James Madison University
JMU Scholarly Commons
Masters Theses
The Graduate School
Summer 2014
The Effects of Chronic, Partial Sleep Deprivation
on Risk-Taking Behavior in Rats
Ashley Marie Shemery
James Madison University
Follow this and additional works at: http://commons.lib.jmu.edu/master201019
Part of the Psychology Commons
Recommended Citation
Shemery, Ashley Marie, "The Effects of Chronic, Partial Sleep Deprivation on Risk-Taking Behavior in Rats" (2014). Masters Theses.
Paper 323.
This Thesis is brought to you for free and open access by the The Graduate School at JMU Scholarly Commons. It has been accepted for inclusion in
Masters Theses by an authorized administrator of JMU Scholarly Commons. For more information, please contact [email protected].
The Effects of Chronic, Partial Sleep Deprivation on Risk Taking Behavior in Rats
Ashley M. Shemery
A thesis submitted to the Graduate Faculty of
JAMES MADISON UNIVERSITY
In
Partial Fulfillment of the Requirements
for the degree of
Master of Arts
Department of Graduate Psychology
July 2014
Table of Contents
List of Tables .........................................................................................................................iii
List of Figures ........................................................................................................................iv
Abstract ..................................................................................................................................v
I.
Introduction ................................................................................................................1
Sleep Deprivation and Decision Making .......................................................2
Sleep Deprivation and Risk Taking ...............................................................3
Defining Risk Taking.....................................................................................3
Measuring Risk Taking ..................................................................................6
Face and Construct Validity of RGT .............................................................8
Rationale ........................................................................................................9
II.
Methodology ..............................................................................................................10
III.
Subjects ..........................................................................................................10
Rodent Gambling Task (RGT).......................................................................11
Sleep Deprivation Wheel ...............................................................................13
Training ..........................................................................................................13
Results ........................................................................................................................15
IV.
Discussion .................................................................................................................16
Limitations ....................................................................................................22
Future Directions ..........................................................................................24
V.
References ..................................................................................................................29
ii List of Tables
Table 1: Means and Standard Deviations of Risk-Taking Behavior
Table 2: Displays the most favored hole each day of the experimental procedure for F-1
Table 3: Displays the most favored hole each day of the experimental procedure for F-2
Table 4: Displays the most favored hole each day of the experimental procedure for F-3
Table 5: Displays the most favored hole each day of the experimental procedure for F-4
Table 6: Displays the most favored hole each day of the experimental procedure for F-5
Table 7: Displays the most favored hole each day of the experimental procedure for Z-1
Table 8: Displays the most favored hole each day of the experimental procedure for Z-2
Table 9: Displays the most favored hole each day of the experimental procedure for Z-3
Table 10: Displays the most favored hole each day of the experimental procedure for Z-4
Table 11: Displays the most favored hole each day of the experimental procedure for Z-5
iii List of Figures
Figure 1: Represents the four variations of the RGT that were coded into MED-PC.
Figure 2: Displays mean number of pellets obtained across the duration of the experiment
Figure 3: Displays the performance of each rat individually. The y-axis displays the total
number of pellets obtained. The axis displays the day of the experiment. Note that the yaxis of the graph for rat Z2 is much larger due to his unique performance
Figure 4: Displays the performance of each rat individually. The y-axis displays the
percentage of choices made. The solid line with filled triangle markers represents
advantageous choices. The dashed line with open triangle markers represents
disadvantageous choices. The x-axis represents day of experiment.
iv Abstract
The effects of sleep deprivation on risk-taking behavior have been minimally
investigated, observing only the effects of total sleep deprivation in human models.
Additionally, the research has shown mixed results. In a chronically sleep deprived
society where many people (e.g. military, medical doctors) require rapid decision making
to ensure the safety and welfare of others, it is of interest to investigate the effects of
chronic partial sleep deprivation on risk-taking behavior in a rodent model. The current
study examined the effects of 5 days of partial sleep deprivation on risk-taking behavior
in Wistar Han rats as measured by the Rodent Gambling Task. Ten rats were placed in a
slowly rotating wheel for 18 hours a day for 5 consecutive days followed by 2 days of
recovery sleep. The rats’ risk-taking behavior was measured each day using the Rodent
Gambling Task (RGT). This task allows rats to nose-poke among 4 holes to obtain an
immediate reinforcer along with a probabilistic delay. Larger, immediate reinforcers are
associated with longer probable delays whereas smaller immediate reinforcers are
associated with shorter probable delays. Thus, across time, selecting the larger,
immediate reward is considered to be a risky decision, as the rats lose the opportunity to
gain reinforcers. We measured behavior across the entire experimental period and during
two “recovery” days where they could obtain ad lib sleep. There was no significant effect
of sleep deprivation on risk-taking behavior. There are multiple explanations for these
results, one of which suggests that the RGT is not a valid measure of risk-taking
behavior. More biologically relevant measures of risk-taking behavior should be
explored.
v The Effects of Chronic Partial Sleep Deprivation on Risk-taking Behavior in Rats
Extensive research has revealed that at least one third of Americans are
chronically, partially sleep deprived, and many people sleep less than 6.5 hr each night
(Bonnet & Arand, 1995; Durmer & Dinges, 2005; Dinges, Rogers & Baynard, 2010,
Webb & Agnew, 1975). Chronic sleep deprivation is defined as sleep deprivation that
occurs repeatedly over the course of several days. Partial sleep deprivation is defined as
getting some sleep each 24 hr period, but less than 8 hr each time. Thus chronic, partial
sleep deprivation refers to obtaining less than 8 hr of sleep per day for several days in a
row. Chronic, partial sleep deprivation differs from that of acute, total sleep deprivation,
which entails refraining from sleep entirely, usually for 1 to 3 days at a time (Drummond
& McKenna, 2009). Chronic, partial sleep deprivation has pronounced negative effects
on cognitive function. More specifically, after just one night of sleep deprivation,
reaction times are slower, there is an increase in mood disturbances, vigilance decreases,
working memory worsens, and decision-making becomes impaired (Harrison & Horne,
2000; Killgore, Balkin, & Wesensten, 2006; Venkatraman, Chuah, Huettel, & Chee,
2007). Additional research has shown that cumulative nights of partial sleep deprivation
accrue an effect on cognitive functions that is proportional to the number of nights of
sleep deprivation (Dinges et al., 1997; Durmer & Dinges, 2005; Van Dongen, Maislin,
Mullington, & Dinges, 2003). In other words, cognitive function becomes increasingly
worse with each additional night of short-duration sleep. This accumulation described as
a sleep debt, results in deficits in cognitive function that may only be completely restored
after sleeping over and above a typical night of sleep (Van Dongen, Rogers, & Dinges,
2003).
2 Sleep Deprivation and Decision Making
Early research on decision-making focused on simple, monotonous tasks that
measure constructs such as cognitive speed, psychomotor skills, short-term memory, and
attention. Researchers found that performance on these tasks declined significantly after
just one night of sleep deprivation (Dinges & Kribbs, 1991; Kjellberg, 1975). However,
many researchers have recognized that these tasks are particularly sensitive to sleep
deprivation because of their monotony and lack of novelty rather than some other
underlying characteristic of the task (Harrison & Horne, 2000). Additionally, tasks that
are well learned may be sensitive to sleep deprivation for the same reasons. Although it is
useful to measure the effects of sleep deprivation using these simple, monotonous tasks
these findings may have limited utility because they are not as intricate as real-world
situation. Using more complex tasks to measure decision-making processes may provide
more insight regarding how sleep deprivation affects every day decision making.
Compared to simple decision-making tasks, more complex tasks seem to have a
variable relation with sleep deprivation. Specifically, performance on complex tasks that
require participants to follow rules and demonstrate logic (IQ tests & critical reasoning)
are less impacted by sleep deprivation. Researchers attribute this to the novelty of the
tasks (Harrison & Horne, 2000). Harrison and Horne (2000) stated that if complex, novel
tasks become well learned, they then become susceptible to sleep deprivation. Thus it
might be logical to conclude that novel, “on-the-fly”, real-world decisions should not be
susceptible to sleep deprivation. However, research has shown this is not the case. In fact,
just one night of sleep deprivation causes impairments in skills required for everyday
decision making (Harrison & Horne, 2000).
3 Sleep Deprivation and Risk-Taking Behavior
Although decision making appears to be negatively impacted by chronic, partial
sleep deprivation, the literature is not clear about the relationship between risky decision
making and chronic partial sleep deprivation. In fact, studies of the effects of sleep
deprivation on risk-taking have neglected chronic partial sleep deprivation completely;
instead, research has focused on the effects of total sleep deprivation. Within this limited
literature, however, some research has shown that risk-taking behavior either does not
change, or actually decreases when participants are sleep deprived (Acheson, Richards,
de Wit, 2007; Killgore, 2007), whereas other research has shown that sleep deprivation
increases risk-taking behavior (Harrison & Horne, 1998; Killgore et al., 2006; McKenna,
Dickinson, Orff, & Drummond, 2007). Note that even within the same author (Killgore),
the results have been mixed.
Defining Risk Taking
A reason that these findings are mixed may come from an inconsistency in
defining and measuring the construct of risk-taking. Risk-taking is a multidimensional
construct that may be defined and measured in different ways (Leigh, 1999). Leigh
(1999) states that risk-taking behavior includes at least five possible dimensions of risktaking behavior. One possible dimension of risk-taking weighs the negative aspects with
the positive aspects. Positive aspects might be creativity or bravery whereas negative
aspects might be failure or danger. A second possible dimension of risk-taking observes
the differences between potential acute or chronic harm. Some behaviors could result in
acute harm such as gambling or jumping from high levels whereas other behaviors may
4 result in chronic harm such as a poor diet or smoking cigarettes. A third possible
dimension of risk-taking addresses the generalizability of risk-taking measures. For
example, someone may take risks with money but not with drugs. A fourth possible
dimension of risk-taking assesses knowledge of probability of harm. Risk entails
knowing the probability of harm, whereas uncertainty implies not knowing the
probability. Decision theory suggests that people are more sensitive to risk than
uncertainty (Tversky & Fox, 1995). Finally, a fifth possible dimension of risk-taking
compares objective versus subjective risk. Epidemiologically populations may be at risk
for contracting diseases, but on the individual level, this risk may be nearly non-existent
(Leigh, 1999). In other words, an objective risk that exists on the epidemiological level
(e.g. the number of people at risk for contracting a disease) may not exist on the
subjective level (e.g. a person that lives in an area where this disease does not exist).
It is interesting that these dimensions of risk-taking, and most definitions of risktaking, are usually dependent upon human mental capacities. The way in which risktaking is defined and measured changes necessarily from humans to animal models.
Humans possess language, which allows for far more complex behaviors than can be seen
in animals. For example, language allows humans to make rules about how to proceed in
specific situations. In fact, many decisions humans make are governed by rules (Catania,
Shimoff, & Matthews, 1989). Because of their lack of language, rats do not develop and
respond to rules. Instead, rats and other animals make decisions based on the true
contingencies in place (Catania et al., 1989). Because of this range of complex behaviors
available to humans, risk-taking can encompass many different things. Risk-taking can be
seen in social settings when someone engages in a “dare” or desires to be seen as brave or
5 heroic. Risk-taking can also be seen in more economic contexts as when humans gamble,
buy, sell, or trade goods. Thus, it is not surprising that the dimensions of risk-taking that
can be defined and measured in humans differ from those of a rat (Womack et al., 2013).
Do animals gamble? Specific to gambling tasks, there are two fundamental
differences in which measuring risk-taking behavior in humans differs from that of
animals. First, there is nothing that can be perceived as valuable and worthy of saving for
potential use as a token to a rat. When food deprived rats receive sucrose pellets as a
reward, they will likely consume those pellets immediately, rather than save them and
note the value of them. Second, in human gambling tasks, money is used as a reinforcer
as opposed to food in rat gambling tasks. In other words, humans are gambling with
secondary reinforcers whereas animals are gambling with primary reinforcers. Thus to
develop a gambling task for rodents, a flexible definition should be used.
Leigh (1999) defines risk-taking behavior entails an action that has the potential
for both (a) danger, loss, or harm; and (b) gain of some reward (Leigh, 1999). The
measure developed by Rivalan et al. (2009) and used in the current study, the Rodent
Gambling Task (RGT), has the potential for loss and gain of reward. The potential losses
come in the form of losing an opportunity to gain reward (having a probabilistic time-out
period). The potential gain comes in the form of always receiving a reinforcer for a
response, but not during the probabilistic time-out periods. The RGT is discussed in more
depth within the method section of this document.
6 Measuring Risk Taking
In addition to the RGT, several other rodent models of risk-taking behavior exist,
including the elevated-plus maze task, the light-dark emergent task, delay discounting,
and the risky-decision making task. For a variety of reasons these tasks were not selected
for the current study. The elevated-plus maze task and the light-dark emergence task have
not been implicated in the risk-taking literature. Instead, they have been used as an
industry standard for measuring anxiety in rodents (Ardayfio & Kim, 2006; Lister, 1987;
Pellow, Chopin, File, & Briley, 1985).
Delay discounting tasks were considered but not selected because the tasks
require several weeks to achieve a stable response in an animal (Green, Myerson, Holt,
Slevin, & Estle, 2004). With the nature of sleep research, measuring the effects of a night
of sleep deprivation must occur immediately after that night of sleep deprivation.
Otherwise, by the passage of time, the organism will become more sleep deprived by
remaining awake even longer, or less sleep deprived by falling asleep.
Finally, the risky-decision making task was considered. In this task, animals lever
press to obtain sucrose pellets, but also run the risk of receiving an electrical shock.
However, much literature as shown that pain perception increases with sleep deprivation
(Cohen & Dement, 1965; Lautenbacher, Kundermann, & Krieg, 2006). Furthermore, this
task requires a few days to achieve stable responses (Simon, Gilbert, Mayse, Bizon, &
Setlow, 2009).
The RGT appeared to be the most appropriate measure to observe the effects of
partial sleep deprivation on risk-taking behavior. It has depth and complexity by
7 providing multiple choices for the rat to choose from, it produces a stable and reliable
measurement in a single one-hour session, and it does not use painful stimuli (Rivalan et
al., 2009).
The RGT is an animal adaptation to the commonly used human gambling task, the
Iowa Gambling Task (IGT). In the IGT participants choose cards among four separate
decks to earn money. Two of the decks are considered advantageous as they result in the
immediate reward of $50 but also the probability of losing $50-250. The other two
decks are considered disadvantageous because although they result in $100 immediately,
they are correlated with the probability of loosing between $250-$1250. Ultimately,
across time, selecting cards from the advantageous deck results in more money than
selecting cards from the disadvantageous deck. Typically, participants select from the
decks at random until they have learned the contingencies, at which point they select
primarily from the advantageous decks (de Visser et al., 2011).
There are currently four versions of the RGT that have been adapted from the IGT
(Pais-Vieira, Lima, & Galhardo, 2007; Rivalan, Ahmed, & Dellu-Hagedorn, 2009; van
den Bos, Homberg, Gijsbers, Den Heijer, & Cuppen, 2006). Although each version of the
RGT differs from the others, they all maintain important characteristics of the IGT. Of
the four RGTs available, we selected the Rivalan et al. (2009) version. The distinguishing
feature of this RGT was the ability to obtain reliable results within a single 1hr session.
As previously mentioned, this is a crucial characteristic when assessing the effects of
sleep deprivation.
8 Face and Construct Validity of RGT
The face and construct validity of Rivalan’s (2009) RGT has been addressed.
Rivalan et al. (2009) found that rats tended to select among the four choices at random
until becoming acquainted with the contingencies. By the end of a single one-hour
session, most of the rats recognized and preferred the more advantageous choices.
However, 40% of the rats failed to show this pattern of behavior after a single one-hour
session. Instead this subset of rats had either no preference, or a strong preference for
disadvantageous choices. It has been suggested by de Visser et al. (2011) that this is
evidence of face validity, as humans tend to perform on the IGT in the same manner.
Most human participants will initially select among the four decks at random and most
will quickly learn to select from the advantageous decks. Approximately 20-30% of
human participants prefer the disadvantageous decks even after substantial opportunity to
become familiar with the contingencies in place (similarly to the 40% of rats who show
this same pattern) (Brechara, Damasio, Tranel, & Anderson, 1998; Crone & van der
Molen, 2004; Dunn, Dalgleish, & Lawrence, 2006). These participants are thought to be
impulsive (Davis, Tweed & Curtis, 2007; Franken, Van Strien, Nijs, & Muris, 2008).
Davis et al. (2007) and Franken et al. (2008) found that participants who scored high on
impulsiveness scales were unable to optimize their gains on the Iowa Gambling Task.
Such traits may be considered analogous to risk-taking and sensitivity to reward in rats
that are deemed poor decision makers.
Construct validity for the Rivalan et al. (2009) version of the RGT is currently
being evaluated. In human studies, damage to the ventromedial prefrontal cortex
(vmPFC) has been associated with poor performance on the IGT (Bechara et al., 2000).
9 Additionally, human studies have revealed a relation between orbitofrontal cortex (OFC)
activation and good performance on the IGT (Bolla, Eldreth, Matochik, & Cadet, 2004).
Specifically, higher OFC activation is related to good performance on the IGT. de Visser
et al. (2011) presented unpublished data suggesting that lesions to the orbitofrontal cortex
impair rats’ ability to quickly recognize and develop a preference for the more
advantageous choice. Furthermore, c-fos expression in the OFC of rats differentiated
good decision makers and poor decision makers (de Visser et al., 2011).
Rationale
Healthcare providers, military commanders, aircraft pilots, policemen, truck
drivers and other shift workers depend on their ability to make sound decisions, often in
the spur of the moment, and may need to make decisions that require them to take risks.
Individuals in these positions often suffer from a large sleep debt due to their work
schedules. When lives may be at risk, it is important to understand the relationship
between sleep deprivation and decision making and risk-taking behavior. Understanding
the effects of chronic, partial sleep deprivation on risk-taking behavior could help guide
sleep practices in high-stakes careers. Animal models may bee especially helpful in
understanding the effects of sleep deprivation on decision-making. Not only are human
sleep studies expensive and difficult to coordinate, but also animal models allow for
much tighter experimental control. Additionally, when animal models are developed it is
possible to conduct studies to understand the brain-behavior mechanisms involved.
10 Conclusion
We mimicked the human work week, and observed the effects of 18 hr of sleep
deprivation, for five consecutive days on the risk-taking behavior of Wistar Hans rats.
We hypothesized that sleep deprivation would increase risk-taking behavior in rats as
measured by the RGT. Additionally, we observed the effects of two days of sleepdeprivation recovery on risk-taking behavior. We hypothesized that the rats would
gradually make more risky decisions on the RGT with each additional day of sleep
deprivation. During the recovery phase, we hypothesized that risk-taking would trend
towards baseline, but not reach it. In other words, as the rats became more sleep deprived
they would nosepoke more on the disadvantageous choices than the advantageous
choices, and during recovery the amount of sleep generated during the two-day period
may not be enough to return to baseline levels of risk-taking behavior.
Method
Subjects
Twelve Wistar Hans rats aged 12 to 13 weeks were used. This particular strain
was selected based on use in previous literature (Rivalan et al., 2009). The rats were
housed individually in a temperature-controlled room (22-23°C) on an inverted 12 hr
light/dark cycle (lights on at 1200). They had free access to water and were food deprived
(80%-85% of free-feeding weight) throughout the duration of the experiment.
11 Rodent Gambling Task (RGT)
An operant chamber 12.0" L x 9.5" W x 11.5" H (Med Associates) adapted from
the five-choice serial reaction time chamber was used for the RGT. Four circular holes
that permitted nose pokes were available on a curved wall and could be dimly lit with a
white light-emitting diode (LED) located at their rear. A fifth hole that was not used in
this experiment was positioned between the four holes that were used. A food receptacle
on the opposite wall was connected to an external food pellet (45mg sucrose pellets)
dispenser. A partition with a central opening was placed in the center of the chamber
parallel to the food wall. This allowed the rats to be oriented at an equal distance to all
nose pokes avoiding thigmotaxic behavior (or in other words, side preferences).
In the RGT rats can nose poke among four different holes to obtain food rewards
during a 1 hr session. Two of the options are considered advantageous options while the
other two options are considered disadvantageous options. A poke in hole is immediately
rewarded (gain) by either 1 or 2 pellets, but has the probability of a time-out (loss) of
variable intensity during which the rat may not poke for pellets.
Of the two disadvantageous options, a poke in one hole results in a gain of two
pellets immediately, but has the probability of relatively long (342s), frequent timeout
that occurs 50% of the time that it is selected. The second disadvantageous hole also
results in a gain of two pellets immediately, but has the probability of a relatively very
long (684s), less frequent timeout that occurs 25% of the time that it is selected. The hole
that results in more frequent timeouts (50% of the time) has a shorter timeout period
(342s) than the hole that results in less frequent (25% of the time) timeouts (684s). This
12 allows for the two disadvantageous holes to be equal in the number of possible
reinforcers that may be obtained over the 1 hr session.
Of the two advantageous options, a poke in one hole results in a gain of one pellet
immediately, but has the probability of a relatively short (36s), less frequent timeout that
occurs 25% of the time that it is selected. The second advantageous hole also results in
the gain of one pellet immediately, but has the probability of a relatively very short (18s),
more frequent timeout that occurs 50% of the time that it is selected. The hole that results
in more frequent timeouts has a shorter timeout period (18s) than the hole that results in
less frequent timeouts (36s). This allows for the two advantageous holes to be equal in
the number of possible reinforcers that may be obtained over the one-hour session.
Although the advantageous choices produce less pellets immediately, the much shorter
delays associated with the advantageous choices allow for significantly more pellets to be
obtained within a 1 hr session compared to the disadvantageous choices. The exact
amount of time for each timeout was calculate using a formula provided by Rivalan et al.
(2009). The theoretical maximum number of pellets a rat may obtain if it only poked the
advantageous holes is 200 pellets. The theoretical maximum number of pellets a rat may
obtain if it only poked the disadvantageous holes is 40 pellets.
To avoid spatial preferences and order effects, four variations of the task were
created and pseudo-randomly assigned to each rat on each of their eight experimental
days. In an attempt to replicate the original methodology as strictly as possible we
maintained a fundamental aspect of the task in which the two advantageous choices are
always grouped together and the two disadvantageous choices are always group together
(see Figure 1).
13 Sleep Deprivation Wheel
The sleep deprivation wheel consisted of aluminum rungs covered in a plastic
mesh (to keep the rat’s tail from getting caught outside the wheel) and polycarbonate
sides. The internal width is 4.4in with an internal diameter of 13.38 inches. There are 82
rungs 0.188 inches in diameter with 0.526 inch spacing. The wheel is driven by a motor
at 1.5 meters per min (Lafayette Instrument: Neuroscience). There is a hole on one side
of the wheel through which the mouthpiece of a water bottle fits, allowing the rat ad lib
access to water. Food was placed inside of the wheel.
Training
All rats completed two different shaping procedures before performing on the
RGT. First, rats were hand shaped using successive approximations to poke on a fifth
hole located between the four holes used in the RGT. Because this hole is centered and
not used in the RGT, it allowed for control over the development of potential spatial
preferences. After reaching the criteria of making 50 consecutive nosepokes, the rats
were then shaped on a response generalization procedure.
In the generalization procedure, each of the four holes used in the RGT were
presented randomly without replacement until 25 responses were made for each hole.
Each response was reinforced with a single pellet, totaling 100 pellets at the end of each
generalization session. The rats performed on the generalization procedure for three
consecutive days. All rats met the criteria on the first day but two additional days allowed
for extinction of responses on holes not signaling a scheduled reinforcer.
14 Procedure
Data were collected over a period of 8 days for each rat. This included 1 day for
baseline testing, 5 days of sleep deprivation, and 2 days of recovery. Due to not having
enough equipment, the rats were divided into two groups. The two groups performed all
of the same tasks, but one hr apart. All of the “F” rats performed the RGT at 2:00 P.M.
and were placed in the sleep deprivation wheel until 9:00 A.M. All of the “Z” rats
performed the RGT at 3:00 P.M. and were placed in the sleep deprivation wheel until
10:00 A.M.
On the first day, the rats were tested on the RGT at either 2:00 P.M. or 3:00 P.M.
to obtain baseline risk-taking performance. The times, 2:00 P.M. or 3:00 P.M. were
selected because they fall a few hours after lights off (12:00 P.M.) during the normal
waking phase for the rats. Following a 1 hr RGT session the rats were placed in a sleep
deprivation wheel for 18 hr until either 9:00 A.M. or 10:00 A.M. The rats were fed once
each day immediately after being placed in the wheel. The amount of food was calculated
by factoring in how much food was obtained during the RGT. The rats were kept at 85%
of their free feed weight. This time was chosen to allow for maximum food deprivation
during the next day’s RGT. At 9:00 A.M. or 10:00 A.M., the rats were removed from the
wheel and placed into their home cages with 5 hr until their next experimental session.
We observed that the rats spent most of this time sleeping. This procedure continued for
the 5 sleep deprivation days. During the recovery days, the animals were tested at their
same respective times, but put back into their home cages after performing on the RGT
rather than the sleep deprivation wheel.
15 The number of trials per 1 hr session varied by session. The number of trials in a
session depends solely on the successive choices the rat makes. The more
disadvantageous choices a rat makes, the more likely it is to experience longer time out
causing an overall lower number of trials in a session compared to a rat that makes more
advantageous choices.
Because there was only one operant box outfitted for the RGT and two sleep
deprivation wheels, two rats performed every 8 days. During the first 8 days of the
experiment rats F-1 and Z-1 performed, during the second 8 days of the experiment F-2
and Z-2 performed, etc.
Results
To evaluate the effects of sleep deprivation on risk taking behavior a one-way
repeated measures ANOVA was performed. The independent variable included eight
levels: baseline, 5 days of sleep deprivation, and 2 days of recovery sleep. The dependent
variable was total number of pellets obtained each session. Mauchly’s test of sphericity
was significant (p < .001) indicating a violation of the assumption of sphericity. The
Greenhouse-Geisser correction for degrees of freedom was used. The omnibus test
yielded no significant differences, F(1.695, 15.258) = .632, p = .520 (see Table 1 for
means and standard deviations). There was no effect of sleep deprivation on risk-taking
behavior as measured by the RGT. Because the aggregate data displayed a cubic function
(see Figure 2), a trend analysis was considered. Despite the visual appearance of a cubic
trend, there was not a significant cubic relation between sleep deprivation and risk-taking
as measured by the RGT, F(1, 9) = .742, p = .412.
16 To better visualize the large amount of variability contributing to the lack of
significance, individual plots were created for each rat (see Figure 3). Based on the
individual plots, it seems as though only two rats (F1 and Z5) show a similar pattern of
behavior in which they start out earning the most pellets, and gradually across time earn
less pellets.
Discussion
This study evaluated the effect of chronic, partial sleep deprivation on risk-taking
behavior as measured by the RGT. This task was expected to produce results that are
analogous to human performance on the Iowa Gambling Task (Rivalan et al., 2009).
Unlike previous research on sleep deprivation and risk-taking behavior, this study
examined partial sleep deprivation rather than total sleep deprivation. We examined
behavior change during five consecutive days of partial sleep deprivation followed by
two days of recovery, which mimics the human workweek. We were not able to
demonstrate that partial sleep deprivation significantly changes risk-taking behavior as
measured by the RGT. In fact, it appears that under the contingencies of the study, each
of the rats displayed a unique response pattern.
There are at least four possible explanations for this finding: 1) the rats were not
sufficiently sleep deprived to impact their performance; 2) sleep deprivation does not
affect risk-taking behavior in any capacity; 3) sleep deprivation does not affect the
dimension of risk-taking behavior that is measured by the RGT; or 4) the RGT is not a
valid measure of risk taking, and is therefore not sensitive to sleep deprivation. Each
potential explanation is discussed in further detail below.
17 One possible explanation for our non-significant results might be that the rats
were not sufficiently sleep deprived to impact their performance. Unfortunately there are
no published studies showing the effects of partial sleep deprivation on risk-taking to use
for comparison. However, indirect evidence from human studies suggests that allowing
rats only 5 hr of sleep per night should be enough to impact risk-taking behavior,
assuming that risk-taking behavior is sensitive to sleep deprivation. A study by Van
Dongen et al., (2003) showed that when humans are allowed only 4 hr of sleep per night
for several nights, their performance on many neurocognitive tasks suffers immensely.
Humans need approximately 8 hr of sleep per night, while laboratory rats need
approximately 13 hr of sleep per night (Antrobus et al., 1997). Restricting rats to only 5
hr of sleep per night from their typical 13 hr (38% of typical amount of sleep) is
theoretically more taxing than restricting humans to only 4 hr of sleep per night from
their typical 8 hr (50% of typical amount of sleep). Thus if humans show severe
neurocognitive dysfunction when obtaining only 4 hr of sleep each night for several
nights, one might expect rats to also show severe neurocognitive dysfunction when
obtaining only 5 hr of sleep each night for several nights. However, this is indirect
evidence, and it may be possible to conclude that the rats were not sleep deprived enough
for their risk-taking behavior to be affected.
A second explanation for our non-significant results might be that sleep
deprivation does not affect risk-taking behavior in any capacity. In other words, sleep
deprivation may not have any affect on risk-taking behavior no matter how much sleep
deprivation occurs. However, this idea may be untrue for at least two reasons. First, a
comprehensive literature review performed by Womack et al. (2013), showed notably
18 mixed results. More specifically, four studies reported that sleep deprivation increases
risk-taking behavior (Killgore et al., 2006; Killgore et al., 2007; Killgore et al., 2010,
Killgore, et al., 2012), two studies reported that sleep deprivation decreases risk-taking
behavior (Killgore, 2007; Killgore et al., 2008), and seven studies reported mixed
findings where risk-taking behavior increased for some instances but decreased in other
instances (Acheson et al., 2007; Chaumet et al., 2009; Jones et al., 2006; Menz, Buchel,
& Peters, 2012; Venkatraman, Chuah, Huettel, & Chee, 2007; Venkatraman, Huettel,
Chuah, Payne, & Chee, 2011). While these may all be human studies, there seems to be
at least some effect of sleep deprivation on risk-taking behavior.
A second reason it may be unlikely that sleep deprivation has no effect on risktaking behavior is that our measure is only capable of measuring one dimension of risktaking behavior. If sleep deprivation has no effect on rat performance on the RGT, we
cannot conclude that sleep deprivation has no effect on rat performance on other
measures of risk-taking behavior.
Furthermore, it is known that sleep deprivation severely affects many aspects of
behavior and cognition. In addition to changes in risk-taking behavior, human studies
have shown that those who are sleep deprived cause more fatal accidents, have more
lapses in attention, and are unable to direct attention to relevant or appropriate cues
(Arnedt, Wilde, Munt, MacLean, 2001; Dawson & Reid, 1997; Doran, Van Dongen,
Dinges, 2001; Dorrian, Rodgers, Dinges, 2005; Killgore 2006; Venkatraman, 2007).
Supporting this, fMRI studies have shown an overall decrease in cerebral activation
relative to a rested state suggesting that the sleep deprived brain cannot allocate resources
appropriately (Chee & Choo, 2004; Drummond et al., 1999). Finally, animal studies have
19 shown that if forcibly subjected to sleep deprivation, rats will inevitably die within 2-3
weeks despite receiving adequate food and water (Rechtschaffen & Bergmann, 1995). It
seems that a rat would be likely to perform differently on a valid risk-taking task a day
prior to death by forcible sleep deprivation than when fully rested. In this situation sleep
deprivation may not directly affect risk-taking behavior, but rather it affects areas of the
brain necessary for mediating in risk-taking behavior. Ultimately, it may be that
regardless of how, sleep deprivation in some capacity could have an effect on risk-taking
behavior.
A third, more appropriate explanation for our non-significant results could be that
sleep deprivation does not affect the dimension of risk-taking behavior that is measured
by the RGT. Although it may seem reasonable to make this conclusion, the patterns of
responding seen in the individual plots (Figure 3) suggest otherwise. If there were no
effect of sleep deprivation on RGT performance, the results should display a stable
pattern of responding in which the total number of pellets obtained on each experimental
day should be approximately equal to that of the total number of pellets obtained during
baseline. Instead, each rat displays a variable and inconsistent pattern of responding.
This unpredictable and variable pattern of responding seems to indicate a flaw in the
measure as opposed to insignificant results due to no effect of sleep deprivation.
A fourth possible explanation for our non-significant results could be that the
RGT is not a valid measure of risk taking and is therefore not sensitive to sleep
deprivation. As previously mentioned, the rats’ performances on the RGT appeared to
show an unpredictable, random pattern. Half of the rats (F2, F4, Z1, Z2, and Z4) showed
sporadic “up-and-down” responding, but not in similar ways as one another (see Figure
20 3). On some days, the rats respond mostly on the disadvantageous choices, whereas other
days, the rats responded mostly on the advantageous choices. In other words, the rats’
respond unpredictably and inconsistently with one another. Also notable, one rat (Z2)
performed at levels beyond that of the other rats. Typically, the rats obtained between 50100 pellets per day, whereas Z2 obtained between 44 to 268 pellets per day. Thus the
overall instability in responding suggests that the RGT may not be a valid tool for
measurement.
In addition to the unstable, unpredictable performance patterns, there are
conceptual reasons for why the RGT may not be a valid measure of risk-taking behavior.
In the well-validated IGT used with humans, participants select among four decks of
cards (de Visser et al., 2011). Two decks are considered advantageous, and two decks are
considered disadvantageous. When selecting from the disadvantageous deck, a larger
amount of money is earned immediately, and there is a risk of losing a higher sum of
money. When selecting from the advantageous deck, a smaller amount of money is
earned immediately, and there is a risk of losing a much smaller amount of money. Thus
across time, selecting from the advantageous deck results in earning more money (de
Visser et al., 2011).
Although the RGT maintains many characteristics of the IGT, there are two
fundamental characteristics it fails to incorporate. First, the RGT uses a primary
reinforcer, food, which may have inherent biological importance unlike secondary
reinforcers such as money (Dewey, 2007; Green et al., 2004). Green at al. (2004) explain
that rats and pigeons seem to discount the value of rewards more steeply as a function of
time compared to humans. In simplified terms, rats and pigeons respond more
21 “impulsively” than humans do. Green et al. (2004) offer multiple explanations for why
these differences may exist, one of which suggests that primary reinforcers such as food
are necessary for survival whereas secondary reinforcers may not be.
Further illustrating the magnitude of the difference between primary and
secondary reinforcers, Dewey (2007) described an experiment carried out by Dr. Sarah
Boysen on chimpanzees. In Boysen’s experiment, there were two chimpanzees. One
chimp was presented with two plates of gumballs (primary reinforcer). One plate had
more gumballs that the other. Once the chimp selected a plate, the plate he selected was
given to the other chimp. Thus, to get the most gumballs for himself, the chimp needed to
point to the plate with less gumballs on it. However, the chimp could not learn this.
According to Boysen, the chimp would even display signs of frustration immediately
upon selecting the plate with more gumballs. When Boysen replaced the gumballs with
tokens however, the chimp could easily maximize his gains by pointing to the plate with
fewer tokens on it (Dewey, 2007). In other words, primary reinforcers may cause animals
to respond more impulsively or even more riskily than if secondary reinforcers could be
used. This is particularly relevant to the RGT because rats may be responding in ways
that appear to be “risky” because of an innate drive to respond impulsively to food.
Second, it is not possible to give a rat an item that the rat would deem valuable
and worthy of saving for use as a token later. Thus, when performing on the RGT, the
animal does not risk losing a valuable item; instead, the animal risks losing the
opportunity to gain a reinforcer. Certainly, by lacking these two characteristics, we may
conclude that the RGT and the IGT are not measuring the same dimension of risk-taking.
However, due to the species differences, along with the way we defined risk-taking
22 behavior, it may be debatable whether the risk of losing a secondary reinforcer is
qualitatively different than the risk of losing an opportunity to gain a primary reinforcer.
Ultimately, the inconsistent and variable performance on the RGT along with the
restricted way in risk-taking behavior must defined to use the RGT contribute to evidence
against the RGT as a valid measure of risk-taking behavior.
Limitations
There are several limitations to this study. First, we did not have a control group.
Although we used a within-subjects design in which each rat served as its own control,
having a separate control group could have provided insight into the unpredictable and
inconsistent RGT performance seen in our experimental group. For example, if the
control group also showed unpredictable and inconsistent RGT performance, it would
have further strengthened the idea that the RGT may not be a valid tool for measurement.
However, had the control group shown consistent and predictable performance on the
RGT, we could use the insight to further investigate the unique and variable performance
of the experimental group.
Second, we did not perform an exact replication of the original RGT experiment.
Rivalan et al. (2009) sought to validate the RGT by measuring the performance of over a
hundred rats. Most of their rats were not exposed to the RGT more than once, as one of
the major features of the RGT was that a rat could perform on the RGT one time and
produce results that would be reliable for several months. In other words, their RGT was
designed to provide a quick, reliable, one-time measure of risk-taking behavior.
However, we exposed the rats to the RGT for 8 days straight. To compensate for
23 potential side and/or hole preferences due to repeated exposure, we created four different
presentations of the holes and pseudo-randomly presented them to each rat over the
course of eight days (Figure 1). To maintain similarity to the original presentation of the
holes, the advantageous holes were always placed together and the disadvantageous holes
were always next to each other.
Although this was not an exact replication of the use of the RGT as proposed by
Rivalan et al., it should not be an issue of concern. Rivalan et al. explained that although
the rats make their first few choices, they learn the association of the holes and only
begin to make advantageous or disadvantageous preferences after the learned association.
Thus, having a new arrangement of holes each day mimics the same situation as the first
day of RGT exposure. One may argue though, that having any previous exposure at all to
the RGT differentially impacts the learning process taking place. It may be that by
changing the presentation of the holes each day no longer makes the task a quick
measure, but instead the rats must first learn that the presentation of the holes will always
vary.
Third, our sample size was far too small. Although an initial power analysis
indicated only nine subjects would be needed for sufficient power, this was based off
assumptions about the effect size. Because this version of the RGT is new, there were no
other publications at the time of the experimental planning to advise an appropriate effect
size in the power analysis. With such extreme variability in our data, a much larger
sample would be needed to discern with any certainty the overall pattern of responding.
Because of our small sample size, the claims we have made about the validity of the RGT
should be taken lightly.
24 Finally, it is possible that some of the rats developed hole preferences. As
previously mentioned, we developed four variations of the presentation of the four
options (ADV 50%, ADV 25%, DIS 50%, DIS 25%). These four variations (see Table 1)
were pseudorandomly presented across the 8 experimental days. Thus, a hole bias would
be choosing one hole, and consistently poking that hole each session despite the changing
order of the contingencies in place. Tables 2 through 11 provide the most preferred hole
per rat, per day of the experiment. Eight of the 10 rats show potential preferences for a
particular hole with selection of that hole ranging from 50% to 88% of the time.
However, for some rats, the “most preferred” hole was only selected once or twice more
than another hole. Additionally, there are factors weighting the options available
including sleep deprivation and the actual contingencies in place, making it difficult to
determine the expected percentage of hole preferences to which to compare the 50%88%. A possible solution to preventing potential hole biases in the future would be to
being each session with forced choice trials.
Future Directions
Animal models are useful in understanding human behavior. They are relatively
inexpensive and allow for a high level of experimental control. To fully understand the
effects of partial sleep deprivation on decision making in humans, it may be helpful to
establish a reliable and valid animal analogy.
The biggest challenge in measuring the effects of sleep deprivation on risk-taking
behavior in animal models is establishing a quick and valid measure of risk-taking
behavior. There are two directions that future research may explore. First, if the
25 discrepancies in the way of defining risk-taking during the rodent gambling task can be
accepted (i.e., not experiencing a loss of a valuable as well as using primary reinforcers)
the RGT could be altered for improvement. Second, if the discrepancies in the way of
defining risk-taking during the RGT cannot be accepted, more biologically relevant tasks
should be explored.
If the restricted definition of risk-taking that accompanies the use of the RGT can
be accepted, the RGT could be altered for improvement. Rivalan et al. (2009) claimed
that the task only needed to be performed once for the animal to discern the schedules of
reinforcement available. However, our data do not support this. As seen in the individual
plots, there is no consistency in responding (see Figure 3). The pattern of responding
seems unpredictable. Additionally, the aggregate data suggest there is a learning curve
that spans at least four sessions (see Figure 2). Thus one way to improve the RGT would
be to allow for at least five sessions during which the animal is exposed to the task.
A second way that the RGT may be improved is to change the dependent variable.
Currently, the dependent variable of the RGT is total number of pellets obtained.
Theoretically, if a rat is obtaining more pellets, it is making more advantageous choices
(see Method for explanation of probabilities). However, it is possible for a rat to obtain
less pellets than previously, yet be making more advantageous choices. If a rat is slower
to respond (overall less responding), but responds mostly on the advantageous holes, it
might appear that the rat is actually make more disadvantageous choices based on total
number of pellets obtained. A more appropriate dependent variable for the RGT may be
percentage of advantageous choices made. By using the dependent variable of percentage
of advantageous/disadvantageous choices made rather than total number of pellets
26 obtained, the performance of the rat will no longer rest on theoretical assumptions.
Instead, it will be an accurate representation of the rat’s performance regardless of the
time it takes to respond. This is particularly important in sleep deprivation research, as a
sleep-deprived rat might be lethargic in responding. Figure 4 shows the rats’ individual
plots using percentage of advantageous choices and disadvantageous choices as a
function of time. In a side-by-side comparison to the plots from Figure 3, the figures look
quite similar. However, there are a few instances in which the plots using the original
dependent variable (Figure 3)
First, in Figure 3, F-5’s performance is portrayed as poor on days 1-3 as well as
on days 7 and 8. However, according to Figure 4, F-5 made more advantageous choices
than disadvantageous choices on days 1-3. Additionally, on days 7 and 8, F-5 made an
equal amount of advantageous and disadvantageous choices. Second, although Z-3’s
performance appears to drop from day 3 to day 6, Z-3 consistently made more
advantageous choices than disadvantageous choices each of those days (see Figure 4).
Finally, despite making more advantageous choices than disadvantageous choice on day
6 (see Figure 4), Figure 3 portrays Z-5’s performance as poor compared to other days.
Although the plots in Figure three do not appear to be drastically different from those in
Figure 4, Figure 4 more accurately represents the performance of the rat on the RGT.
Rather than attempting to improve the RGT, it may be more appropriate to use an
entirely different measure of risk-taking behavior in sleep deprivation research. More
specifically, it may be helpful to use a more biologically relevant measure. For animals, a
true risk might entail exposing oneself to predators to potentially obtain food or crossing
27 a tall, narrow pathway to get to somewhere safe. Unlike humans, most risks for animals
entail risking their lives (Weber, Shafir, & Blais, 2004).
There are a couple of well-known, biologically relevant measures of risk-taking
or anxiety-like behaviors in animals: the elevated plus maze and the light-dark emergence
task (Misslin & Belzung, 1989; Pellow, Chopin, File, & Briley, 1985). The elevated plus
maze is a plus-shaped maze that is elevated approximately 50 cm and consists of two
closed arms and two open arms. Rats or mice that spend more time exploring open arms
are thought to have lower levels of anxiety and are more likely to take risks (Giorgetta et
al., 2012; Pellow, Chopin, File, & Briley, 1985). The other task known as the light-dark
emergence task considers the innate aversion rodents have to brightly lit areas. The lightdark emergence box has two compartments, a dark compartment and a light
compartment. The rodent is placed in the light compartment and the amount of time it
takes for the animal to run into the dark compartment as well as the amount of time spent
in each compartment reflects their level of anxiety (Bourin & Hascoet, 2003; Crawley &
Goodwin, 1980).
Initially, these tasks were considered but not chosen on the basis that they might
be to simplistic to capture the decision making process of risk-taking. Additionally, these
tasks are mostly implicated in research focused on anxiety rather than risk-taking.
However, human research has shown a strong relation between risk-taking and anxiety.
Specifically, those with high anxiety tend to take less risks (Giorgetta et al., 2012;
Simmons, Howard, Simpson, Akil, & Clinton, 2012). Furthermore, a couple of animal
studies have shown that there are markers of risk-taking behavior that can be assessed
when observing animal performance in the elevated plus maze, such stretched postures
28 and head dips (Laviola, Macrı̀, Morley-Fletcher, & Adriani, 2003; Macrı̀, Adriani,
Chiarotti, & Laviola, 2002).
A third possible task that could be used but is not currently practiced in the animal
risk-taking field is one that involves the rat jumping over water to a platform with food.
There could be a couple of options available to the rat: 1) Platforms placed relatively
close together in water where the food reward on the other side is small and 2) Platforms
place relatively far from one another in the water where the food reward on the other side
is large. If the rat falls in, it then suffers a time out before being able to choose again. In
this task, the risker decision would be to jump farther for a larger reward. It may even be
of interest to probabilistically vary the amount of reward on the other side. In other
words, the larger leap may come with a larger reward than the smaller leap, but not
always.
Overall, future research should further investigate the relationship between
anxiety behaviors and risk-taking behaviors in animals in hopes of finding a more
biologically relevant way to assess risk-taking behavior. With more biologically relevant
measures of risk-taking behavior, researchers can develop animal models to better
understand the effects of chronic, partial sleep deprivation on risk-taking behavior.
29 References
Acheson, A., Richards, J. B., & de Wit, H. (2007). Effects of sleep deprivation on
impulsive behaviors in men and women. Physiology & Behavior, 91(5), 579-587.
Antrobus, J., Balter, M., Battles, J., Berger, R., Bliwise, D., Carskadon, M., et al. (1997,
January 1). The phylogeny of sleep. Retrieved June 26, 2014, from
http://www.sleepsources.org/uploads/sleepsyllabus/b.html
Ardayfio, P., & Kim, K. S. (2006). Anxiogenic-like effect of chronic corticosterone in the
light-dark emergence task in mice. Behavioral neuroscience, 120(2), 249.
Arnedt, J. T., Wilde, G. J., Munt, P. W., & MacLean, A. W. (2001). How do prolonged
wakefulness and alcohol compare in the decrements they produce on a simulated
driving task?. Accident Analysis & Prevention, 33(3), 337-344.
Bechara, A., Damasio, H., Tranel, D., & Anderson, S. W. (1998). Dissociation of
working memory from decision making within the human prefrontal cortex. The
journal of neuroscience, 18(1), 428-437.
Bechara, A., Damasio, A. R., Damasio, H., & Anderson, S. W. (1994). Insensitivity to
future consequences following damage to human prefrontal cortex. Cognition, 50(1),
7-15.
Bechara, A., Damasio, H., Damasio, A. R., & Lee, G. P. (1999). Different contributions
of the human amygdala and ventromedial prefrontal cortex to decision-making. The
Journal of Neuroscience, 19(13), 5473-5481.
30 Bechara, A., Tranel, D., & Damasio, H. (2000). Characterization of the decision-making
deficit of patients with ventromedial prefrontal cortex lesions. Brain, 123(11), 21892202.
Bolla, K. I., Eldreth, D. A., Matochik, J. A., & Cadet, J. L. (2004). Sex-related
differences in a gambling task and its neurological correlates. Cerebral
Cortex, 14(11), 1226-1232.
Bonnet, M. H., & Arand, D. L. (1995). We are chronically sleep deprived. Sleep-New
York, 18, 908-911.
Bourin, M., & Hascoet, M. (2003). The mouse light/dark box test. European Journal of
Pharmacology, 463(1), 55-65.
Catania, A. C., Shimoff, E., & Matthews, B. A. (1989). An experimental analysis of rulegoverned behavior. In Rule-governed behavior (pp. 119-150). Springer US.
Chaumet, G., Taillard, J., Sagaspe, P., Pagani, M., Dinges, D. F., …, Philip, P. (2009).
Confinement and sleep deprivation effects on propensity to take risks. Aviation,
Space, and Environmental Medicine, 80, 73–80. doi:10.3357/ASEM. 2366.2009
Chee, M. W., & Choo, W. C. (2004). Functional imaging of working memory after 24 hr
of total sleep deprivation. The Journal of Neuroscience, 24(19), 4560-4567.
Crone, E. A., & van der Molen, M. W. (2004). Developmental changes in real life
decision making: performance on a gambling task previously shown to depend on
31 the ventromedial prefrontal cortex. Developmental Neuropsychology, 25(3), 251279.
Crawley, J., & Goodwin, F. K. (1980). Preliminary report of a simple animal behavior
model for the anxiolytic effects of benzodiazepines. Pharmacology Biochemistry
and Behavior, 13(2), 167-170.
Davis, C., Patte, K., Tweed, S., & Curtis, C. (2007). Personality traits associated with
decision-making deficits. Personality and Individual Differences, 42(2), 279-290.
Dawson, D., & Reid, K. (1997). Fatigue, alcohol and performance
impairment. Nature, 388(6639), 235-235.
de Visser, L., Homberg, J. R., Mitsogiannis, M., Zeeb, F. D., Rivalan, M., Fitoussi, A., ...
& Dellu-Hagedorn, F. (2011). Rodent versions of the Iowa gambling task:
opportunities and challenges for the understanding of decision-making. Frontiers In
Neuroscience, 5.
Dewey, R. (2007). Animal behavior and cognition. Psychology: An introduction.
Retrieved from http://www.intropsych.com/index.html
Dinges, D., Kribbs, N. (1991). Performing while sleepy: Effects of experimentally
induced sleepiness. In T. H. Monk (Ed.) Sleep, sleepiness, and performance (pp. 97128). Winchester, England: Wiley.
32 Dinges, D., Rogers, N., & Baynard, M. (2010). Chronic sleep deprivation. In M. Kryger,
T. Roth & W. Dement (Eds.), Principles and practices of sleep medicine (5 ed., pp.
67-75). Philadelphia: Elsvier.
Doran, S. M., Van Dongen, H. P. A., & Dinges, D. F. (2001). Sustained attention
performance during sleep deprivation: evidence of state instability. Archives
Italiennes de Biologie, 139(3), 253-267.
Dorrian, J., Rogers, N. L., & Dinges, D. F. (2005). Psychomotor vigilance performance:
Neurocognitive assay sensitive to sleep loss (Doctoral dissertation, Marcel Dekker).
Dunn, B. D., Dalgleish, T., & Lawrence, A. D. (2006). The somatic marker hypothesis: A
critical evaluation. Neuroscience & Biobehavioral Reviews, 30(2), 239-271.
Durmer, J. S., & Dinges, D. F. (2005). Neurocognitive consequences of sleep
deprivation. Paper presented at the Seminars in Neurology, 25(1) 117-129.
Drummond, S. P., Brown, G. G., Gillin, J. C., Stricker, J. L., Wong, E. C., & Buxton, R.
B. (2000). Altered brain response to verbal learning following sleep
deprivation. Nature, 403(6770), 655-657.
Drummond, S. P., Brown, G. G., Stricker, J. L., Buxton, R. B., Wong, E. C., & Gillin, J.
C. (1999). Sleep deprivation-induced reduction in cortical functional response to
serial subtraction. Neuroreport, 10(18), 3745-3748.
33 Drummond, S., & McKenna, B. (2009). Sleep deprivation/restriction: Human studies. In
Amlaner, C., & Fuller, P. (2nd ed.). Basics of sleep guide (249-256). Westchester, IL:
Sleep Research Society.
Franken, I. H., Van Strien, J. W., Nijs, I., & Muris, P. (2008). Impulsivity is associated
with behavioral decision-making deficits. Psychiatry Research,158(2), 155-163.
Giorgetta, C., Grecucci, A., Zuanon, S., Perini, L., Balestrieri, M., Bonini, N., ... &
Brambilla, P. (2012). Reduced risk-taking behavior as a trait feature of
anxiety. Emotion, 12(6), 1373.
Green, L., Myerson, J., Holt, D. D., Slevin, J. R., & Estle, S. J. (2004). Discounting of
delayed food rewards in pigeons and rats: Is there a magnitude effect?. Journal of
the experimental analysis of behavior, 81(1), 39-50.
Harrison, Y., & Horne, J. (1998). Sleep loss affects risk-taking. Journal of Sleep
Research, 7(2), 113.
Harrison, Y., & Horne, J. A. (2000). The impact of sleep deprivation on decision making:
A review. Journal of Experimental Psychology: Applied, 6(3), 236.
Jones, C. B., Dorrian, J., Jay, S. M., Lamond, N., Ferguson, S., & Dawson, D. (2006).
Self-awareness of impairment and the decision to drive after an extended period of
wakefulness. Chronobiology International, 23, 1253–1263.
doi:10.1080/07420520601083391
34 Kjellberg, A. (1975). Effects of sleep deprivation on performance of a problem solving
task. Psychological reports, 37, 479-485.
Killgore, W. D. (2007). Effects of sleep deprivation and morningness-eveningness traits
on risk-taking. Psychological reports, 100(2), 613-626.
Killgore, W. D., Balkin, T. J., & Wesensten, N. J. (2006). Impaired decision making
following 49 h of sleep deprivation. Journal of Sleep Research, 15(1), 7-13.
Killgore, W. D., Grugle, N. L., & Balkin, T. J. (2009). Gambling when sleep deprived:
don't bet on stimulants. Chronobiology International, 29(1), 43-54.
Killgore, W. D., Grugle, N. L., Killgore, D. B., Leavitt, B. P., Watlington, G. I., McNair,
S., & Balkin, T. J. (2008). Restoration of risk-propensity during sleep deprivation:
caffeine, dextroamphetamine, and modafinil. Aviation, Space, and Environmental
Medicine, 79(9), 867-874.
Killgore, W. D., Kamimori, G. H., & Balkin, T. J. (2011). Caffeine protects against
increased risk-­‐‑taking propensity during severe sleep deprivation. Journal of Sleep
Research, 20(3), 395-403.
Killgore, W. D., Kelley, A., & Balkin, T. J. (2010). So you think you're bulletproof:
Development and validation of the Invincibility Belief Index (IBI). Military
Medicine, 175(7), 499-508.
35 Killgore, W. D., Lipizzi, E. L., Kamimori, G. H., & Balkin, T. J. (2007). Caffeine effects
on risky decision making after 75 hours of sleep deprivation. Aviation, space, and
Environmental Medicine, 78(10), 957-962.
Laviola, G., Macrı̀, S., Morley-Fletcher, S., & Adriani, W. (2003). Risk-taking behavior
in adolescent mice: psychobiological determinants and early epigenetic
influence. Neuroscience & Biobehavioral Reviews, 27(1), 19-31.
Leigh, B. C. (1999). Peril, chance, adventure: concepts of risk, alcohol use and risky
behavior in young adults. Addiction, 94(3), 371-383.
Lister, R. G. (1987). The use of a plus-maze to measure anxiety in the
mouse. Psychopharmacology, 92(2), 180-185.
Macrı̀, S., Adriani, W., Chiarotti, F., & Laviola, G. (2002). Risk taking during
exploration of a plus-maze is greater in adolescent than in juvenile or adult
mice. Animal Behaviour, 64(4), 541-546.
Manes, F., Sahakian, B., Clark, L., Rogers, R., Antoun, N., Aitken, M., & Robbins, T.
(2002). Decision‐making processes following damage to the prefrontal
cortex. Brain, 125(3), 624-639.
Mckenna, B. S., Dickinson, D. L., Orff, H. J., & Drummond, S. (2007). The effects of
one night of sleep deprivation on known-risk and ambiguous-risk decisions. Journal
of Sleep Research, 16(3), 245-252.
36 Menz, M. M., Buchel, C., & Peters, J. (2012). Sleep deprivation is associated with
attenuated parametric valuation and control signals in the midbrain during valuebased decision making. The Journal of Neuroscience, 32(20), 6937–6946.
doi:10.1523/jneurosci.3553-11.2012
Misslin, R., Belzung, C., & Vogel, E. (1989). Behavioural validation of a light/dark
choice procedure for testing anti-anxiety agents. Behavioural processes, 18(1), 119132.
Rechtschaffen, A., & Bergmann, B. M. (1995). Sleep deprivation in the rat by the diskover-water method. Behavioural brain research, 69(1), 55-63.
Rivalan, M., Ahmed, S. H., & Dellu-Hagedorn, F. (2009). Risk-prone individuals prefer
the wrong options on a rat version of the Iowa Gambling Task. Biological
psychiatry, 66(8), 743-749.
Reynolds, B., & Schiffbauer, R. (2004). Measuring state changes in human delay
discounting: An experiential discounting task. Behavioural processes, 67(3), 343356.
Roman, V., Walstra, I., Luiten, P. G., & Meerlo, P. (2005). Too little sleep gradually
desensitizes the serotonin 1A receptor system. Sleep-New York then Westchester,
28(12), 1505.
37 Pellow, S., Chopin, P., File, S. E., & Briley, M. (1985). Validation of open: Closed arm
entries in an elevated plus-maze as a measure of anxiety in the rat. Journal of
Neuroscience Methods, 14(3), 149-167.
Petiau, C., Harrison, Y., Delfiore, G., Degueldre, C., Luxen, A., Franck, G., ... & Maquet,
P. (1998). Modification of fronto-temporal connectivity during a verb generation
task after a 30 hour total sleep deprivation. A PET study. Journal of Sleep
Research, 7(2), 208.
Simmons, R. K., Howard, J. L., Simpson, D. N., Akil, H., & Clinton, S. M. (2012). DNA
methylation in the developing hippocampus and amygdala of anxiety-prone versus
risk-taking rats. Developmental Neuroscience, 34(1), 58-67.
Simon, N. W., Gilbert, R. J., Mayse, J. D., Bizon, J. L., & Setlow, B. (2009). Balancing
risk and reward: a rat model of risky decision making.
Neuropsychopharmacology, 34(10), 2208-2217.
Tversky, A., & Fox, C. R. (1995). Weighing risk and uncertainty. Psychological Review,
102(2), 269.
Van Dongen, H. P., Maislin, G., Mullington, J. M., & Dinges, D. F. (2003). The
cumulative cost of additional wakefulness: Dose-response effects on
neurobehavioral functions and sleep physiology from chronic sleep restriction and
total sleep deprivation. Sleep-New York then Westchester, 26(2), 117-129.
38 Van Dongen, H., Rogers, N. L., & Dinges, D. F. (2003). Sleep debt: Theoretical and
empirical issues. Sleep and Biological Rhythms, 1(1), 5-13.
Venkatraman, V., Chuah, Y. L., Huettel, S. A., & Chee, M. W. (2007). Sleep deprivation
elevates expectation of gains and attenuates response to losses following risky
decisions. Sleep-New York then Westchester, 30(5), 603.
Venkatraman, V., Huettel, S. A., Chuah, L. Y. M., Payne, J. W., & Chee, M. W. L.
(2011). Sleep deprivation biases the neural mechanisms underlying economic
preferences. The Journal of Neuroscience, 31, 3712–3718. doi:10.1523/
jneurosci.4407-10.2011
Webb, W., & Agnew, H. (1975). Are we chronically sleep deprived?. Bulletin of the
Psychonomic Society, 6(1), 47-48.
Weber, E. U., Shafir, S., & Blais, A. R. (2004). Predicting risk sensitivity in humans and
lower animals: risk as variance or coefficient of variation. Psychological
Review, 111(2), 430.
39 Table 1
Means and Standard Deviations of Risk-Taking Behavior
Condition
M
SD
Baseline
Sleep Dep 1
Sleep Dep 2
Sleep Dep 3
Sleep Dep 4
Sleep Dep 5
Recovery 1
Recovery 2
74.5
77.7
79.8
89.1
58.4
63
77.4
77.5
34.6
27.2
54.19
65.45
12.1
20.97
43.38
61.52
40 Table 2 Displays the most favored hole each day of the experimental procedure. F-­‐1 shows a potential 50% bias towards hole 3. Day Hole Type Code 1 2 3 4 5 6 7 8 ADV 1/2 ADV 1/4 ADV 1/2 ADV 1/4 DIS 1/2 DIS 1/4 ADV 1/2 ADV 1.2 B A D C A D B C Hole Number 3 2 2 3 3 3 3 4 41 Table 3 Displays the most favored hole each day of the experimental procedure. F-­‐2 shows a potential 75% bias towards hole 3. Day 1 2 3 4 5 6 7 8 Hole Type ADV 1/4 ADV 1/4 DIS 1/4 ADV 1/2 ADV 1/2 & 1/4 DIS 1/4 ADV 1/4 DIS 1/2 Code C A D B B D C A Hole Number 3 2 3 3 3 & 4 3 3 3 42 Table 4 Displays the most favored hole each day of the experimental procedure. F-­‐3 shows no potential biases towards any hole. Day 1 2 3 4 5 6 7 8 Hole Type DIS 1/4 DIS 1/2 DIS 1/2 DIS 1/4 DIS 1/4 DIS 1/4 ADV 1/2 ADV 1/4 Code D C A B B C A D Hole Number 3 2 3 2 2 1 1 1 43 Table 5 Displays the most favored hole each day of the experimental procedure. F-­‐4 shows a potential 75% bias towards hole 1. F-­‐4 Day 1 2 3 4 5 6 7 8 Hole Type ADV 1/4 ADV 1/2 ADV 1/2 & 1/4 DIS 1/4 DIS 1/2 ADV 1/4 DIS 1/4 ADV 1/2 ADV 1/4 Code D A C B D C A B Hole Number 1 1 4, 3, & 1 1 1 1 1 4 44 Table 6 Displays the most favored hole each day of the experimental procedure. F-­‐5 shows a potential 63% bias towards hole 3. Day 1 2 3 4 5 6 7 8 F-­‐5 Code D B C A A B D C Hole Type ADV 1/4 ADV 1/2 ADV 1/4 DIS 1/2 DIS 1/2 ADV 1/2 ADV 1/4 DIS 1/4 Hole Number 1 3 3 3 3 3 1 1 45 Table 7 Displays the most favored hole each day of the experimental procedure. Z-­‐1 shows a potential 63% bias towards hole 3. Day 1 2 3 4 5 6 7 8 Z-­‐1 Code B D A C D B A C Hole Type DIS 1/4 ADV 1/2 DIS 1/4 ADV 1/4 DIS 1/4 ADV 1/2 DIS 1/2 ADV 1/4 Hole Number 2 2 4 3 3 3 3 3 46 Table 8 Displays the most favored hole each day of the experimental procedure. Z-­‐2 shows a potential 88% bias towards hole 4. Day 1 2 3 4 5 6 7 8 Z-­‐2 Code D A B C A D C B Hole Type ADV 1/4 DIS 1/4 ADV 1/4 ADV 1/2 DIS 1/4 DIS 1/2 ADV 1/2 ADV 1/4 Hole Number 1 4 4 4 4 4 4 4 47 Table 9 Displays the most favored hole each day of the experimental procedure. Z-­‐3 shows no potential biases towards any holes. Day 1 2 3 4 5 6 7 8 Z-­‐3 Code C D B A A B C D Hole Type ADV 1/4 DIS 1/2 ADV 1/2 ADV 1/4 ADV 1/4 ADV 1/2 DIS 1/2 DIS 1/4 Hole Number 3 4 3 2 2 3 2 3 48 Table 10 Displays the most favored hole each day of the experimental procedure. Z-­‐4 shows a 50% bias towards hole 4. Day 1 2 3 4 5 6 7 8 Hole Type DIS 1/2 & 1/4 ADV 1/2 DIS 1/2 ADV 1/4 DIS 1/4 ADV 1/2 DIS 1/4 DIS 1/2 Code A C D B B A C D Hole Number 3 & 4 4 4 4 2 1 1 4 49 Table 11 Displays the most favored hole each day of the experimental procedure. Z-­‐5 shows a potential 75% bias towards hole 2. Day 1 2 3 4 5 6 7 8 Hole Type DIS 1/4 ADV 1/2 ADV 1/4 DIS 1/2 DIS 1/2 DIS 1/4 ADV 1/4 DIS 1/2 Z-­‐5 Code B D A C D B A C Hole Number 2 2 2 2 4 4 2 2 Figure 1. Represents the four variations of the RGT that were coded into MED-PC.
50 51 Figure 2. Displays mean number of pellets obtained across the duration of the experiment
52 Figure 3. Displays the performance of each rat individually. The y-axis displays the total
number of pellets obtained. The axis displays the day of the experiment. Note that the y-axis
of the graph for rat Z2 is much larger due to his unique performance
53 Figure 4. Displays the performance of each rat individually. The y-axis displays the percentage of
choices made. The solid line with filled triangle markers represents advantageous choices. The
dashed line with open triangle markers represents disadvantageous choices. The x-axis represents day
of experiment.