who were totally sleep deprived. The participants
were randomly assigned to experience either total sleep deprivation or sleep at home
according to their natural sleep patterns. Thus, in the home sleep group, we did not
control the amount of sleep obtained. As a result, some participants who slept at
home had less sleep on the night prior to the Final Assessments than they reported
needing in order to feel fully rested, as indicated by the actigraph data. In essence,
the control group was given the opportunity for full sleep but instead experienced
a naturally-induced sleep deficit.
Physiological and cognitive changes consistent with some sleep loss, including fatigue,
are likely to have been experienced in this group. In the current study, increased
levels of fatigue were predictive of effort-related performance impairments, with
the greatest effort-related impairments produced following no sleep. The inadequate
amount of sleep experienced by the control group is consistent with other studies
which have found that less than one third of college students receive eight hours
or more of sleep each night[52].
Thus, the average sleep length for the control group may offer a realistic representation
of college students’ sleep habits. Nonetheless, our results suggest that even under
natural sleep conditions, when participants are permitted to sleep as they would normally,
increased fatigue is related to greater perceptions of task difficulty and use of
heuristics on effort-related tasks. Thus, total sleep loss impairs effort-related
performance when compared with a naturally-experienced minor sleep deficit. Our findings
also suggest that having some sleep confers benefits on effort-related performance
in comparison to the total absence of sleep. Future experimental studies designed
to enforce adequate sleep will clarify differences in effort between the full complement
of sleep and naturally-experienced sleep, which may include naturally-experienced
sleep loss.
The absence of between-group differences on some variables could be explained by the
partial sleep loss experienced by those who slept at home. Specifically, perception
of difficulty in the Puzzle Task, quality rating and purchase likelihood for the attractive
refrigerator with the unfavorable review, and the Math Difficulty-Time Choice did
not significantly differ between groups. Future studies which compare the use of heuristics
by those who have had their full sleep complement with those who have been sleep-deprived
might show greater between-groups differences in these variables.
Participants in NES and TSD experienced different settings for the overnight session
on the night prior to the Final Assessments; the former slept in their home environment,
and the latter remained awake in the sleep laboratory. This experimental design allowed
for the examination of NES participants› performance following a night in their natural
sleep environment as opposed to the unfamiliar lab setting. Sleeping in an unfamiliar
environment may result in poorer sleep quality or quantity, also known as the first
night effect[53]. Having participants sleep at home was intended to reduce the first night effect.
However, exposure to different settings prior to the Final Assessments could affect
performance in subtle ways. The effects, if any, of differences in pre-assessment
context on perception of task difficulty and use of heuristics could be assessed in
future studies. Furthermore, participants’ sleep duration and quality were not assessed
in the period prior to the overnight monitoring session; thus, we cannot conclude
that both groups had similar sleep patterns immediately prior to entering the study.
The groups, however, did not differ on any of the self-reported variables measured
through the sleep questionnaire.
Moreover, participants were randomly assigned to groups which mitigated any potential
sleep differences between the groups. In addition, though participants were randomly
assigned to conditions, future studies would benefit from a within-subjects assessment
to provide an unequivocal understanding of the impact of no sleep on the use of heuristics.
The available tasks used to assess perception of difficulty and use of heuristics
do not have equivalent parallel forms, which precluded a within-subjects design in
the current study.
CONCLUSION
Our findings demonstrate the effects of sleep deprivation and fatigue on perception
of task difficulty and use of heuristics. Sleep deprivation induces greater self-reported
fatigue, which is associated with perception of greater task difficulty. Due to this
change in perception, sleep-deprived individuals may attempt to compensate for their
limitations by using heuristics rather than complex mental processes. In the current
study, the sleep-deprived participants perceived the Article Task as more difficult
and used the what-is-beautiful-is-good, greedy algorithm, and speed-accuracy trade-off
heuristics. The results from this study emphasize the importance of examining the
various ways in which sleep deprivation and fatigue affect the perceived difficulty
of tasks, effort expenditure, and critical real-world outcomes.
