An Investigation into Predictors of Competitive Success
in a Large-Scale Group Foraging Task
By
Copyright 2016
Bryan Tomes Yanagita
Submitted to the graduate degree program in the Applied Behavioral Science Department
and the Graduate Faculty of the University of Kansas in partial fulfillment of the
requirements for the degree of Master of Arts.
__________________________
Chairperson Derek D. Reed
__________________________
David P. Jarmolowicz
__________________________
Pamela L. Neidert
Date Defended:
Wednesday, June 15th, 2016
ii
The Thesis Committee for Bryan Tomes Yanagita
certifies that this is the approved version of the following thesis:
An Investigation into Predictors of Competitive Success
in a Large-Scale Group Foraging Task
__________________________
Chairperson Derek D. Reed
Date approved: Wednesday, June 15th, 2016
iii
Abstract
Ideal Free Distribution Theory (IFD) (Fretwell & Lucas, 1970) suggests that the allocation of
organisms to two or more resource sites is a function of the available resources at each site.
Experimental arrangements involving human participants consist of a series of trials in which
participants choose between two resource sites with differing resource values. Past research has
investigated the relationship between delay discounting and performance in an IFD task, noting
correlations between higher discounting and poorer individual performance. The current study
sought to investigate the predictive capabilities of various psychological assessments, including
an individual and group-context delay discounting task. Results demonstrate that the allocation
of participants conformed to the IFD theory. Additionally, results suggest that, while controlling
for all other assessments, the individual delay discounting assessment, competitive index score,
and proportion of trials switched significantly predicted performance.
Keywords: delay discounting, group context delay discounting, ideal free distribution
theory, competitive success
iv
Acknowledgements
I would like to thank Drs. Derek Reed, Dave Jarmolowicz, and Pam Neidert for
dedicating their time, expertise, and support while serving on my Master’s committee.
I cannot express how thankful I am to earn my Master’s degree from KU, and I owe this
opportunity to Derek for seeing my potential and being my mentor throughout the process. I will
always look fondly on my adventure at KU, and am happy that I was able to work with you. You
have provided me with a life and career path, and I will confidently leave KU knowing that I was
able to learn from you.
I have appreciated Pam’s generosity, mentorship, and supervision, as she provided me
with the opportunity to gain applied experience. Although I am not your graduate student, you
made time for me and provided guidance during my time at Sunnyside. My experience in the
CDC has been both challenging and exciting, and I am thankful for what you have taught me.
I would like to thank Dave for his support and mentorship in classes, and for providing
feedback on projects along the way. I have appreciated your insight, approachability, and
willingness to assist me with any and all questions.
I would like to thank Amel, Brent, Gideon, Alec, Kelsey, and Andi for helping me
transition to graduate school, supporting my research and GTA positions, and most importantly,
your friendship. You have all been a large part of my graduate experience, and I will always
appreciate what you have done for me.
Lastly, I am forever grateful for the love and support of my family. Thank you for
everything.
v
Table of Contents
Introduction 1
Methods 13
Results 19
Discussion 22
References 28
Tables 33
Figures 37
Appendices 44
vi
List of Illustrative Materials
Table Captions
Table 1. Breakdown of extra credit earned for the in-class group foraging
task.
33
Table 2.
Table 3.
Table 4.
Point availability for each resource site across all six conditions.
Correlation matrix for the online survey assessments and
performance measures from the in-class group foraging task.
Summary of multiple regression analysis for predicting
individual competitive success (i.e., total points earned during
the group foraging task).
34
35
36
Figure Captions
Figure 1. Natural log k-values across small, medium, and large magnitude
hypothetical monetary rewards for both the group context and
individual delay discounting assessments.
37
Figure 2. Distribution of k-values for both group context and individual
delay discounting assessments.
38
Figure 3. The number of people selecting resource site “A” across all 15
trials for all six conditions. The dashed line on each graph
indicates the predicted number of individuals selecting resource
site “A” based on the IFD equation 4.
39
Figure 4. The proportion of individuals as a function of the proportion of
available points. The left panel shows the average of the first
five trials for each condition, the middle panel shows the
average of the middle five trials for each condition, and the right
panel shows the average of the last five trials for each condition.
40
Figure 5. The observed number of individuals that selected resource site
“A” plotted as a function of the predicted number of individuals
that selected resource site “A” based on the IFD equation 4.
Data points show the average of the last five trials for each
condition.
41
Figure 6. The proportion of trials switched as a function of the log ratio of
points earned on the previous trial.
42
vii
Figure 7. Data collection method comparison between iClicker total
switches plotted as a function of paper-pencil switches for each
participant.
43
viii
Appendix Index
A) Informed consent form 44
B) Online survey flow 45
C) Example individual delay discounting assessment 46
D) Example group context delay discounting assessment 47
E) Social Value Orientation assessment 48
F) Competitiveness index 49
G) iClicker 2 50
H) Participant data collection sheet for the in-class group foraging task 51
I) Group foraging task example trial 52
J) Condition change slide 55
K) Example of live feedback during group foraging task trial 56
1
An Investigation into Predictors of Competitive Success
in a Large-Scale Group Foraging Task
As a quantitative model of choice (Herrnstein, 1961), the matching law states that the
relative proportion of responses towards an alternative is equal to (i.e., matches) the relative
proportion of reinforcement available for that alternative. Typically, matching is assessed by
presenting two concurrently available response alternatives, each corresponding to a different
variable interval schedule of reinforcement. All behaving is choice, in that at any instance an
organism may continue to engage in a response or begin engaging in a number of different
responses (Herrnstein, 1970). Put simply, the paradigm is representative of the choices that
organisms continuously encounter. The matching law predicts that the relative rate of
reinforcement for a given choice alternative dictates the response allocation of an organism,
where
𝑅1
𝑅2
=
𝐵1
𝐵2
Equation 1
R1 and R2 refer to the relative rate of reinforcement, and B1 and B2 refer to the relative response
rates.
Baum (1974) provided a generalized equation of the matching law that accounts for the
sensitivity to reinforcement of the organism and whether there are biases in the organism’s
responding. Quantitatively, the generalized matching equation is expressed as
log(
𝐵1
𝐵2
) = 𝑠 log(
𝑅1
𝑅2
) + log(𝑏) Equation 2
where B refers to the response alternatives and R refers to the available reinforcement for each
alternative. Sensitivity, or s, refers to the degree to which an organism allocates its behavior to a
reinforcing choice alternative that is accounted for by reinforcement. Perfect matching, as
denoted by s=1.00, states that an organism’s allocation of behavior perfectly matches the
2
available reinforcement for each reinforcing alternative. For instance, in a concurrent choice
arrangement, perfect matching would suggest that twice the behavior would be allocated to the
choice alternative that has twice the available reinforcement. Undermatching, as denoted by
s<1.00, suggests that less behavior would be allocated to the response alternative than is
predicted by the matching law. Conversely, overmatching, as denoted by s>1.00, suggests that
more behavior would be allocated to the response alternative than predicted by the matching law.
Lastly, bias (b) refers to any preferences for a response alternative that is unaccounted for by
reinforcement. A positive bias refers to a preference towards the response alternative associated
with the numerator of the response portion of equation 2 that is unaccounted by reinforcement,
while a negative bias indicates a preference towards the denominator that is unaccounted by
reinforcement.
The matching law has been applied to understand the behavioral allocation of a diverse
range of behaviors including sports (e.g., Alferink, Critchfield, Hitt, & Higgins, 2009; Reed,
Critchfield, & Martens, 2006), academics (e.g., Reed & Martens, 2008), and social dynamics
(e.g., Borrero, Crisolo, Tu, Rieland, Ross, Francisco, & Yamamoto, 2007). Although the
matching law has been helpful in understanding the distribution of an individual’s behavior in a
variety of contexts, recently, ecological biology has been applied to understand the distribution
of organisms operating within a group, known as the Ideal Free Distribution (IFD) theory.
The IFD theory (Fretwell & Lucas, 1970) is an extension of the matching law that
accounts for the behavior of organisms operating in a group setting, and suggests that the
allocation of organisms to two or more resource sites is a function of the available resources at
those sites. There are distinct similarities between both the qualitative and quantitative aspects of
the matching law and the IFD theory. First, the dependent variable in a matching law paradigm is
3
an organism’s behavior, while the dependent variable in an IFD paradigm is the number of
organisms. Instead of manipulating the relative rate of reinforcement (i.e., matching law) for the
two different response alternatives, the typical IFD paradigm involves the manipulation of the
relative amount of resources available at the two different resource sites. Lastly, the quantitative
models are represented by the same equation; instead of the proportion of behavior as a function
of reinforcement, it is the proportion of organisms as a function of the available resources.
Quantitatively, its simplest form states that the proportion of individuals at a resource site
is equal to the proportion of available resources (Baum & Kraft, 1998):
𝑁1
𝑁2
=
𝐴1
𝐴2
Equation 3
where N refers to the number of individuals at a resource site, and A refers to the available
resources. Therefore, it is predicted that if there are twice as many resources available at site one
than site two, there will be twice as many organisms at site one than site two. This result of
perfectly equal proportion of individuals to resources is referred to as habitat matching (Pulliam
& Caraco, 1984). As with the matching law, however, there are often deviations from habitat
matching. That is, the perfect distribution of organisms to available resources rarely occurs,
resulting in a need for a generalized version of the equation to account for the deviations from
the IFD theory. In 1987, Fagan modified the generalized matching equation to account for
groups of organisms (cf., Baum, 1974; equation 2):
log(
𝑁1
𝑁2
) = 𝑠 log(
𝐴1
𝐴2
) + log(𝑏) Equation 4
In the generalized ideal free distribution model, N refers to the number of individuals at a
resource site. A refers to the amount of resources available at each site. The value, s, refers to the
sensitivity to differences in the amount of resources available in each resource site. Lastly, b
4
refers to the group bias that influences the relative allocation of organisms to each resource site.
A potential bias, for instance, could be the presence of a predator at a resource site.
Perfect matching, or habitat matching, is represented by s=1. Undermatching (i.e., s <
1.00) refers to a situation in which the number of individuals at a resource site is less extreme
than what is predicted based on the amount of available resources. . That is, there are less
individuals at the resource site than predicted by equation one. Conversely, when there are more
organisms at a resource site than predicted by equation one, the pattern of responding is referred
to as overmatching (i.e., s > 1.00). Undermatching is the most common finding in the literature,
with an average s value of 0.70 across a wide range of species (Kennedy & Gray, 1993). These
data suggest organisms are typically less sensitive to resource allocations than is predicted by the
IFD theory.
Initial investigations of the IFD theory originated in ecological and biological research
aimed at assessing the relative distribution of non-human animals towards two or more resource
sites in contrived and natural environments (e.g., Milinksi, 1984; Harper, 1982; Abrahams,
1989), concluding that IFD theory predicted the distribution of organisms. As one example,
Harper (1982) manipulated the availability of resources (i.e., pieces of white bread) for a group
of mallard ducks in their natural habitat (i.e., a pond). Both the rate at which resources were
presented and the amount of resources presented (i.e., weight of bread) were manipulated,
finding that the distribution of ducks closely approximated the predictions based on the IFD
theory.
Although IFD has been experimentally studied with non-humans, there have been
relatively few investigations into the IFD of humans (e.g., Kraft & Baum, 2001; Kraft, Baum, &
Burge, 2002; Sokolowski, Tonneau, & Baque, 1999; Madden et al., 2002; Critchfield &
5
Atteberry, 2003). Despite the lack of research, the initial data provide support of the IFD theory
for humans in both experimental and natural settings.
In a typical human operant IFD arrangement, participants are instructed to make choices
between two resource sites differing in the amount or rate of rewards provided. There are
numerous experimental manipulations which result in different levels of conformity to the
predicted distribution of individuals, including the ability to switch resource sites after seeing
how many individuals are in each site, the ability to communicate (i.e, verbal behavior), and
whether it is a discrete trial or free operant arrangement. Participants are told that they are to
compete for limited resources (i.e., hypothetical points that are typically exchangeable for
monetary rewards or extra credit) by making a series of choices between one of two resource
sites, in which they evenly split the resources with however many individuals are at each
resource site. Each choice is a trial, and there are typically 15-20 trials per condition. Each
condition is associated with a constant rate or amount of rewards for each site, with most
experiments employing four to six conditions (including a direct replication condition).
In 1999, Sokolowski et al. conducted the first experimental investigation of the
distribution of human participants between two resource sites. Each resource site had a specific
number of token rewards that were available, and these rewards were evenly split between the
number of individuals that were at the resource site. The participant that had the most tokens was
awarded a cash prize for each condition. The ratio of available tokens for each resource site was
manipulated between each condition. The results of the experiment demonstrated conformity to
the ideal free distribution theory, with undermatching patterns of distribution being observed
(s=0.67).
6
Madden et al. (2002) conducted a series of three experiments investigating various
procedures used to assess the ideal free distribution theory. In all three experiments, 12
undergraduate students served as participants. Participants were instructed to make a series of
choices between two resource sites of varying undisclosed point values that would be evenly
split between each other individual in that resource site, and were told that they were to try and
earn as many points as they could. In the first experiment, a systematic replication of Kraft and
Baum (2001), participants were instructed to make a choice between red or blue (i.e., each
resource site), and could then freely switch between resource sites. The second experiment
followed the same procedure as experiment one, but participants were unable to switch after their
initial choice and they were instructed not to talk. This was done to ensure that participants made
decisions based on their own judgements, rather than on what others said. In the third
experiment, rewards were delivered on concurrent variable time schedules. In this free operant
arrangement, participants were able to freely walk between resource sites. The first two
experiments consisted of 5 conditions, each consisting of 20 trials. In the third experiment, each
of the five conditions consisted of at least 20 minutes of responding and when the behavior of the
group was deemed stable.
The results from experiment one of Madden et al. (2002) show undermatching for both
initial and final choices, with the final choice distribution of responses (s=0.92) much closer to
ideal matching than initial selections (s=0.40), and virtually no bias for either the initial or final
selection. Unsurprisingly, these data suggest initial selections were less sensitive to the available
resources than the final selections, as participants were able to adapt their choices following
initial selections to adapt to the distribution of individuals at a given resource site.
7
The Madden et al. (2002) manuscript noted that participants communicated—thus, they
may have strategized—in the first experiment. Therefore, experiment two investigated the IFD
theory without being able to communicate or freely switch between resource sites following their
initial selections. Results from experiment two demonstrated undermatching (s=0.82) taken from
the final six trials of each condition (i.e., stability criteria). Interestingly, as a novel analysis, the
probability of switching as a function of the points earned on a previous trial was plotted for each
condition. Analysis indicated participants were a) more likely to switch after low point earnings
(i.e., 6-8 points), b) would likely stay at the resource site for medium point earnings (i.e., 10-18
points), and c) would be more likely to switch after big-win point earnings (i.e., 20+ points).
Madden et al. (2002) data suggest the orderly relation between group allocation and
available resources in a discrete trial format. However, the authors noted that a free operant
format, in which organisms are able to freely travel between each resource site, is more
analogous to the natural environment. Therefore, experiment three employed a free operant
paradigm in which two response alternatives (i.e., two sides of a room) were programmed to
deliver rewards on concurrently available variable time schedules of reinforcement. Results from
experiment three demonstrated a correspondence between the predicted allocation of individuals
(i.e., predicted by the IFD theory) and the obtained number of participants at each resource site.
These data lend support to the predictive utility of the IFD theory. Additionally, results of the
final 5-minutes of each condition (i.e., stability criteria) demonstrated undermatching (s=0.71).
During group foraging tasks, there is variability between participants in regards to
competitive success. Critchfield & Atteberry (2003) suggested that an individual’s sensitivity to
delays may account for the variability in these tasks. Conceptually, individuals that are relatively
sensitive to delays (i.e., self-controlled) may be more likely to sample the rich resource site,
8
switch less often, and earn more overall points. Put simply, individuals that respond in more self-
controlled patterns might be expected to demonstrate more competitive success than individuals
that are relatively insensitive to delays (i.e., impulsive). To assess sensitivity to delays,
Critchfield & Atteberry (2003) employed a delay discounting assessment in conjunction with a
group foraging task.
Delay discounting refers to the decline in value of a reward based on the delay to which
that reward is received (Odum, 2011). In a typical human operant delay discounting procedure,
the participant is provided with a series of dichotomous choices between a smaller hypothetical
rewards delivered immediately versus larger hypothetical rewards delivered after some delay.
Preference for the smaller reward across trials is representative of impulsive choice, while
preference towards the larger yet delayed reward across trials is representative of self-controlled
choice. Steep rates of discounting (i.e., a pattern of responding towards smaller-immediately
available rewards) has been associated with maladaptive behaviors including obesity, gambling,
and substance abuse (Epstein, Salvy, Carr, Dearing, & Bickel, 2010; Reynolds, 2006).
Critchfield and Atteberry (2003) investigated the relation between delay discounting and
performance in an IFD activity with college undergraduates in a human operant arrangement.
The rationale for their study was based upon past literature that noted the relationship between
discounting and sensitivity to reinforcement (Critchfield & Atteberry, 2003), as well as
differences in the rates of discounting that were correlated with prevalence or intensity of
impulsive choices (Critchfield & Kollins, 2001). Participants completed a six-page paper-pencil
delay discounting assessment, in which participants made a series of dichotomous choices
between $1000 received after a delay, or a smaller amount available now. The initial choice was
between $1000 delivered after a delay and $500 delivered immediately. If the participant
9
selected the larger-delayed reward, the smaller amount decreased on the subsequent trial.
Conversely, if the participant selected the smaller amount, the smaller amount increased on the
subsequent trail. The delay was manipulated across each page (i.e., condition), and consisted of a
1 month, 6 month, 1 year, 3 years, 5 years, and 10 year delays. All rewards and delays were
hypothetical.
Participants in the Critchfield and Atteberry (2003) study were unsystematically assigned
to one of 17 groups, each consisting of 9-12 individuals, to participate in the group foraging task
(i.e,. IFD paradigm). Each participant was provided with both a red and a blue card used for
responding during the task, each colored card corresponding with a resource site. Participants
were instructed to raise either the blue or red card to make a choice. Participants were then told
that they would be responding to earn hypothetical points exchangeable for extra credit, with the
points being split evenly between the other participants that were at the same resource site. For
instance, if there were 100 points available for selecting the red resource site, and if 10
individuals selected red, they would each split the 100 points for a total of 10 points each. The
participants were not told how many points each resource site was worth. Each group completed
4 conditions, with each condition containing 16 trials.
Critchfield and Atteberry (2003) hypothesized that participants demonstrating steeper
rates of discounting (i.e., a relative preference towards smaller-immediately available rewards)
would be (a) more likely to sample the lean resource site, (b) less likely to switch resource sites,
and (c) earn less overall points (i.e., perform worse in the group foraging task).
Critchfield and Atteberry’s (2003) results demonstrated undermatching across the
majority of groups, with group slopes ranging from s=0.36-1.30 (median s=0.72). Switch point
probability as a function of relative point gain was consistent with the results of Madden et al.
10
(2002), in that individuals were likely to switch with low relative point earnings, likely to stay
with medium relative point earnings, and there was a slight increase in switch probability at
relatively high point earnings. Additionally, results suggest a slight difference between the most
and least impulsive individuals, as determined by temporal discounting rates, with the most
impulsive individuals having less pronounced (i.e., less extreme) switch probabilities. Although
there were substantially different results for individuals in relation to competitive success (i.e.,
rich patch choices, number of switches, frequency of better outcomes, and total point earnings), a
correlational analysis suggested that temporal discounting accounted for a significant proportion
of the variance in the data. There were significantly negative correlations between discounting
rates and log rich/lean choices, log better/worse choices, and total points earned, while there
were significant correlations between the number of switches and discounting rates in both
studies. Simply put, steeper discounters tended to choose the lean resource site more often, select
the worse option more often, switch more often, and earn fewer overall points.
The IFD theory provides an experimental and conceptual framework that has potential
applied relevance. The allocation of individuals as a function of the available resources has been
observed in naturalistic environments, suggesting that environments may be contrived to
manipulate the relative distribution of individuals. For instance, in one demonstration of the
distribution of humans in an undisturbed natural environment, Disma, Sokolowski, and Tonneau
(2011) observed the distribution of children in Istanbul, Turkey, in relation to the available
resources (i.e., high traffic areas in which the children could sell water bottles). They found that
the IFD theory was predictive of the proportion of children towards the available resource sites.
Put simply, the proportion of children at each intersection matched the proportion of cars at those
intersections. In this sense, the manipulation of the available resource sites (e.g., location of site,
11
rate of available resources, the amount of available resources) may result in changes in the
proportion of individuals at each resource site. The manipulation of these variables could
influence issues of societal importance by changing the distribution of individuals to areas that
are more socially appropriate. For instance, social cohesion (i.e., cooperation between groups of
people from diverse backgrounds) could be enhanced by increasing the available resources in
areas that supports these interactions (e.g., community centers, parks).
Past research has looked at social engagement (Borrero, Crisolo, Tu, Rieland, Ross,
Francisco, & Yamamoto, 2007; Conger & Killeen, 1974; McDowell & Caron, 2010; McDowell
& Caron, 2010), but these analyses have only been useful for engagement between two people.
Despite this, it is important to look at group cohesion. Psychological assessments, including
social discounting measures (Charlton, Yi, Porter, Carter, Bickel, & Rachlin, 2013) and Social
Value Orientation (Van Lange, Otten, De Bruin, Joireman, 1997), have been used to investigate
the relative variables that contribute to group cohesion. These assessments are easy to administer
in an online survey format, and provide an assessment of an individual’s preference for outcomes
that are face valid in comparison to a group foraging task (e.g., competition, social dynamics).
Similar to other behavioral indicators of cooperation (see Hursh and Roma, 2013), IFD can
provide a quantitative application to group cohesion.
The purpose of the current study is to systematically extend the previous research by
Critchfield and Atteberry (2003) by assessing the predictive utility of three different empirically
validated assessments (i.e., delay discounting, social value orientation, competitiveness index) in
comparison with performance on a group foraging task. Additionally, in an effort to improve the
novel discounting assessment in Critchfield and Atteberry (2003), the current study employed
empirically validated delay discounting assessments. Furthermore, Critchfield and Atteberry
12
(2003) only assessed individual discounting, while the current study investigated both individual
(Kirby, Petry, & Bickel, 1999) and group-context delay discounting (Charlton et al., 2013).
Lastly, a recent experimental investigation (i.e., Sokolowski, Tonneau, & Cordevant, 2015)
demonstrated the use of a video game-based data collection tool, finding orderly relations
between available resources and the distribution of individuals responding. Despite this
demonstration, the current study provided potential improvements by employing a novel data
collection tool that allows for automated data collection, live feedback for participants, and can
support 150+ participants in a single experimental session.
13
Methods
Online Survey
Participants
160 undergraduate students (138 women, 20 men, 1 prefer not to disclose, 1 gender non-
conforming) enrolled in an introductory behavioral science course at a large Midwestern
university served as participants, ranging in age from 19 to 45 years old (M= 21.05, SD= 3.29).
Participant’s received 0.5% extra credit towards their overall grade in the course for the
completion of the online survey. The research procedures were approved by the University of
Kansas’ Human Subjects Committee (HSCL #00002182).
Materials and Procedure
Participants were notified via email and an in-class announcement about the availability
of the online survey. The survey, generated via Qualtrics online survey database, was available
for 11 days (March 24th to April 3rd, 2015) and required approximately 15 minutes to complete.
The online survey consisted of demographics, an individual (Kirby et al., 1999) and group
context delay discounting assessment (Charlton et al., 2013), a Social Value Orientation
assessment (Van Lange et al., 1997), and the Revised Competitiveness Index (Houston, Harris,
McIntire, & Francis, 2002).
The survey began with an informed consent page (see appendix A) and demographic
questions (e.g., age, gender). Following the completion of the demographic questions,
participants were randomly and evenly assigned to both the individual or group context delay
discounting assessment. After completing each of these discounting assessments, participants
were randomly assigned to the Social Value Orientation and the Competitiveness Index that were
evenly counterbalanced to mitigate order effects. Lastly, participants were required to type in
14
their unique iClicker ID number, allowing the experimenter to match their online survey
responses with their performance in an in-class group foraging task.
Multiple-choice Questionnaire:
The individual delay discounting assessment was adapted from the 27-item Kirby delay
discounting assessment (Kirby et al., 1999), and consisted of 27 questions regarding decisions
between smaller-immediate hypothetical monetary amounts and larger hypothetical monetary
amounts delivered after a specified delay that benefits the responding individual (see appendix
C). The following assumption was provided prior to the administration of the individual delay
discounting:
“For each of the next 27 choices, please mark which hypothetical reward you would
prefer: the smaller reward today, or the larger reward in the specified number of days.
While you will not actually receive the rewards, pretend you will actually be receiving
the amount you indicate and answer honestly.”
The group-context delay discounting assessment (Charlton et al., 2013) followed the same
format (i.e., same delays and monetary amounts), however, the choices were presented as
benefiting the responding individual and 9 other students enrolled in the Introduction to Applied
Behavioral Science course (see appendix D). Participants were provided with the following
assumption:
“Imagine you must make decisions that affect you and 9 other classmates in ABSC 100.
For each of the next 27 choices, please mark which hypothetical reward you would
prefer: the smaller reward today, or the larger reward in the specified number of days.
However, the choices that you will make will benefit you and 9 other classmates in
ABSC 100. While neither you nor each of the classmates will not actually receive the
rewards, pretend that you all will actually be receiving the amount you indicate and
answer honestly as if the rewards were real.”
The Social Value Orientation assessment consisted of a 9-item trichotomous choice paradigm
assessing an individual’s prosocial, individualistic, and competitiveness orientations (Van Lange
et al., 1997; see appendix E). Participants were presented with three options in which they
15
selected hypothetical outcomes for themselves and an unknown stranger. Participants were
provided with an assumption regarding the task, provided below.
“In this task we ask you to imagine that you have been randomly paired with another
person, whom we will refer to simply as the "Other." This other person is someone you
do not know and that you will not knowingly meet in the future. Both you and the
"Other" person will be making choices by selecting either the letter A, B, or C. Your own
choices will produce points for both yourself and the "Other" person. Likewise, the
other's choice will produce points for him/her and for you. Every point has a value: the
more points you receive, the better for you, and the more points the "Other" receives, the
better for him/her.
Here's an example of how this task works:
A B C
You get: 480 540 480
Other gets: 80 280 480
In this example, if you chose A you would receive 480 points and the other would receive
80 points; if you chose B, you would receive 540 points and the other 280; and if you
chose C, you would receive 480 points and the other 480. So, you see that your choice
influences both the number of points you receive and the number of points the other
receives. Before you begin making choices, please keep in mind that there are no right or
wrong answers - choose the option that you, for whatever reason, prefer most. Also,
remember that the points have value: The more of them you accumulate, the better for
you. Likewise, from the "other's" point of view, the more points s/he accumulates, the
better for him/her.”
Participants’ pattern of choices was classified as one of three categories. Prosocial involves
taking a lesser value of points thereby providing the “other” individual with an equal amount of
points (i.e., choice C in the provided assumptions). Competitive involves taking a lesser amount
of points thereby providing the “other” individual with a very small amount of points (i.e., choice
A). Lastly, individualistic involves taking the highest amount of points, thereby providing the
“other” individual with a medium amount of points (i.e., choice B). If more than four responses
were for a single pattern of responding (e.g., 5 responses in the prosocial category), the
individual responded in ways that are designated as social orientation.
16
The Revised Competitiveness Index consists of 14-items divided into two subscales;
enjoyment of competition and contentiousness (Houston et al., 2002; see appendix F), in which
participants rated themselves on a five-point Likert scale.
Group foraging task
Participants
One week after the online survey was made available, students in the Introduction to
Applied Behavioral Science course were allowed to participate in an optional in-class extra
credit opportunity that took place in their normal classroom during a regularly scheduled class
period. 154 participants completed the in-class portion of the study. Demographic data were not
taken, although participants provided their unique iClicker IDs following the completion of the
in-class study, allowing the researcher to compare results obtained from the in-class portion with
the online survey portion. The amount of extra credit that participants earned was determined by
their performance in the group foraging task and is shown in Table 1. Performance was based
upon the amount of accrued hypothetical points during the task.
Materials and Setting
The group foraging task took place in a large, 272 seat lecture hall with stadium-style
seating. Each participant sat facing three large screens that displayed relevant information.
Participants were instructed to bring their iClicker 2s (https://www.iclicker.com/; see appendix
G), a remote device that is used for live quiz responding during the regularly scheduled class
periods. Each iClicker has a power key, arrows for selecting various options, and 5 response keys
(A-E). For the purpose of the group foraging task, participants were instructed to power on their
iClickers and only respond using either the “A” or “B” response keys. The iClickers were synced
with the computer in the classroom, allowing for automated data collection. Each participant was
17
provided with the information statement (see Appendix A), instructions, and a data collection
sheet (see Appendix H) for each condition (six) to record their point earnings for each trial.
Two undergraduate research assistants provided inter-observer agreement during data
collection. One research assistant sat next to the researcher and recorded the number of
individuals that chose either option “A” or “B” during the task on an Excel spreadsheet. The
Excel spreadsheet was programmed to calculate the point values that the participants earned after
each trial, by dividing the number of people selecting each option by the pre-determined
condition values. The condition point values can be found in Table 2. This research assistant
helped ensure the accuracy of data by checking that the number of responses on the clicker
software matched what the researcher entered in the feedback spreadsheet. If any errors
occurred, the research assistant was trained to fix these issues before the feedback was presented
to the participants. Despite this, there were no reported errors, and the procedures went according
to plan. The second research assistant helped ensure fidelity by blacking out the trial screens
while providing feedback (i.e., the number of points and individuals at each resource site). The
primary investigater recorded any errors in procedural fidelity, however, there were no errors in
feedback delivery.
Procedure
Participants were instructed to read through the information statement prior to the study.
Following the information statement, the instructions were displayed on the middle screen at the
front of the lecture hall, and the instructions were read aloud to the participants (see below).
“We are interested in how people make choices and decisions. Today, you will make a
series of choices using your iClickers. You will be able to freely choose and change your
responses within each trial, but when the trial ends, you must record your choice by
circling either “A” or “B” in the resource selection column on your data sheet. Following
this, you will record the number of points that you earned on that trial, as well as the
number of people in each resource site. This information will be provided to you after
18
every choice trial. There will be 6 conditions with 15 trials each, totaling 90 trials. Your
task is to earn as many points as you can, which will then be calculated into extra credit.
There will be a certain amount of points delivered for each resource site after each trial.
These points will be divided equally between each person in the zone. For example, if
there are 5 people in zone “A” and there is 1 point available, then each person in the zone
will be rewarded (1/5) or .2 points. There will be many choices today, so it is up to you
how to respond. If you falsify your records to try to earn more points, you will receive a
score of zero for this task. Yet, if you record your responses accurately, you will receive
an additional 0.1% extra credit added to your overall final grade. Finally, once the task
begins, there is no talking allowed. If you are being disruptive during this task, we will
respectfully ask you to leave the classroom. This is done to ensure that your choice
responses are based on your own judgments, rather than based on what others say. If you
have any clarifying questions, now is the time to ask them. We will not answer questions
or comments during the task.”
After reading the instructional statement, participants were led through two practice trials to help
participants understand the task. Participants were instructed to make a choice between resource
sites “A” or “B”, and were told that they had 15 seconds before the trial ended. Immediately after
this trial, the experimenter provided feedback to the participants via an Excel spreadsheet on the
screen for the current trial. The information provided consisted of the number of people in each
resource site (A or B), as well as how many points were earned per person for A and B.
Participants were then instructed that they had 15 seconds to record this information on their data
sheets. The purpose of self-recorded data sheets were a) to help promote attending to the task, b)
to provide an additional layer of feedback, and c) to provide an additional measure of data
reliability (in addition to the iClicker software and excel spreadsheet).
As per Madden et al.’s (2002) experiment two, participants were not allowed to talk. Two
research assistants monitored the class to ensure there were no distractions or interruptions.
The experiment began following the completion of the practice trials. Participants
completed 15 trials per condition, with feedback provided after each trial. Additionally, live
feedback regarding the number of individuals at each resource site was provided via the iClicker
software during each trial (see appendix K). Participant distribution was presented as a bar graph
19
during the live feedback within each trial, with live adjustments automatically made as
participants selected or switched resource sites. Each trial took roughly 15-seconds to collect
responses, and roughly 15-seconds to provide feedback. Following the completion of the
condition (i.e., 15 trials), a condition change slide was presented and read aloud (see appendix J).
The condition change slide stated that “The number of points allocated to ‘A’ and ‘B’ has now
changed. It is up to you to determine the best way to earn points”. This process continued until
all 6 conditions (90 trials) were completed. Following the completion of the 90 trials, participants
were required to turn in their paper-pencil data sheets to the research assistants and were allowed
to leave the classroom.
Results
Delay Discounting
Delay discounting results were obtained for 160 participants. Figure 1 shows the natural
log transformed k-values for small, medium, and large hypothetical monetary magnitudes for
both individual and group context delay discounting assessments. The individual MCQ delay
discounting values were natural log transformed and show a mean small k-value of -3.5646
(SEM=.1200), a mean medium k-value of -4.2374 (SEM=.1290), and a mean large k-value of
-4.7083 (SEM=.1293). Additionally, for the group-context natural log transformed delay
discounting values, the average small k-value of -3.7940 (SEM=.1305), the average medium
k-value of -4.3597 (SEM=.1263), and the mean large k-value of -4.7312 (SEM=.1366).
Individual MCQ discounting natural log transformed values ranged from -8.74034 to -1.38629,
with an average k-value of -4.13943 (see figure 2 for the distribution of natural log transformed
k-values). Group discounting natural log transformed values ranged from -8.74034 to -1.38629,
with an average k-value of -4.27381. Individual delay discounting and group context delay
20
discounting were compared using a 2 (condition) x 3 (magnitude) repeated-measures ANOVA.
These analyses demonstrated a significant main effect of magnitude (F(2, 318) = 141.3,
p < .0001), a non-significant main effect of condition (F(1, 159) = 2.584, p = .1099), and a non-
significant interaction (F(2, 318) = 2.335, p = .0985). Overall consistency scores for the
individual discounting assessment was 96.13%, while the group delay discounting consistency
scores were 95.81%.
Competitiveness and Social Values Orientation
Competitiveness scores were obtained for 159 participants, with an average
competiveness score of 45.95 (SEM=0.716). The maximum competiveness score of a participant
was 67, while the minimum score was 15. Social Value Orientation assessments were completed
by 160 participants. Participants were scored into four different categories based on their pattern
of responding: Competitive, individualistic, prosocial, or mixed. There were 71 participants that
were categorized as prosocial, 57 categorized as individualistic, 24 categorized as competitive,
and 8 categorized as mixed.
Group Foraging
Of the 154 participants that completed the in-class group foraging task, data were
analyzed for 147 participants due to missing data (i.e., no responses). Figure 3 shows group
foraging data for each condition. The number of people that selected resource site “A” is
presented across each trial in the group foraging task. As the trials progressed across all 6
conditions, the number of individuals stabilized near the predicted level of distribution based on
the IFD theory.
The relative proportion of individuals choosing resource site “A” vs. resource site “B” as
a function of the relative proportion of points available at resource site “A” and “B” is
21
represented in Figure 4. The left panel indicates the average distribution of individuals across all
conditions for the first 5 trials (y=.5872x+.0606, r2=.9590), the middle panel represents the
average distribution of individuals across all conditions for the middle 5 trials (y=.8599x-.0040,
r2=.9970), and the right panel represents the average distribution of individuals across all
conditions for the last 5 trials (y=.9626x+.0049, r2=.9993). Each data point is representative of
one condition, totaling 6 data points per trial block (i.e., five conditions plus one direct
replication condition). To assess whether there was a difference between the first five, middle
five, and last five trials in the group matching plot, a comparison was conducted between a free
y-intercept and shared yet constrained y-intercept. Constraining a shared y-intercept between the
three regression lines showed significantly different slopes F(2,14)=28.94, p<.0001.
Additionally, while allowing a non-constrained y-intercept for the three regression lines, the
slopes were found to be significantly different F(2,12)=25.65, p<.0001.
Figure 5 shows the average number of individuals observed in resource site “A” across
the last five trials of each condition were plotted as a function of the predicted number of
individuals based on the ideal free distribution equation (see equation 3). Linear functions
accounted for 99.91% of the variance in the data, with a best fit line of y=0.973x + 2.472.
Figure 6 shows the probability of switching resource sites, defined as the proportion of
individuals that switched resource sites as a function of the proportion of points earned on the
previous trial. The x-axis represents the log-transformed proportion of points earned on the
selected resource site compared to the points earned in the other resource site. The y-axis depicts
probability of switching. Each data point represents an instance in which a point value was
provided and at least one person switched.
22
Correlations between individual discounting, group discounting, competitiveness
assessment, Social Value Orientation, and performance on the group foraging task (i.e., total
points earned, number of switches) are compared in Table 3. There was a statistically significant
relationship between individual and group discounting (r(159) = 0.710, p < .001), and individual
discounting and total points earned (r(159) = 0.186, p < .05). There were no other statistically
significant relationships between these metrics.
Predictive Utility
Table 4 shows results from a multiple regression analysis assessing the predictive utility
of the psychological and behavioral independent variables that were assessed (i.e., individual
delay discounting, group context delay discounting, competitiveness index, SVO) on the
dependent variable of total points earned. Since the purpose of the current study was to assess
competitive success, the dependent variable of total points was selected. Results indicated that, in
combination, the proportion of trials switched (β = -6.366, t(152) = -3.177), p = .002), individual
delay discounting (β = -0.450, t(152) = -2.152, p = .034), and the competitiveness index (β =
.053, t(152) = 2.301, p = .023) were statistically significantly predictive, while controlling for all
other variables.
Discussion
Delay Discounting
Overall, there was a magnitude effect for both individual and group-context delay
discounting that significantly differed between magnitudes (i.e., small, medium, large). There
were no condition differences observed between the individual and group context delay
discounting assessments, despite condition differences in past research (e.g., Bickel,
Jarmolowicz, Mueller, Franck, Carrin, and Gatchalian, 2012). However, there may be limitations
23
in the current study that could account for the inconsistences in these findings. One potential
explanation could be that the social distance of the group in Bickel et al.’s (2012) study was
greater than in the current study. In this sense, the student participants from the introductory
course in the current study may have had a closer social proximity than the groups used in Bickel
et al.’s (2012) study. Additionally, our group context discounting assessment differed from the
traditional group context (i.e., “me-we”) assessment. Traditionally, “me-we” discounting
provides the assumption that individuals are responding for hypothetical rewards for themselves
and a group of anonymous individuals. However, in order to assess the social cohesiveness of
our group foraging task (i.e., introduction to behavioral science students), the assumptions were
adapted to include classmates in the group context. These differences in the group context
assessment may account for the inconsistencies between studies.
There was a similar range of values for both the individual and group context delay
discounting assessments which allowed for comparison between the metrics, showing a
statistically significant correlation between both the individual and group context delay
discounting assessments.
There were a range of values and categories for the competitiveness and social values
orientation assessments, respectively, allowing for comparisons between these metrics.
Ideal Free Distribution Theory
Group foraging was described by the IFD theory, accounting for up to 99.91% of the
variance in the data. Undermatching was observed, suggesting that changes in the number of
individuals at a resource site were less extreme than changes in the available resources at that
site. For each condition, as trials progressed, the distribution of participants got closer to ideal
matching. That is, as participants experienced resource allocation in each condition, they
24
distributed themselves in a way that closely matched the proportion of available resources. This
finding is consistent with past research, in that as participants experience the contingencies
associated with a resource site selection, and are able to freely switch between resource sites
within a trial, their distribution was better accounted for by the IFD theory (e.g., Madden et al.,
2002). Additionally, as seen in Sokolowski et al. (2015), there was a high degree of
correspondence between the average observed and predicted number of individuals selecting
each resource site across the last five trials of each condition (R2 = .9991).
The probability of switching resource sites, as defined by the proportion of individuals
that switched resource sites following trials of varying point earnings, followed a similar pattern
as found in past research (i.e., Critchfield and Atteberry, 2003; Madden et al., 2002). Participants
that experienced lower point values relative to other resource sites were more likely to switch
resource sites, while individuals that experienced medium point values relative to other resource
sites tended to stay. The “big-win” pattern of responding, wherein individuals that earned
relatively high point values relative to the other resource site, had an increased likelihood of
switching sites.
Critchfield and Atteberry (2003) investigated the relationship between delay discounting
and individual success in a group foraging task, hypothesizing that individuals that demonstrate
steeper discounting would perform worse than individuals that demonstrate shallower
discounting. Results from their study showed more impulsive participants tended to choose the
lean patch more often, select the worse of the two resource sites more often, switch more often
between resource sites, and earn fewer overall points, while compared to the other more self-
controlled participants. Despite the same rationale for studying the relationship between delay
discounting and the IFD theory, the current results slightly differ. In our study, there was a
25
significant positive correlation between individual discounting and total points earned,
suggesting that the more impulsive individuals (as defined by steeper discounting) earned more
overall points. Additionally, a multiple regression analysis showed a predictive negative
correlation between total points earned for individual discounting, and the proportion of trials
switched, and a predictive positive correlation between total points and scores on the
competitiveness index, while controlling for all other assessments. However, despite these
differences in the obtained results between these studies, it should be noted that there were
differences in the procedures used. For instance, our study employed two empirically validated
discounting assessments; a standard 27-item MCQ assessment and an adapted group-context
assessment (i.e., Kirby et al., 1999; Charlton et al., 2013). Additionally, in Critchfield and
Atteberry’s (2003) group foraging task, participants were not allowed to switch within trials.
During the current study, participants were allowed to freely switch between resources sites
within a 15-second trial. As past research has shown, experimental manipulations to the ability to
switch within a trial, the amount of feedback provided, and other procedural variations (e.g., the
ability to talk, free-operant vs. discrete trial arrangement) all result in varying levels of
conformity to the IFD theory (e.g., Critchfield & Atteberry, 2003; Madden et al., 2002;
Sokolowski et al., 2015; Kraft & Baum, 2001). Therefore, the difference in experimental
arrangements could have accounted for the differences in findings between the current study and
Critchfield & Atteberry (2003).
Despite finding a statistically significant positive correlation between standard
discounting and total points earned, and a multiple-regression analysis indicating a predictive
correlation between group context discounting and the proportion of switches, there may be other
assessments or processes that could better account for the IFD performance data. These data may
26
support the notion that a single pattern of responding or assessment may not easily explain
performance in this group foraging task (Critchfield & Atteberry, 2003).
These data suggest that there is variation in individual choice patterns while operating in
a group foraging context. That is, despite individuals operating under the same group foraging
contingencies, there is variation in their individual performance such that some participants
maximize their earned resources while others earn relatively few total resources. Understanding
this variability through psychological and behavioral assessments may help us understand why
organisms are more or less sensitive to group contingencies.
Future research could investigate other behavioral processes, alone or in combination, to
see whether there are better predictors of performance in the group foraging task. Additionally,
the effects of group placement are an avenue for future research. That is, to date, group
placement in IFD tasks have been largely arbitrary. However, it may be that because
performance in a group foraging task is dependent on both the available resources and the other
foragers, the arrangements of groups could influence individual performance. For instance, after
identifying the correlation between standard discounting and total points earned, future research
could compare performance between highly impulsive groups, more self-controlled groups, or
groups that are a mix of more and less impulsive individuals. By better understanding how and
why individuals perform in competitive group contexts, and by understanding the interactions
when group placement is a variable, we could better inform group placement in societal contexts
(e.g., military settings, work settings, school settings) and maximize earned resources.
In the group foraging experimental arrangement, manipulations of resource availability
resulted in proportional changes in the number of individuals at each resource site. These data
supported the IFD theory, suggesting that by changing resource availability (e.g., amount,
27
quality, frequency of delivery), there would be changes in the allocation of individuals at these
resource sites. Although the group foraging task represents a simplistic paradigm, if these results
are conceptually scaled up, it suggests that the theory could be applied to the built environment.
For instance, in areas that are have fewer resources allocated to positive outlets (e.g., libraries,
parks, school programs, athletics), and more resources allocated towards negative outlets (e.g.,
gang activity, unhealthy foods, drugs), we will likely see, a higher proportion of individuals
sampling the negative outlet resource sites. However, as suggested by the IFD theory, by
manipulating the availability of resources there will be corresponding changes in the distribution
of individuals, thus influencing the likelihood that these individuals sample negative outlets.
There are a few potential limitations that should be noted. As described in the Method
section, data were obtained using both paper-pencil self-recorded methods and the electronic
automated iClicker software. Following the collection of data, there were unanticipated issues
associated with each data collection method. Paper-pencil data demonstrated inaccurate records
of point values and resource site selections, but each trial was filled in for each participant.
Conversely, the automated iClicker software recorded accurate records of point values and
resource site selections, but there were missing data for some participants. Presumably, this was
because either the software did not record a response from a participant, or the participant did not
make a response (but recorded it on their data sheet). Therefore, the data from both the paper-
pencil and iClicker software were plotted and compared (see figure 3), demonstrating a
statistically significant correlation between these values (r(145) = 0.751, p<.0001). Additionally,
a best-fit line was also plotted (y = .913X + 2.686), with an R2 value of .848. Outliers that were
identified were excluded from the correlational and best-fit line analyses, but are plotted as red
dots on figure 3. The electronic iClicker data were used throughout all analyses because they
28
were automated, eliminated self-report errors, and were used for immediate feedback during the
group foraging task.
Another potential limitation is that the number of participants that completed the online
survey did not match the number of participants that completed the in-class group foraging task.
Because these tasks were offered as extra credit options, they were not required and thus
individuals could opt out of one or both of the research activities. However, given the large
sample sizes of each aspect of the current study, there were enough participants that completed
both the in-class and online tasks to warrant statistical analyses. Any data that were used to
compare the online survey results with the group foraging task were only for participants that
completed both components of the study.
Overall, the current study sought to investigate whether several commonly used
behavioral and psychological assessments were predictive of individual performance in a large-
scale group foraging task. Notably, individual competitive success (i.e., total points earned in the
group foraging task) was statistically significantly predicted by the proportion of trials switched,
individual delay discounting, and total competitiveness index score, when controlling for all
other assessments. As a group, the distribution of participants was well described by the IFD
theory.
29
References
Abrahams, M. V. (1989). Foraging guppies and the ideal free distribution: The influence of
information on patch choice. Ethology, 82, 116-126.
Alferink, L. A., Critchfield, T. S., Hitt, J. L., & Higgins, W. J. (2009). Generality of the matching
law as a descriptor of shot selection in basketball. Journal of Applied Behavior Analysis,
42(3), 595-608.
Baum, W. M. (1974). On two types of deviation from the matching law: bias and
undermatching. Journal of the experimental analysis of behavior, 22(1), 231-242.
Baum, W. M., & Kraft, J. R. (1998). Group choice: Competition, travel, and the ideal free
distribution. Journal of the Experimental Analysis of Behavior, 69(3), 227-245.
Bickel, W. K., Jarmolowicz, D. P., Mueller, E. T., Franck, C. T., Carrin, C., & Gatchalian, K. M.
(2012). Altruism in time: Social temporal discounting differentiates smokers from
problem drinkers. Psychopharmacology, 224, 109-120.
Borrero, J. C., Crisolo, S. S., Tu, Q., Rieland, W. A., Ross, N. A., Francisco, M. T., &
Yamamoto, K. Y. (2007). An application of the matching law to social dynamics. Journal
of Applied Behavior Analysis, 40(4), 589-601.
Charlton, S. R., Yi, R., Porter, C., Carter, A. E., Bickel, W., & Rachlin, H. (2013). Now for me,
later for us? Effects of group context on temporal discounting. Journal of behavioral
decision making, 26(2), 118-127.
Clicker & Audience Response Systems - iClicker. (n.d.). Retrieved May 09, 2016, from
https://www1.iclicker.com/
Conger, R., & Killeen, P. (1974). Use of concurrent operants in small group research: A
demonstration. Pacific Sociological Review, 399-416.
30
Critchfield, T. S., & Atteberry, T. (2003). Temporal discounting predicts individual competitive
success in a human analogue of group foraging. Behavioural Processes, 64(3), 315-331.
Critchfield, T. S., & Kollins, S. H. (2001). Temporal discounting: Basic research and the analysis
of socially important behavior. Journal of applied behavior analysis, 34(1), 101-122.
Disma, G., Sokolowski, M. B., & Tonneau, F. (2011). Children's competition in a natural setting:
evidence for the ideal free distribution. Evolution and Human Behavior, 32(6), 373-379.
Epstein, L. H., Salvy, S. J., Carr, K. A., Dearing, K. K., & Bickel, W. K. (2010). Food
reinforcement, delay discounting and obesity. Physiology & Behavior, 100(5), 438-445.
Fagan, R. (1987). A generalized habitat matching rule. Evolutionary Ecology, 1, 5–10.
Fretwell, S. D., & Lucas, H. L., Jr. (1970). On territorial behavior and other factors influencing
habitat distribution in birds. Acta Biotheoretica, 19, 16–36.
Harper, D. G. C. (1982). Competitive foraging in mallards: “Ideal free ducks.” Animal
Behaviour, 30(2), 575-584.
Herrnstein, R. J. (1961). Relative and Absolute strength of Response as a Function of Frequency
of Reinforcement, 2. Journal of the experimental analysis of behavior, 4(3), 267-272.
Herrnstein, R. J. (1970). On the Law of Effect. Journal of the experimental analysis of behavior,
13(2), 243-266.
Houston, J., Harris, P., McIntire, & Francis, D. (2002). Revising the competitiveness index using
factor analysis. Psychological Reports, 90(1), 31-34.
Hursh, S. R., & Roma, P. G. (2013). Behavioral economics and empirical public policy. Journal
of the experimental analysis of behavior, 99(1), 98-124.
Kennedy, M., & Gray, R. D. (1993). Can ecological theory predict the distribution of foraging
animals? A critical analysis of experiments on the ideal free distribution. Oikos, 158-166.
Kirby, K. N., Petry, N. M., & Bickel, W. K. (1999). Heroin addicts have higher discount rates for
31
delayed rewards than non-drug-using controls. Journal of Experimental psychology:
General, 128(1), 78.
Kraft, J. R., & Baum, W. M. (2001). Group choice: The ideal free distribution of human social
behavior. Journal of the Experimental Analysis of Behavior, 76(1), 21-42.
Kraft, J. R., Baum, W. M., & Burge, M. J. (2002). Group choice and individual choices:
modeling human social behavior with the Ideal Free Distribution. Behavioural processes,
57(2), 227-240.
Madden, G. J., Peden, B. F., & Yamaguchi, T. (2002). Human group choice: Discrete-trial and
free-operant tests of the Ideal Free Distribution. Journal of the Experimental Analysis of
Behavior, 78(1), 1-15.
McDowell, J. J., & Caron, M. L. (2010). Matching in an undisturbed natural human
environment. Journal of the experimental analysis of behavior, 93(3), 415-433.
McDowell, J. J., & Caron, M. L. (2010). Bias and undermatching in delinquent boys’ verbal
behavior as a function of their level of deviance. Journal of the experimental analysis of
behavior, 93(3), 471-483.
Milinksi, M. (1984). Competitive resource sharing: an experimental test of a learning rule for
ESSs. Animal Behavior, 32, 233-242.
Odum, A. L. (2011). Delay discounting: I'm a k, you're a k. Journal of the experimental analysis
of behavior, 96(3), 427-439.
Pulliam, H. R., & Caraco, T. (1984). Living in groups: is there an optimal group size. In: Krebs,
J. R. and Davies, N. B. (eds) Behavioural ecology: an evolutionary approach, 3, 122-
147.
Reed, D. D., Critchfield, T. S., & Martens, B. K. (2006). The generalized matching law in elite
32
sport competition: Football play calling as operant choice. Journal of Applied Behavior
Analysis, 39(3), 281-297.
Reed, D. D., & Martens, B. K. (2008). Sensitivity and bias under conditions of equal and unequal
academic task difficulty. Journal of Applied Behavior Analysis, 41(1), 39-52.
Reynolds, B. (2006). A review of delay-discounting research with humans: relations to drug use
and gambling. Behavioural pharmacology, 17(8), 651-667.
Sokolowski, M. B. C., Tonneau, F., & i Baqué, E. F. (1999). The ideal free distribution in
humans: An experimental test. Psychonomic bulletin & review, 6(1), 157-161.
Sokolowski, M. B., Tonneau, F., & Cordevant, M. A. (2015). A portable system for studying
discrete‐trial group choice. Journal of the experimental analysis of behavior, 103(2), 419-
426.
Van Lange, P. A., De Bruin, E., Otten, W., & Joireman, J. A. (1997). Development of prosocial,
individualistic, and competitive orientations: theory and preliminary evidence. Journal of
personality and social psychology, 73(4), 733.
33
Table 1.
Percentile Rank Extra Credit
Earned
100-75 1.50%
74-50 1.25%
49-25 1.00%
24-0 0.75%
34
Table 2.
Condition Points A Points B Ratio
1 80 40 2:1
2 20 100 1:5
3 100 20 5:1
4 40 80 1:2
5 60 60 1:1
6 20 100 1:5
35
Table 3.
Competitiveness
Score
Group
Discounting
Individual
Discounting
Total
Points
Earned
Proportion
of Trials
Switched
Competitiveness ---
Group
Discounting
0.836 ---
Individual
Discounting
-0.062
0.710***
---
Total Points
Earned
0.144
0.023
0.186*
---
Proportion of
Trials Switched
-0.089
-0.140
-0.051
-0.121
---
*p<.05
**p<.01
***p<.001
36
Table 4.
DV = Total Points β SE t p
Intercept 68.422 1.476 46.342 .000
Prop. Trials Switched -6.336 1.994 -3.177 .002
Competitiveness .053 .023 2.301 .023
Individual Discounting -.450 .209 -2.152 .034
Group Context Discounting .303 .191 1.585 .116
Prosocial SVO
Individualistic SVO -.121 .444 -.272 .786
Mixed SVO .407 .867 .469 .640
Competitive SVO -.669 .622 -1.077 .284
R2 = .187 (Adj. R2= .134)
37
Figure 1.
38
Figure 2.
39
Figure 3.
40
Figure 4.
41
Figure 5.
42
Figure 6.
43
Figure 7.
44
Appendix A
45
Appendix B
46
Appendix C
47
Appendix D
48
Appendix E
49
Appendix F
50
Appendix G
51
Appendix H
Cond. 1 Last 4 digits of KU I.D.:
Trial # Resource Selection # of Points Earned # of People in "A" # of People in "B"
1 A B
2 A B
3 A B
4 A B
5 A B
6 A B
7 A B
8 A B
9 A B
10 A B
11 A B
12 A B
13 A B
14 A B
15 A B
52
Appendix I
53
54
55
Appendix J
56
Appendix K