The Boundaries of Freedom

MISSING IMAGE

Material Information

Title:
The Boundaries of Freedom An Evaluation of the Parameters of Preference for Choice
Physical Description:
1 online resource (117 p.)
Language:
english
Creator:
Rusak,Jennifer M
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Psychology
Committee Chair:
Dallery, Jesse
Committee Members:
Chambers, John
Branch, Marc N
LeBoeuf, Robyn A

Subjects

Subjects / Keywords:
adults -- alternatives -- amount -- better -- certainty -- chains -- choice -- concurrent -- differential -- forced -- free -- humans -- outcomes -- preference -- probability -- reinforcement -- response -- schedule -- versus
Psychology -- Dissertations, Academic -- UF
Genre:
Psychology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Organisms constantly face choices between responses and their consequences. Across a range of species and conditions, organisms prefer to make those choices between a greater number of alternatives as opposed to fewer or only one alternative; that is, they prefer free choice to restricted or forced choice. A number of variables contribute to preference for free choice, such as the number and quality of available alternatives, and reinforcer probability and amount. The purpose of the present set of five experiments was to extend previous work that has evaluated each of these variables. Specifically, previous research was replicated using within-subject methods with adult human participants, and several procedural variations among previous studies were directly assessed. In each experiment, participants made selections for either free or forced choice of response alternatives in the context of a concurrent-chains schedule. Experiments 1 and 2 examined the effect of number of alternatives and probability of reinforcement on preference for choice; results did not indicate consistent effects of either variable on preference for free choice. Experiments 3 and 4 were designed to test the hypothesis that a history of differential outcomes produces preference for free choice options. When outcomes (amount and probability of reinforcement) were better for free or forced choice, participants preferred the better option, but there was little carry-over of this history to conditions in which outcomes were equal. Between-subject variability with respect to free choice preference was high in Experiments 1-4. Experiment 5 tested the effect of presenting higher quality alternatives (as indicated by participants? rankings) in free and forced choice options, and this is the only experiment in which consistent preference for free choice was exhibited when free and forced choice outcomes were equal. Experiments 1-4 did not provide conclusive evidence to support and/or qualify the findings of previous research. These outcomes indicate a need to continue to study preference for choice of response alternatives and reinforcers with adult humans.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Jennifer M Rusak.
Thesis:
Thesis (Ph.D.)--University of Florida, 2011.
Local:
Adviser: Dallery, Jesse.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-08-31

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2011
System ID:
UFE0043156:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 THE BOUNDARIES OF FREEDOM: AN EVALUATION OF THE PARAMETERS OF PREFERENCE FOR CHOICE By JENNIFER M. RUSAK A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREM ENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011

PAGE 2

2 2011 Jennifer M. Rusak

PAGE 3

3 ACKNOWLEDGMENTS My sincere thanks go to my Ph.D. advisors : to Tim Hackenberg who encouraged and guided my start and to Jesse Dallery who kindly s upported my finish. I thank all my lab mates for their instruction, assistance and feedback and especially Anne Macaskill for her enthusiasm and generous programming assistance. My research assistants Michael Blakey, Joanna Lee, and Kerri Alexander gre atly facilitated the collection of these data and their work is much appreciated. I would like to thank Larry Winner for his patience and expertise in guiding me through the statistical analyses of my results. Thanks go to Marc Branch, John Chambers, and Robyn LeBoeuf for their time and input as members of my doctoral committee. I thank the Office of Graduate Minority Programs at the Univers ity of Florida for easing my financial burden and allowing me to focus on completing this dissertation. My friends especially Erik, Jonelle, Justine, and Scott have consistently been sources of reassurance, humor, commiseration and caring and I will enough for their contribution to both my academic and personal life, and above all I thank my parents for their unconditional understanding and support of all my pursuits no matter where I went or how long I took.

PAGE 4

4 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 3 LIST OF TABLES ................................ ................................ ................................ ............ 6 LIST OF FIGURES ................................ ................................ ................................ .......... 7 ABSTRACT ................................ ................................ ................................ ..................... 8 CHAPTER 1 GENERAL INTRODUCTION ................................ ................................ .................. 10 Evidence for Preference for Choice ................................ ................................ ........ 11 Why is Choice Preferred? ................................ ................................ ....................... 13 Number of Free Choice Alternatives ................................ ................................ 15 Certainty of Reinforcement ................................ ................................ ............... 16 Pa st History with Multiple Alternatives ................................ .............................. 17 The Present Study ................................ ................................ ................................ .. 20 2 EXPERIMENT 1: NUMBER OF ALTERNATIVES ................................ .................. 24 Method ................................ ................................ ................................ .................... 25 Participants, Setting, and Apparatus ................................ ................................ 25 Experimental Design ................................ ................................ ........................ 26 Concurrent Chains Procedure ................................ ................................ .......... 27 Within and Between Subjects Manipulations ................................ .................. 28 Pr actice Exposure Trials ................................ ................................ ................... 29 Data Analysis ................................ ................................ ................................ ... 29 Results ................................ ................................ ................................ .................... 30 Discussion ................................ ................................ ................................ .............. 32 3 EXPERIMENT 2: REINFORCER CERTAINTY ................................ ...................... 40 Method ................................ ................................ ................................ .................... 42 Participa nts, Setting, and Apparatus ................................ ................................ 42 Experimental Design ................................ ................................ ........................ 42 Concurrent Chains Procedure ................................ ................................ .......... 43 Within and Between Subjects Manipulations ................................ .................. 44 Practice Exposure Trials ................................ ................................ ................... 45 Data Analysis ................................ ................................ ................................ ... 46 Results ................................ ................................ ................................ .................... 46 Discussion ................................ ................................ ................................ .............. 48

PAGE 5

5 4 EXPERIMENT 3: AMOUNT OF REINFORCEMENT ................................ .............. 54 Method ................................ ................................ ................................ .................... 57 Participants, Setting, and Apparatus ................................ ................................ 57 Experimental Design ................................ ................................ ........................ 58 Concurrent Chains Procedure ................................ ................................ .......... 59 Within Subject Conditions ................................ ................................ ................ 60 Betwee n Subjects Manipulation ................................ ................................ ....... 61 Data Analysis ................................ ................................ ................................ ... 62 Results ................................ ................................ ................................ .................... 62 Discussio n ................................ ................................ ................................ .............. 65 5 EXPERIMENT 4: PROBABILITY OF REINFORCEMENT ................................ ...... 75 Method ................................ ................................ ................................ .................... 75 P articipants, Setting, and Apparatus ................................ ................................ 75 Experimental Design ................................ ................................ ........................ 75 Concurrent Chains Procedure ................................ ................................ .......... 76 Within Subject Conditions ................................ ................................ ................ 78 Between Subjects Manipulation ................................ ................................ ....... 79 Data Analysis ................................ ................................ ................................ ... 80 Results ................................ ................................ ................................ .................... 80 Discussion ................................ ................................ ................................ .............. 83 6 EXPERIMENT 5: AVAILABILITY OF PREFERRED ALTERNATIVES ................... 93 Method ................................ ................................ ................................ .................... 94 Participants, Setting, and Apparatus ................................ ................................ 94 Experimental De sign ................................ ................................ ........................ 94 Preference Ranking ................................ ................................ .......................... 95 Concurrent Chains Procedure ................................ ................................ .......... 96 Exp erimental Conditions ................................ ................................ ................... 98 Data Analysis ................................ ................................ ................................ ... 99 Results ................................ ................................ ................................ .................. 100 Discussion ................................ ................................ ................................ ............ 101 7 GENERAL DISCUSSION ................................ ................................ ..................... 109 Limitations ................................ ................................ ................................ ............. 109 Future Directions ................................ ................................ ................................ .. 111 Conclusions ................................ ................................ ................................ .......... 113 LIST OF REFERENCES ................................ ................................ ............................. 114 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 117

PAGE 6

6 LIST OF TABLES Table page 2 1 Experiment 1: Post session questionnaire contents ................................ ........... 37 2 2 Experiment 1: Mean proportions of free choice selections (non transformed) for the Certain and Uncertain groups in each condition for both condition sequences. ................................ ................................ ................................ ......... 37 3 1 Experiment 2: Post se ssion questionnaire contents ................................ ........... 51 3 2 Experiment 2: Mean proportions of free choice selections (non transformed) for the All Stimuli and Single Stimulus groups in each condition for both conditi on sequences. ................................ ................................ .......................... 51 4 1 Experiment 3: Post session questionnaire contents ................................ ........... 72 4 2 Experiment 3: Results of two tailed t test evaluat ing whether mean proportions of free choice selections (non transformed) are different from 0.5. 72 5 1 Experiment 4: Post session questionnaire contents ................................ ........... 90 5 2 Experiment 4: Results of two tailed t test evaluating whether mean proportions of free choice selections (non transformed) are different from 0.5. 90 6 1 Exper iment 5: Post session questionnaire contents ................................ ......... 106 6 2 Experiment 5: Mean proportions of free choice selections (non transformed) in each condition for both condition sequences. ................................ ............... 106

PAGE 7

7 LIST OF FIGURES Figure page 1 1 Schematic of a simpl e concurrent chains schedule. ................................ .......... 23 2 1 Schem atic of the concurrent chains schedule in place during choice trials for all conditions in Experiment 1.. ................................ ................................ ........... 38 2 2 Proportion of free choice selections in the Two and Five conditions in Experime nt 1.. ................................ ................................ ................................ .... 39 3 1 Schematic of the concurrent chains schedule in place during choice trials for all conditions in Experiment 2 ................................ ................................ ............ 52 3 2 Proportion of free choice selections in the Certain and Uncertain conditions in Experiment 2. ................................ ................................ ................................ ..... 53 4 1 Schematic of the concurrent chains schedule in place during choice trials for in the Mix ed Colors Baseline conditions in Experiment 3.. ................................ 73 4 2 Proportion of free choice selections in the Baseline, Forced Better and Free Better conditions in Experiment 3. ................................ ................................ ...... 74 5 1 Schematic of the concurrent chains schedule in place during choice trials for in the Mixed Colors Baseline conditions in Experiment 4.. ................................ 91 5 2 Pro portion of free choice selections in the Baseline, Forced Better and Free Better conditions in Experiment 4. ................................ ................................ ...... 92 6 1 Schematic of the concurrent chains schedule in place during choice trials fo r all conditions in Experiment 5. ................................ ................................ .......... 107 6 2 Proportion of free choice selections in the Baseline, Forced Better and Free Better conditions in Experiment 5. ................................ ................................ .... 108

PAGE 8

8 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE BOUNDARIES OF FREEDOM: AN EVALUATION OF THE PARAMETERS OF PREFER ENCE FOR CHOICE By Jennifer M. Rusak August 2011 Chair: Jesse Dallery Major: Psychology O rganisms constantly face choices between response s and the ir consequences Across a range of species and conditions organisms prefer to make those choices betwee n a greater number of alternatives as opposed to fewer or only one alternative; that is, they prefer free choice to restricted or forced choice. A number of variables contribute to preference for free choice such as the number and quality of available al ternatives, and reinforcer probability and amount The purpose of the present set of five experiments was to extend previous work that has evaluated each of these variables. Specifically, previous research was replicated using within subject methods with adult human participants and several procedural variations among previous studies were directly assessed. In each experiment participants made selections for either free or forced choice of response alternatives in the context of a concurrent chains sc hedule. Experiments 1 and 2 examined the effect of number of alternatives and probability of reinforcement on pre ference for choice; results did not indicate consistent effects of either variable on preference for free choice. Experiments 3 and 4 were de signed to t est the hypothesis that a history of differential outcomes produces preferenc e for free choice options. When outcomes (amount and probability of

PAGE 9

9 reinforcement) were better for free or forced choice, participants preferred the better option, but there was little carry over of this history to conditions in which outcomes were equal. Between subject variability with respect to free choice preferenc e was high in Experiments 1 4. Experiment 5 tested the effect of presenting higher quality alternati and this is the only experiment in which consistent preference for free choice was exhibited when free and forc ed choice outcomes were equal. Experiments 1 4 did not provide c onclusive evidence to support and/or qualify the findings of previous research. The se outcomes indicate a need to continue to study preference for choice of response alternatives and reinforcers with adult human s

PAGE 10

10 CHAPTER 1 GENERAL INTRODUCTION A coup le is undecided about the movie they will see and travels to a multiplex instead of a single screen cinema; a child spends more time working on a task for which she can choose her reward than one in which the same reward is chosen for her; a squirrel inh ab its a densely wooded area rather than a residential neighborhood where trees are fruitful but sparse. The common thread running through each of these scenarios is that the actors are choosing an option with more alternatives. Scientific studies conducted in different disciplines and using a variety of procedures, measures, and subject populations have demonstrated the same behavioral phenomenon : organisms prefer more choice to less choice Because choice is a ubiquitous concept with several possible mea nings, we need to define Catania (1975 ) point s out In the psychological vocabulary, the availability of alternatives has long been the basis for distinguishing among free choices and forced choices (p. 1). Therefore, to prefer (free) choice is to prefer access to multiple alternatives versus a single or fewer alternatives. Surprisingly, preference for choice has been observed across species even when response alternatives and outcomes in free and forced choice options are ident ical. The determinants of this preference are not well understood and require further examination in both humans and non humans. The present set of experiments conducted with humans further evaluated the roles of several variables that have been shown or hypothesized to contribute to preference for choice: namely, number of alternatives, current certainty of reinforcement, and past history of better outcomes when choosing among multiple alternatives.

PAGE 11

11 Evidence for Preference for Choice The alternatives from which organisms may choose are generally of two types: response alternatives or reinforcers. That is, organisms may choose among alternative behaviors in which to engage and/or alternative consequences of those behaviors. Most studies with non huma ns have investigated preference for choice of response alternatives. In an early study Voss and Homzie (1970) found that rats preferred a major pathway of a maze that offered two subsequent minor pathways before reaching the goal to one that did not pres ent a further choice between pathways. Later studies of preference for choice employed a more sophisticated procedure called a concurrent chains schedule In a simple concurrent chains arrangement, two stimuli (called initial links) are concurrently ava ilable and responding to those stimuli produces a new set of stimuli called the terminal links. Responding in the terminal links then produces reinforcement. A schematic of a simple concurren t schedule is shown in Figure 1 1 In studies of preference f or choice, responding on one initial link will produce a single stimulus in the terminal link link will produce an array of multiple stimuli in the terminal link ( Because each initial link is associated with its own distinct terminal link, relative rates of responding on concurrent initial links is used as a measure of preference for the terminal links, either forced choice or free choice. An organism prefers (free) choice if a greater proportion of its responding is allocated to the initial link associated with more stimuli in the terminal link.

PAGE 12

12 In several studies using concurrent chains arrangements pigeons have shown preference for a terminal lin k in which two response keys are available over a terminal link with only one key illuminated (Catania, 1975 1980 ; Cerutti & Catania, 1986, 19 9 7; Ono, 2000, 2004) M onkeys preferred free choice over forced choice w hen presented with terminal links in whi ch reinforcement delivery was probabilistic (Suzuki, 1 999). T hese experiments suggest that organisms prefer multiple alternatives to fewer. In addition Catania and Sagvolden (1980) conducted an experiment in which both terminal links consisted of four i lluminated keys. In the forced choice terminal link, three keys were red and signaled extinction and one was green and signaled reinforcement; in the free choice terminal link, three keys were green and one was red. Pigeons in this experiment preferred t he free choice link even though the number of keys in each terminal link was the same The experimenters concluded that the subjects preferred not just more visual stimuli, but more response alternatives (in this case more that would produce reinforcement ). Research with adult human participants has demonstrated preference for free choice of response alternatives in undergraduate students in the U.S. (Skowronski & Carlston, 1982) and Japan (Suzuki, 1997 2000 ) Fo r example, Suzuki (2000 ) used a concurr ent chains procedure in whic h participants chose between two initial links, one that led to a terminal link in which one response alternative was available and one t hat led to a terminal link in which multiple response alternatives were available. Partici pants in this study p referred both two and five terminal link alternatives versus one alternative.

PAGE 13

13 Organisms also make choices between reinforcers, and research suggests that similar processes control preference for choice of reinforcers and response alt ernatives. Both t ypically developing preschoolers (Fenerty & Ti ger, 2010; Tiger et al., 2006) and children with developmental disabilities (Fisher et al., 1997) have shown preference for free choice of reinforcers For example, Tiger, Hanley, and Hernan dez (2006) used a concurrent chains procedure in which the initial links consisted of different academic tasks. Completing each task resulted in a different consequence, and Tiger et al. observed that most of the preschooler participants c ompleted more ta sk s after which they could choose one reinforcer from an array of multi ple identical reinforcers than task s for which only one reinforcer was available. P revious research suggest s that choice is a powerful controlling variable affecting the behavior of h umans and non humans alike. It remains to be determined, however, exactly how or why choice functions in this capacity. A lthough several hypotheses have been offered, there is limited evidence to support particular explanations regarding what feature(s) of choice contribute to it being preferred. Therefore it is not clear why organisms prefer to make choices among multiple response alternatives and /or reinforcers. Why is Choice Preferred? P re ference for choice might arise in phylogenetic or in the ont og enetic history of organisms Catania and Sagvolden (1980) suggest a possible phylogenetic mechanism: more food supplies are available will probably have a survival advantage over one th at chooses patches of the environment containing o n p. 85). Such

PAGE 14

14 behavior is not necessarily a matter of an organism choosing more reinforcement over less. All available food supplies could be equally rich and an organism may for age in only one, but environmental changes could easily threaten this food supply. An organism is in danger if it chooses an area in which no alternative food source exists and is more likely to survive if it chooses an area in which it could switch to an unthreatened source with relatively little effort. It could be hypothesized that the organism that chooses an area with multiple food sources is more sensitive to the availability of multiple alternatives, a characteristic that may have been selected in its evolutionary history and can be passed on to its offspring. In addition Catania (1980) points out responding than does forced choice, the free choice preference is at least c onsistent p. 139). In short, or ganisms that seek out situations in which multiple response alternatives and/or reinforcers are available may be more likely to survive than those that do no t, and their progeny may then be predisposed to prefer conditions that offer choice. E xperience within the lifetime of an organism is also likely to contribute to preference for choice. Specifically, choice may be preferred over no choice i f an organis m has experiences in which th e outcomes associated with the availability of choice are more beneficia l than those associated with its unavailability For example, when organisms have access to more than one response alternative or reinforcer, it may be mo re likely that they prefer one of those alternatives compared to situations in which only one alternative is available. In addition, the availability of multiple alternatives may better accommodate temporary fluctuations in preference or changes

PAGE 15

15 in motiva tion In this case, the situation in which multiple alternative s are available is also more likely to be associated with reinforcing stim uli that are curr ently valued, and would be preferred to a s ituation in which fewer or only one alternative are availa ble. Having had these experiences, the organism learns to prefer conditions in which multiple response alternatives an d/or reinforcers are available, especially when a available The phylogenetic and ontogenetic processes that c ontribute to preference for choice are limited as explanations until tested. Although empirical evaluation of the phylogenetic nature of preference for choice is difficult, it is possible to control and environments in order to understand what ontogenetic experiences produce preference for choice. Along these lines, a number of studies have been conducted to determine some variables in the immediate environment that control preference for free choice ver sus forced choice. Number of Free Choice Alternatives Increasing the number of alternatives in a free choice option may increase the and result in an increase in preference for choice. P reference for free choice over forced choice has been demonstrated when the free choice option contains as few as two alternatives versus one in the forced choice option (Catania, 1975, 19 80; Cerutti & Catania, 1986, 199 7; Fisher et al., 1997; Ono, 2000, 2004; Skowronski & Carlston, 1982; Suzuki, 1999, 2000 ; Voss & Homzie, 1970 ) and some studies have found that increasing the number of free choice alternatives produces an increase in the degree of free choice preference. Catania (1980) observed a system atic increase in preference for a free choice option in one pigeon as the number of response keys in a free choice terminal link increased

PAGE 16

16 from two to four alternatives. Suzuki (2000) found that human participants exhibited a higher preference for free ch oice when three stimuli were present in the free choice terminal link compared to two stimuli, and Suzuki (1999) observed a similar effect with monkeys Tiger et al. (2006) found that increasing the number of identical reinforcers in an array from which p articipants could choose one resulted in an increase in the p roportion of trials in which participants chose the free choice option Interestingly, there is also evidence to suggest that too many alternatives in a free choice option may shift preference to no choice options (Iyengar & Lepper, 2000; Shah & Wolford, 2007) For example, in Experiment 1 of their study, Iyengar and Lepper (2000) set up displays of jams for grocery store patrons to sample. They found that 30% of the patrons who had sampled f rom an array of 6 jams went on buy a jar of jam compared to only 3% of the patrons who had sampled from an array of 24 jams. Similar results were observed in the other two experiments by Iyengar and Lepper and by Shah and Wolford who used simi lar dependen t measures However, both of these studies used i ndirect measures of preference making it difficult to draw conclusions regarding the number of alternatives that shifts preference from free choice to forced choice. Certainty of Reinforcement When rein forcement delivery is uncertain, orga nisms may prefer access to multiple response alternatives among which to switch Therefore, preference for free choice over forced choice may be higher when reinforcement is uncertain Studies conducted by Catania and colleagues showed preference for choice under a variety of conditions w hen reinfor cement delivery was certain in both the free and forced choice options. human participants showed preference for choice when the probability of reinfo rcement across free and forced choice options was 0.4. Similarly,

PAGE 17

17 Ono (2004 ) found preference for choice in pigeons when reinforcement probability was 0.5 in the free and forced choice options However, Ono (2000) observed differences in degrees of prefe rence for choice depending on whether reinforcement delivery was certain or uncertain when he conducted an explicit test of the effect of reinforcement certainty using a concurrent chains schedule Two groups of pigeons were exposed to a condition in whic h reinforcement delivered at the end of both free and for ced choice terminal links was certain and another condition in which reinforcement was delivered at a probability of 0.5. One group experienced certain reinforcement first and the other experienced uncertain reinforcement first. Pigeons in both groups were indifferent between free and forced choice when reinforcement was certain. Pigeons that experienced uncertain reinforcement first showed preference for the free choice option in that condition but pigeons who were exposed to certain reinforcement first continued to be indifferent between the two options when reinforcement was uncertain. These results indicate that reinforcement certainty influences free choice preference but its effect may be moderated by other variables (such as past experience). Past History with Multiple Alternatives Free choice is preferred even when it offers no obvious advantage over forced choice. In most of the studies with non humans cited above, whether subje cts c hose free or forced choice they were always presented with the same reinforcer (e.g. access to food) at the same probability (Catania, 1975, 1980; Catania & Sagvolden, 1980; Cerutti & Catania, 1986, 1997; Ono, 2000, 2004; Voss & Homzie, 1970). Subjects pr eferred free choice even though neither free nor forced choice was associated with the availability of a more preferred reinforcer. H umans have also shown preference for choice when identical reinforcers were delivered in free and forced choice options (T iger

PAGE 18

18 et al., 2006). Fisher et al. (1997) yoked reinforcers in the forced choice option to the reinforcers that were chosen when participants were given an array of multiple reinforcers. Under these conditions, participants preferred the free choice optio n. These studies indicate that preference for choice persists even when outcomes in free and forced choice option s are identical Although number of free choice alternatives and certainty of reinforcement control preference for choice, it is necessary to explain why these and any other controlling variables produce free choice preference under conditions in which free choice is no more beneficial than forced choice The explanations are likely to e merge fr om an understanding of the past history of organis ms and the effect of this his tory on current behavior. As described earlier, the most common suggestion regarding ontogenetic processes that contribute to free choice preference is that organisms have had a history in which free choice options afforded th forced choice options. When no differential outcomes exist between free and forced choice options in the current environment, that past history may be used to explain curre nt preference for free choice. The inf luence of past history can be appreciated by analyzing a choice between a free and forced choice option in the context of the operant contingencies in play. Consider a concurrent chains schedule in which the free choice terminal link contains two or more response alternatives indicated by distinct discriminative stimuli For example, a pigeon that responds to obtain a free choice terminal lin k would be confronted with two or more illuminated response keys, each of which functions as a discriminative stimu lus signaling that responding to the key will produce reinforcement.

PAGE 19

19 R esearch on the function of discriminative stimuli has indicated that, in addition to occasioning responding, these stimuli function as conditioned reinforcers. If a free choice termina l link contains multiple discriminative stimuli that occasion multiple response alternatives, it might also be said that the free choice terminal link contains multiple conditioned reinforcers. Therefore, preference for a free choice terminal link could s imply be a preference for multiple conditioned reinforcers versus the single conditioned reinforcer in the forced choice terminal link. Preference for free choice of reinforcers demonstrated when the free choice array consists of multiple reinforcers fr om which subjects choose one ca n be explained by appealing to similar principles. That is, in both cases, the subject may be attracted to the free choice option simply because the visual array signaling free choice contains more reinforcing stimuli than the forced choice visual array. This account of free choice preference as a preference for multiple conditioned reinforcers does not necessarily require that the subject have any prior history in which free choice produced a be tter outcome than forced ch oice. I t does require, however, that the subject had some history with the discriminative stimuli or reinforcers in order for them to have been established as conditioned reinforcers. One problem for this account, then, is explaining preference for choic e in novel situations when particular stimuli have not yet been established as conditioned reinforcers. An alternate account of the role of conditioned reinforcement in preference for free choice involves a process in which a history with differential ou tcomes in free and forced choice options is a key variable. As Fisher et al. (1997) explain : From an ontogenic perspective, Catania (1980) suggested that individuals may prefer choice over no choice because it provides a mechanism for

PAGE 20

20 adjusting reinforce r delivery in relation to momentary fluctuations in reinforcer over time because it is often correlated with increa sed access to preferred stimuli ( p. 435). In other words, preference for c hoice may not be due to the fact that subjects are attracted to particular stimuli that have been established as conditioned reinforcers; rather, the choice context itself comes to function as a conditioned reinforcer due to experiences in which free choic e provided some better outcome than forced choice (either access to more preferred stimuli, as Fisher et al. suggest, or greater magnitude or better probability of reinforcement, etc.) In this view, multiple stimuli do not function independently as condi tioned reinforcers that control preference for choice but instead the presence of multiple stimuli function as a unit as a conditioned reinforcer. The difference is subtle but important because if the latter is the case, then preference for free choice is more likely to generalize to any situation ( even a novel one) in which an array of multiple stimuli (either discriminative stimuli that occasion multiple response alternatives or multiple reinforcing stimuli) is present The Present Study The purpose of the present investigation was to assess the conditions under which preference for choice is exhibited. The first two experiments extend ed previous research on the role of number of alternatives and certainty of reinforcement in preference for choice. Sp ecifically, these experiments tested whether higher numbers of free choice alternatives and uncertain reinforcement would produce increases in free choice preference. They were conducted using within subject methods with undergraduate students in order to systematically replicate experiments conducted with non humans, preschoolers, and individuals with developmental disabilities

PAGE 21

21 E xperiments 3 and 4 were designed to test the hypothesis that a history of better outcomes in a free choice option leads to a p reference for free choice when outcomes in free and forced choice options are the same. These experiments also included conditions in which a forced choice option provided a better outcome in order to test the persistence of free choice preference and det ermine whether experiencing better outcomes from forced choice c ould lead to forced choice preference when outcomes were equal Finally, Experiment 5 evaluated the role of the availability of preferred alternatives as a contributor to preference for free choice. Preference for choice was available in both the free and forced choice option, in free choice only and in forced choice only. The procedures of Experiments 1 4 were designed to be comparable to previous non human and human studies that used arbitrary stimuli and provided choices between response alternatives instead of reinforcers. Participants in E xperiments 1 4 made choices between free and forced choice response alternatives that were represented as shapes shown on a computer screen. Choices were made via mouse click s on the shapes o n the screen and choices were repeated across multiple trials across experimental conditions. Using procedures similar to those of previous studies f acilitates comparisons between the present set of experiments and previous research. In addition, requiring participants to engage in repeated choices allows for a more sensitive measure of preference because the experimenter can observ e whether choices change over time. If choices remain stable, the experimenter can have more

PAGE 22

22 However, the results of Experiments 1 4 suggested that the use of arbitr ary stimuli may have limited the control over behavior of the independent variables. Therefore, while continuing to require repeated choices between free and forced choi ce options.

PAGE 23

23 Figure 1 1. Schematic of a simple concurrent chains schedule. In this example, in the initial links, responding to the single key on the left produces a forced choice terminal link and responding to the single stimulus on the left produces a free choice terminal link. Responding to these terminal link stimuli produces reinforcement. After reinforcement is delivered a new trial begins with the presentation of the initial links. Initial links Terminal links Forced C hoice Free Choice Reinforcement Response Requirement Response Requirement Reinforcement Response Requirement Response Requirement

PAGE 24

24 CHAPTER 2 EXPERIMENT 1: NUMBER OF ALTERNATIVES Several studies have indicated that preference for choice may be related to the number of alternatives available in a free choice option. Catania (1980) presented three pigeons with a choice between f ree and f orced choice terminal links u nder conditions in which the free choice link consisted of two, three, or four illuminated respon se keys. For one subject, increasing in the number of alternatives produced an increase in the degree of preference for the free choice link. Suzuki ( 2000 ) c onducted a comparison across groups of undergraduate students and observed a higher proportion of responding for a free cho ice option by participants that had five identical response alternatives compared to participants that had only two of those same alt ernatives in the free choice option Finally, Tiger et al. (2006) conducted a study in which preschoolers could complete either a task that gave them access to a choice of reinforcers or a task that gave them access to one reinforcer. Tiger et al. found that increasing the number of reinforcers from which to choose resulted in a higher proportion of responding on the task that led to a choice of reinforcers This was the case both for p articipants that initially preferred a choice among five reinforcers to one reinforcer and for p articipants who did not initially exhibit preference for choice of reinforcers These studies indicate that simply increasing the number of alternatives in a free choice option may result in an increase in preference for that o ption. The present experiment was de signed to replicate and extend previous studies by evaluating the effect of changing the number of alternatives in a free choice option on a within subject basis. Catania (1980) and Tiger et al. (2006) employed within subject design s but the only studies that have manipulated number of alternatives in a free

PAGE 25

25 choice option with adult human participants have done so on a between groups basis (Suzuki, 1997, 2000). Therefore, an ABAB reversal design was used in this s tud y to determine whether participan will change when they are exposed to conditions in which the free choice option contains both two and five response alternatives. C ertainty of reinforcement has also been shown to contribute to free choice preference. Ono (2000) found that pigeon subjects exposed to certain reinforcement were indifferent between free and forced choice options, but pigeons exposed to uncertain reinforcement (without a prior history of certain reinforcement) s howed a strong preference for free choice The certainty of reinforcement has differed among studies that have evaluated preference for choice and so it is useful to compare the results of these studies in order to draw conclusions about the influence of reinforcement certainty on preference for choice However, these studies have used different subje ct populations and methods so direct comparisons of the ir results is cautioned In the present experiment, a between subjects manipulation was used to asse ss the role of reinforcement certainty on preference for choice. Therefore, the present experiment extends previous research by direct ly assessing the effects of reinforcement certainty on preference for choice using adult human participants drawn from th e same population and exposed to the same methods. Method Participants, Setting, and Apparatus Thirty six students (2 6 females and 10 males) at the University of Fl orida served as participants. Participants were recruited from a pool comprised of stude nts enrolled in General Psychology and from two sections of Applied Behavior Analysis and they

PAGE 26

26 received course credit in exchange for participation. P to 2 2 years All participants completed the experimental task using th e mouse and keyboard of a PC in one of two rooms containing the PC a table, and a chair and were alone in the room for the duration of the task T he task was presented via a program designed in Microsoft Visual Basic 6.0. Experimental Design Participa concurrent chains procedure. Choices were made by moving a cursor to one of two options that differed in the number of visual stimuli available: only one alternative (forced choice) or sev eral alternatives (free choice). An ABA B reversal design was used to evaluate the effect of the number of free choice alternatives on preference for the free and forced choice options Participants were also divided into two groups (n=1 8 /group) in which the probability of point delivery after selection of an alternative was either certain (1.0) or uncertain (0.5). Experimental events and conditions are described in detail below. Before completing any trials, all participants were given a set of brief i nstructions describing the general structure of the experiment and telling them how to make responses. As a part of these instructions, they read the following statement: Clicking on the circle(s) on the screen may result in the delivery of points. The total number of points you have earned will be displayed on the screen. Your goal should be to try to earn as many points as you can. At the end of the session the total number of points you earned will be recorded by the researcher. Once all data from al l participants have been collected, if you have earned the highest total number of points you will win a $50 gift card to the store of your choice. This statement was included to help ensure that participants were sufficiently motivated to attend to the t ask and respond to earn points.

PAGE 27

27 Concurrent C hains Procedure In the choice trials of all conditions of the present experiment, a sequence of visual stimuli were presented on a computer screen according to a concurrent chains schedu le (as depicted in Figure 2 1). In the first component of this schedule, two arrays of visual stimuli were shown on the computer screen. One array contained one circle and the other contained multiple circle s These arrays were placed above two clickable buttons labeled A and B. Button A was always on the left and button B was always on the right, but the positions of the circle arrays were quasi randomized across trials. That is, sometimes one circle a ppeared above button A a nd multiple circles appeared above button B and vice versa; neither array appeared above the same button more than two times in a row. These initial arrays of s timuli are called initial links as they are the first links in the chain of stimuli that event ually lead to a reinforcer. A fixed ratio (FR) 1 response schedule was in place in the initial links: a single mouse click to b utton A or B removed both buttons and the array that had appeared above the non chosen button and constituted as a response fo r either the array with one ci rcle or for the array with multiple circles. A click to the button associated with one circle initiated the forced choice terminal link and a click to the button associated with multiple circles initiated the free choice term inal link. In the forced choice terminal link, one circle was positioned randomly on the screen and a fixed interval (FI) 5s schedule was in place. That is, the first click to the circle after 5 seconds had elapsed removed t he circle from the screen and end ed the terminal link. In the free choice terminal link, multiple circles were positioned randomly on the screen (without overlapping) and an FI 5s schedule was in place All of the circles remained visible for the entirety of the FI 5s

PAGE 28

28 schedule and t he first click on any of the circles after the fixed interval elapsed removed all circles from the screen and terminated the link. At the end of every terminal link a point was delivered, if scheduled. When a point was del ivered ear displayed on the screen during the initial link of every trial and continued to accumulate across conditions. An inter trial interv al (ITI) occurred after every terminal link, whether or not a point was delivered. During the ITI the screen remained blank for 3 seconds. At the end of the ITI a new trial began. Each condition ended when 2 4 choice trials were completed. Once all t he trial s in a condition were completed a new set of brief instructions appeared on the screen to remind the participant how to complete the task and to begin the next condition. When all four conditions (either Two Five Two Five or Five Two Five Two ) we re c ompleted and before they left the room, participants responded to a brie f questionnaire consisting of four questions. These questions are listed in Table 2 1. Within and Between S ubjects Manipulation s Within subject conditions All participants wer e exposed to both the Two and Five conditions. In the Two condition, two circles were present in the initial link associated w ith the free choice option and in the free choice terminal link In the Five condition, five circles were present in the initial link associated with the free choice option and in the free choice terminal link Between subjects manipulation. A between subjects manipulation was in place following the completion of the FI 5s schedule in all terminal links in all conditions. For h alf of the participants (Certain group) responding on a circle after the FI 5s

PAGE 29

29 schedule elapsed always resulted in the delivery of a point (1.0 probability of reinforcement) and for the other half of the participants (Uncertain group) a point was deliver ed with a probability of 0.5. Practice Exposure Trials At the beginning of each condition and before choice trials began, participants completed a set of 1 0 practice exposure trials. These trials were included to gu arantee that participants experience d the contingencies that were in place in the f ree and f orced choice terminal links in the subsequent choice trials. During the exposure trials, only and eithe r one or multiple (two or five, depending on the condition) circles appeared above the button. One mouse click to that button initiated either a f ree or f orced choice terminal link (depending on the trial type) The contingencie s in the terminal links of the practice exposure trials were the same as those in the choice trials. That is, FI 5s schedules were in place and points were delivered during these trials at a probability of either 1.0 or 0.5 ( Each set of practice exposure trials consisted of five f ree choice trials and five f orced choice trials with their order quasi randomized. Data Analysis D ata from individual participants and data averaged across participants were analyzed. For each participa nt, data from the last ten trials in each condition were compared to those from the first ten trials and to overall condition averages. No notable differences/trends were observed, and so data from all trials were used to calculate condition proportions. Individual data were graphed and visually assessed for trends, patterns and differences across conditions both within and between subjects. In

PAGE 30

30 addition, statistical analyses were conducted using proportion of free choice selections as the dependent meas ure. These proportions were transformed prior to statistical analysis (except for certain t tests as noted) P roportion s of free choice selections were converted using an arcsine transformation in which the arcsine was multiplied by the square root of th e proportion. This transformation is conducted on proportional data to better meet the assumptions of normal distribution and homogeneity of variance. The means and standard deviations reported for statistical tests in the Results section are those of th e transformed data (except where noted). Means and standard deviations reported in Table 2 2 are of the raw, and not transformed, data. Results The results of Experiment 1 a re presented in Figure 2 2. P roportion of free choice selections is graphed as a function of condition. This proportion was calculated for each participant by dividing the sum of free and forced choice selections by the total number of free choice selections. Bars indicate the mean proportion s of free choice selections for all p arti cipants in each condition and individual proportions are plotted as data points The left panels show data from participants in the Certain group and the right panels show data from the Uncertain group ; data in the top row are from participants wi th the order Two Five Two Five and data in the bottom row are from participants with the order Five Two Five Two Visual an alysis of these graphs indicated only a few difference s between t he mean proportion s of free choice selections across the Two and Five con ditions within group and within condition order for both the Certain and Uncertai n groups. On average, p reference for free choice was higher in the initial Five condition compared to the initial Two condition for the participants in the Uncertain group wh o had Five first (bottom right

PAGE 31

31 panel of Figure 2 2). In the Uncertain group, mean proportion of free choice selection s in the first Five condition for participants who had the Five condition first (bottom right panel in Figure 2 2) was higher than the mea n proportion of free choice selections in the first Two condition for participants who had the Two condition first ( top right panel). There also appeared to be an effect of reinforcement certainty when looking at the first Five conditions for the particip ants who received that condition first in the Certain and Uncertain groups (bottom row in Fig ure 2 2) Preference for free choice was higher in the first Five condition in the Uncertain group. In addition, mean proportions were near or below 0.5 for near ly all conditions in both groups. Analyses were conducted to reveal statistically significant results. C ondition m eans and standard deviations for all conditions are shown in Table 2 2. A univariate ANOVA did not reveal any significant main effects of condition, reinforcement certainty or condition order an d no significant interaction between condition and reinforcement certainty was indicated In addition no difference s determined by visual analysis w ere statistically significant according to inde pendent samples t tests A two tailed t test that used non transformed data indicated that the only condition in which the mean proportion of free choice selections was stat istically larger than 0.5 was the first Five condition in the subgroup of the Unce rtain group for whom the Five condition was the first condition ( M = .699, SD = .220), t ( 8 ) = 2 .712, p = .027. The lack of statistically significant results in the cases in which visual analysis determined a difference between conditions was likely due to the large degree of between subject variability observed in most of the experimental conditions across condition order and group This variability was evident by the wide spread of individual

PAGE 32

32 proportions as reflected in the standard deviations shown in Ta ble 2 2 and the spread of individual data points in Figure 2 2. Discussion The results of this experiment indicate that altering the number of response alternatives in the free choice option did not produce a consistent effect on sel ection of that option The results also indicate that certainty of reinforcement did not have an effect on preference for free choice. These results differ from those of several previous studies in which changes in number of alternatives and certainty of reinforcement produ ced changes in free choice preference. Suzuki (2000) compared free choice preference across two groups, one that had two free choice respo nse alternatives and another that had five free choice response alternatives. In that study, mean proportion of free choice selections in the Five Choice group was significantly greater than that for the Two Choice group. These differences were not reliably replicated within subject in the in the present experiment. Only 3 out of 18 participants in the Certain group an d 3 out of 18 participants in the Uncertain group showed a pattern in which proportion of free choi ce selections was higher in both Five conditions compared to both Two conditions. In addition, t he fact that a difference in mean proportions between the Fi ve and Two conditions was only observed across participants in the Uncertain group suggests that the effect of number of response alternatives may not be seen when reinforcement is certain. In fact, probability of reinforcement was but there was no comparison of number of alternatives under certain reinforcement in that study ). In the present experiment, n o differences in mean proportion s of free choice selections were observed between the Two and Five cond itions within the Uncerta in

PAGE 33

33 and Certain groups These results replicate the findings of Suzuki (1997) in which no differences were observed between a condition in which participants had two identical free choice alternatives and a condition in which a different group of participa nts had three identical free choice alternatives (with probability of reinforcement at 0.4) In addition, Catania (1980) did not see a consistent parametric increase in preference for a free choice option when he exposed pigeons to each of four conditions in which the number of free choice response alternatives was increased from one to four (with reinforcement probability at 1.0) ; this increase was only observed in one out of three subjects in one out of three replications of the comparison. There is evi dence that changing number of free choice alternatives is only effective for some individuals when that number is higher than five, which could account for the lack of consistent higher preference for five free choice alternatives in this experiment. For example, Tiger et al. (2006) found that three out of six participants were indifferent between free and forced choice of reinforcers when five items were in the reinforcer array; preference for the choice option only increased when the number of reinforcer s was increased to 10 Other studies have used indirect measures of preference to determine the effect of number of alternatives in a choice option. Shah and Wolford (2007) presented different participants with arrays consisting of different numbers of p ens. The number of participants that bought a pen from their array increased as the number of pens in the array increased The number of participants who bought a pen was highest when there were 10 pens in the array Similarly, Reutskaja and Hogarth (20 09) asked participants to choose a gift box from arrays of 5, 10, 15, or 30 boxes R atings of choice satisfaction and enjoyment were highest for the

PAGE 34

34 participant s who chose from 10 boxes These studies indicate that, for some participants, five alternativ es may not be enough to change their preference for a free choice option. Future research should investigate the effect of parametrically increasing the number of response alternatives in a free choice option on a within subject basis. Such an investiga tion would provide a more complete function along which number of alternatives changes free choice preference That is, it would be possible to see at what point changing the number of response alternatives produces free and forced choice preference. Studies that have indirectly evaluated preference for choice have indicated that choice may not be preferred when high numbers of alternatives are available in a free choice option (Iyengar & Lepper, 2000; Reutskaja & Hogarth 2009; Shah & Wolford 2007). The results of the present experiment also show that the certainty of reinforcement did not influence preference for the free choice option as wa s demonstrated by Ono (2000). No studies with human participants have directly investigated the effect of r einforcement certainty on free choice preference and so it is difficult to conclude why reinforcement certainty did not have an effect on a between subject s basis in this experiment. In general, t he mean proportion s of free choice selections observed in this study we re lower than those observed in other studies in which reinforcement was uncertain (Suzuki, 1997; 2000). One difference between the pre sent experiment and those other arned points exchangeable for money. It may be the case that participants in the present

PAGE 35

35 experiment were not sufficiently motivated to respond, which may have contributed to a lack of sensitivity to both reinforcement certainty and number of free choice a lternatives. An important finding in this experiment was the large degree of between subject variability in proportion of free choice selections. There were very few consistent patterns observed in the indi vidual data suggesting that number of altern atives was not a robust enough independent variable to exert control over behavior and/or that the experim ental procedure was ineffective In several cases, in both groups, participants maintained proportions of responding for free choice at around 0.5 ac ross all conditions Responses to the post session questionnaires indicated that several participants found the task boring, and when they experienced the fact that both options produced the same outcomes, they simply alternated responding for the fr ee an d forced choice options. This indifferent responding suggests that in the absence of differential consequences, number of free choice alternatives is not a powerful enough variable to alter preference for a free choice option. Several participants also the respond, even though there was no such pattern may have generated their own rules with regard to the task and this verbal behavior may have had more control over their behavior than the experimentally arranged contingencies. Several studies have found that when participants were given or generated their own rules with regard to reinforcement schedules, their behavior was insensitive to changes in those schedules. Rath er, behavior persisted in accordance with rules that did no t match actual contingencies ( c.f. Galizio, 1979; Hackenberg & Joker, 1994; Hayes, Brownstein, Zettle, Rosenfarb, & Korn,1986 ; Rosenfarb, Newland,

PAGE 36

36 Brannon, & Howey, 1992 ). It is possible that the use of arbitrary stimuli made it more likely for behavior to com e instruction rather than the experimental task making their behavior insensitive to the variables that were manipulated F uture research with adult h uman participants might be more successful in demonstrating an effect of number of alternatives on free choice preference if more l or simulated response alternatives (e.g. dishes on a menu, products in a grocery store, etc.) are used

PAGE 37

37 Table 2 1. Experiment 1 : Post session questionnaire contents Question # Question text 1 Which of the arrays did you choose more often, the one with one circle or the one with more than one circle? Why? 2 Did you prefer having more circles available? That is, were you more likely to choose the array with more than one circle when there were 5 versus when there were 2? Why or why not? 3 Did you enjoy earning points? 4 Is there anything else you would like to say about this study? Table 2 2. Expe riment 1: Mean p roportion s of free choice selections (non transformed) for the Certain and Uncertain groups in each condition for both condition sequences. Group Condition M Certain Two .519 (.238) Five .477 (.388) Tw o r ep .417 (.340) Five r ep .58 8 (.376) Five .454 (.288) Two .537 (.259) Five rep .500 (.277) Two rep .431 (.273) Uncertain Two .426 (.200) Five .458 (.359) Two rep .546 (.340) Five rep .565 (.367) Five .699 (.220)* Two .542 (.375) Five rep .565 (.355) Two rep .569 (.389) Note: Standard deviations are in parentheses Asterisks indicate that the mean proportion is significantly larger than 0.5 (p < = .05).

PAGE 38

38 Figure 2 1. Schematic of the concurrent chains schedule in place during choice trials for all conditions in Experiment 1. In the initial links, only buttons A or B could be clicked and the positions of the arrays of one or multiple circles were quasi randomized. The array of multiple circles contained either 2 or 5 c ircles depending on condition. In this example, Button A represents the forced choice option and Button B repr esents the free choice option. In the terminal links, clicks on the circles following the termination of the FI 5s schedule produced a point at a probability of either 1.0 (Certain condition) or 0.5 (Uncertain condition). Initial link Terminal link Forced Choice Free Choice FI 5s 1 pt delivered at 1.0 or 0.5 probability FI 5s 3s ITI FR 1 FR 1 A B

PAGE 39

39 Figure 2 2. Proportion of free choice selections in the Two and Five conditions in Experiment 1 Bars indicate proportions averaged across partic ipant; data points indicate proportions from individual participants The dashed lines are reference lines set at 0.5. Data from the Certain and Uncertain group s are in the top and bottom left and top and bottom right panels, respectively The top row s hows data from participants with the condition order Two Five Two Five and the bottom row shows data from participants with the condition order Five Two Five Two Proportion Free Choice Selections Condition Certain Uncertain

PAGE 40

40 CHAPTER 3 EXPERIMENT 2 : REINFORCE MENT CERTAINTY The results of Experiment 1 did not indicate that participants for whom reinforcement was uncertain preferred free choice more than participants for whom reinforcement was certain. However, the comparison of certain versus uncertain reinforcement in Experiment 1 was conducted on a between subject s basis. Ce rt ainty of reinforcement has been shown to contribute to free choice preference on a within subject basis Ono (2000) used a concurren t chains schedule to test the effect of reinforcement certainty on preference for choice in two groups of pigeo ns. Both groups were exposed to a condition in which reinforcement delivered at the end of both free and forced choice terminal links was certain and another condition in which reinforcement was delivered at a probability of 0.5. One group experienced ce rtain reinforcement first and the other experienced uncertain reinforcement first. Pigeons in both groups were indifferent between free and forced choice when reinforcement was certain. Pigeons that experienced uncertain reinforcement first showed prefe rence for the free c hoice option, but pigeons that were exposed to certain reinforcement first continued to be indifferent between the two options when reinforcement was uncertain. The present experiment was a systematic replication of Ono (2000) using a within subject comparison with undergraduate students as participants. As in Ono (2000), one group of participants was exposed to certain reinforcement first and the other was roups exhibited by the group of pigeons exposed to certain reinforcement first carried over to a subsequent uncertain rei nforcement condition. However, t he pigeons expo sed to

PAGE 41

41 uncertain reinforcement first did not repeat this condition after the certain reinforcement condition. This replication would have been a better test for carry over from the certain reinforcement condition. An ABA reversal design was used in the p resent experiment to expose all participants to both certain and uncertain reinforcement and to a repetition of their initial condition. This extension of Ono provides for a better test for an effect of history with certain reinforcement on prefer ence for free choice A between subjects manipulation was employed in the free choice terminal links of the concurrent chains schedule used in this experiment. For one group of participants, all the stimuli in the free choice terminal link remained visi ble for the entire duration of the FI 5s schedule used in that link. For the other group, one click on any stimulus in the free choice terminal link removed the other two stimuli and left only the selected stimulus visible for the rest of the FI 5s schedu le. This manipulation was included because differences in preferences observed across studies may in part be attributed to differen ces in the presentation of free choice visual stimuli. Ono (2000) hypothesized that if the visual stimuli that appear in the free choice terminal link function as conditioned reinforcers, the n more preference for a free choice link might be observed if these stimuli remain visible during the terminal link FI schedule than if they do not. A test of this hypothesis is importa nt because different studies have used different methods of presenting free choice terminal links. Comparing these methods across groups drawn from the same population and exposed to otherwise similar experimental conditions allow s for a better determinat ion of the influence, if any, of these procedural differences.

PAGE 42

42 Method Participants Setting, and Apparatus Thirty six student s (2 9 females and 7 males) at the University of Florida served as participants. Participants were recruited from a pool compri sed of students enrolled in General Psychology and they received course credit in exchange for participation. All participants completed the experimental task using the mouse and keyboard of a PC i n one of two rooms containing the PC a table, and a chair, and were alone in the room for the duration of the task T he task was presented via a program designed in Microsoft Visual Basic 6.0. Experimental Design choice was assessed using a concur rent chains procedure. An ABA reversal design was used to evaluate the effects of reinforcer probability on preference for free versus forced choice options. Each participant was exposed to both of two conditions in whic h reinforcement was either certain or probabilis tic; the Certain and Uncertain c onditions, respectively. P articipants were also divided into two groups (n=1 8 /group) in which the number of visual stimuli displayed after responding for an alternative in the free choice option differed. Experimental events and conditions are described in detail below. Before completing any trials, all participants were given a set of brief instructions describing the general structure of the experiment and telling them how to make responses. As a part of these instructions, they read the following statement: Clicking on the circle(s) on the screen may result in the delivery of points. The total number of points you have earned will be displayed on the screen. Your goal should be to try to earn as many points as you can. At the end of the session the total number of points you earned will be recorded by the researcher. Once all data from all participants have been collected, if you

PAGE 43

43 have earned the highest total number of points you will win a $50 gift card to the store of your choice. This statement was included to ensure that participants were sufficiently motivated to attend to the task and respond to earn points. Concurrent C hains Procedure A concurrent chains schedul e was in place in the choice trials of all conditions of the present experiment (as depicte d in Figure 3 1 and described in Experiment 1 ). In the initial links of the schedule two arrays of small approximately 1 green circles were shown on the computer screen. One array contained one circle and the other contained three circles. These arrays were placed above two clickable buttons labeled A and B. Button A was always on the left and button B was always on the right, but the positions of the circle arrays were quasi randomized across trials. That is, sometimes one circle appeared above button A and three circles appeared above button B and vice versa; neither array appeared above the same button more than two times in a row. A fixed ratio (F R) 1 response schedule was in place in the initial links: a single mouse click to either button A or B removed both buttons and the array that had appea red above the non chosen button, and constituted as a response for either the array with one circle or for the array with three circles. A click to the button associated with one circle initiated the forced choice terminal link and a click to the button associated with three circles initiated the free choice terminal link. In the forced choice terminal li nk, one circle was positioned randomly on the screen and a fixed interval (FI) 5s s chedule was in place; the first click to the circle af ter 5 seconds had elapsed removed the circle from the screen and ended the terminal link. In the free choice

PAGE 44

44 terminal link, three circles were positioned randomly on the screen (without overlapping) and an FI 5s schedule was in place A between subjects manipulation was in place in the free choice terminal link as described below. At the end of every terminal link a p oint was delivered, if scheduled. When a point was del displayed on the screen during the initi al link of every trial and continued to accumulate across conditions. An inter trial interval (ITI) occurred after every terminal link, whether or not a point was delivered. During the ITI the screen remained blank for 5 seconds. At the end of the ITI, a new trial began. Each condition ended when 28 choice trials were completed. Once all the trials in a condition were completed, a new set of brief instructions appeared on the screen to remind the participant how to complete the task and to begin the next condition. When all three conditions (either Certain Uncertain Certain or Uncertain Certain Uncertain) were completed and before they left the room, participants responded to a brief questionnaire consis ting of six questions. These questions are lis ted in Table 3 1. Within and Between S ubjects Manipulations Within subjec t conditions. All participants were exposed to both the Certain and Uncertain conditions. In the Certain condition, responding on a circle after the FI 5s schedule elapsed always resulted in the delivery of a point (1.0 probability of reinforcement) In the Uncertain condition, a point was delivered at a probability of 0.5. Between subjects manipulation. A between subjects manipulation was in place in the free choice terminal l ink For half of the participants, all three of the circles

PAGE 45

45 remained available for the entirety of the FI 5s schedule and the first click on any of the three circles after the elapse of the FI schedule removed all three circles and terminated the link ( Al l Stimuli arrangement) For the other half of the participants, the first click to any of the three circles in the free choice terminal link (before the FI 5s schedule elapsed) caused the other two to disappear. Therefore, only one circle remained availa ble for the rest of the FI 5s schedule and the first click to this remaining circle after the elapse of the FI schedule removed that circle and terminated the link ( Single Stimulus arrangement) Practice Exposure Trials At the beginning of each condition and before choice trials began, participants completed a set of 12 practice exposure trials. These trials were included to guarantee that participants would experience the contingencies that were in place in the f ree and f orced choice terminal links in t he subsequent choice trials. During these trials, only and either one or three circles appeared above the button. One mouse click to that button initiated eit her a f ree or f orced choice terminal link. The contingencie s in the terminal links of the practice exposure trials were the same as those in the choice trials. That is, FI 5s schedules were in place and points were del ivered at a probability of either 1. 0 or 0.5 (if the practice exposure trials preceded Certain or Uncertain choice trials, respectively). In addition, the visual stimuli in the free choice terminal links were presented as in either the Single Stimulus arrangement or the All Stimuli arrangem ent depending on group assignment. Each set of practice exposure trials consisted of six forced choice trials and six free choice trials with their order quasi randomized.

PAGE 46

46 Data Analysis Data from individual participants and data avera ged across participants were analyzed in the same manner as in Experiment 1. Results The results of Experiment 2 are presented in Figure 3 2. P roportion of free choice selections is graphed as a function of condition. This proportion was calculated by di viding the sum of free and forced choice selections by the total number of free choice selections. Bars indicate the mean proportion s of free choice selections for all participants in each condition and individual proportions are plotted as data points. The left panels show data from participants in the All Stimuli group and the right panels show data from the Single Stimulus group; data in the top row are from participants with the order Certain Uncertain Certain and data in the bottom row are from part icipants with the order Certain Uncertain Certain. Visual analysis of these graphs indicated only a few differen ces between the mean proportion s of free choice selections across the Certain and Uncertain conditions within and across condition order, for both the All Stimuli and Single Stimulus groups. F or participants who experienced the Certain condition first (top row in Figure 3 2) mean proportion of free choice selections in the first Certain condition was higher in the Single Stimulus group compare d to the All Stimuli group. In the subgroups of the All Stimuli group, the mean proportion of free choice selections was higher in the first Uncertain condition compared to the first Certain condition (top and bottom left panels in Figure 3 2). The mean proportion of free choice selections was higher on average in the Uncertain condition compared to both Certain conditions in both term inal link arrangements (top row). For participants with the order Uncertain Certain Uncertain, no

PAGE 47

47 differences between con ditions were observed in the All Stimuli group. For the Single Stimulus group, the mean proportion of free choice selections was higher in the Certain condition compared to both Uncertain conditions. Overall, mean proportions were near or below 0.5 for n early all conditions in both groups. Analyses were conducted to reveal statistically significant results. C ondition m eans and standard deviations for all conditions are shown in Table 3 2. A univariate ANOVA did not reveal significant main effects of c ondition, terminal link arrangement, or condition order, and no significant interaction between condition and terminal link arrangement was indicated. The only comparison that approached st atistical significance was the independent samples t test between the first Certain condition ( M = .686 SD = 345 ) and first Uncertain condition ( M = 945 SD = 214 ) across the subgroups of the All Stimuli group t (16) = 1. 905 p = .0 75 A two tailed t test using non transformed data indicated that the only condition in which the mean proportion of free choice selections was statistically larger than 0.5 was the first Uncertain condition for the participants with the order Uncertain Certain Uncertain in the All Stimuli group ( M = .639, SD = .174), t ( 8 ) = 2 .386, p = .0 44. The lack of statistically significant results in the cases in which visual analysis determined a difference between conditions was likely due to the large degree of between subject variability observed in most of the experimental conditions across c ondition order and group. This variability was evident by the wide spread of individual proportions as reflected in the stan dard deviations shown in Table 3 2 and the spread of individual data points in Figure 3 2.

PAGE 48

48 Discussion The results of the present experiment do not replicate previous findings in which differences in reinforcer certainty produced differences in preference for free choice. Ono (2000) observed that the proportion of free choice selections by pigeons that experienced a condition of unc ertain reinforcement first were high. The mean proportion of free choice selections for participants in the present experiment who experienced t he Uncertain condition first was close to 0.5 However, the proportion of free choice selections was greater t han 0.5 for eight out of nine participants who had Uncertain first in the All Stimuli group compared to only three out of nine in the Single Stimulus group. These data suggest that preference for free choice may be higher when response alternatives remain available in the free ch oice terminal link an d reinforcement is uncertain. Ono (2000) also found that proportion of free choice selections decreased from the uncertain to the certain condition. The results of this experiment do not replicate these find ings. Proportions of free choice selection decrease d from Uncertain to Certain in only 7 of 18 of the participants in both terminal link group s. Most p articipants in the All Stimuli group were indifferent in the Certain condition tha t followed Uncertain, but the degree of decrease was not close to that observed by Ono. This is primarily because the mean proportion of free choice selections was already close to 0.5 in the initial Uncertain condition in that group (even if it was statistically significantl y higher than 0.5) In addition, six out of nine of the participants in the Single Stimulus group showed higher preference for free choice in the Certain condition that followed Uncertain (bottom right panel of Figure 3 2). Post s ession questionnaires re veal that these participants perceived that the free choice option provided more points than the forced

PAGE 49

49 choice option (although this was not scheduled to be the case). This perception may be the result of having just experienced a condition in which reinf orcement was uncertain; consistently preferred even though its outcomes were not better than the other. This does not explain why participants in the All Stimuli group would no t also prefer free choice when reinforcement was certain. Participants who experienced the Certain condition first (top row of Figure 3 2) were, on average, indifferent between the free and forced choice options, though the ranges of individual data indi cate a large degree of between subject variability. Seven out of eight participants in the Single Stimulus group had proportions of free choice selections above 0.5 in the initial Certain condition. On average, proportions of free choice selections were close to indifference, but indifference between forced and free choice was not consistently observed across participant. Ono (2000) observed that subjects that experience certain reinforcement first remained indifferent when exposed to uncertain reinforc e ment. Although mean proportion s of free choice selection s did not differ from each other between the initial Certain and Uncertain conditions for either terminal link group, the individual data reveal a great degree of between subject variability. In fa ct, 11 out of 18 participants who experienced Certain first had a higher preference for free choice in the Uncertain condition that followed Certain. The main finding in this experiment was the large degree of between subject variability in proportion of free choice selections. There were some patterns observed in the individual data. For example, for 7 out of 18 the participants in both terminal link arrangement group s who experienced Certain first, preference increased from Certain

PAGE 50

50 to Uncertain and th en decreased again from Uncertain to the Certain replication. For h alf the participants in the Single Stimulus group preference increased from Uncertain to Certain and then decreased from Certain to the Uncertain replication. However, these patterns wer e not consistent across all participants and the actual proportions of free choice selection differed greatly across participant s in all subgroups making it difficult to draw conclusions about the effect of reinforcement certainty at the group level. A large number of participants reported on the post session questionnaires that they were seeking patterns or tricks to figur e out how th ey should respond to the task. guided their behavior and made them insensitive to the experimentally arranged contingencies. As in Experiment 1, it is possible that the use of arbitrary stimuli made it instruction rather than the experimental task. Future research with adult human participants might be more successful in demonstrating an effect of reinforcer certainty on free choice preference if real or simulated response alternatives (e.g. dis hes on a menu, products in a grocery store, etc.) are used

PAGE 51

51 Table 3 1. Experiment 2 : Post session questionnaire contents Question # Question text 1 When you had to choose between two arrays of green circles, which of the arrays did you choose more oft en, the one with one circle or the one with more than one circle? Why? 2 When there was more than one circle available, did you think there were enough to choose from? Why or why not? 3 In the case where there were more than one circle to click, did you like the fact that clicking on one made the rest go away? Why or why not? 3b In the case where there were more than one circle to click, did you like the fact that they all remained available when you were clicking them? Why or why not? 4 Sometimes yo u always earned a point for clicking, and sometimes you didn't. Did your preference for one circle vs. more than one circle change depending on whether you always got a point or not? Why or why not? 5 Did you enjoy earning points? 6 Is there anything e lse you would like to say about this study? Table 3 2. Experiment 2 : Mean proportions of free choice selections (non transformed) for the All Stimuli and Single Stimulus groups in each condition for both condition sequences. Group Condition M All St imuli Certain .446 (.254) Free Choice Uncertain .603 (.309) Certain rep .492 (.347) Uncertain .639 (.175)* Certain .598 (.210) Uncertain rep .576 (.275) Single Stimulus Certain .517 (.304) Free Choice Uncertain .576 (.299) Certain re p .532 (.415) Uncertain .487 (.186) Certain .599 (.383) Uncertain .500 (.282) Note: Standard deviations are in parentheses. Asterisks indicate that the mean proportion is significantly larger than 0.5 (p < = .05).

PAGE 52

52 Figure 3 1. Schematic of the concurrent chains schedule in place during choice trials for all conditions in Experiment 2 In the initial links, only buttons A or B could be clicked and the positions of the arrays of 1 or 3 circles were quasi randomized. In this example, Button A represents the forced choice option and Button B represents the free choice option. In the terminal links, clicks on the circles following the termination of the FI 5s schedule produced a point at a probability of either 1.0 (C ertain condition) or 0.5 (Uncertain condition). Terminal link Initial link Forced Choice Free Choice FI 5s 1 pt delivered at 1.0 or 0.5 probability FI 5s 3s ITI FR 1 FR 1 A B

PAGE 53

53 Figure 3 2. Proportion of free choice selections in the Certain and Uncertain conditions in Experiment 2 Bars indicate proportions averaged across participant; data points ind icate proportions from individual participants The dashed lines are reference lines set at 0.5. Data from the All Stimuli and Single Stimulus group s are in the top and bottom left and top and bottom right panels, respectively The top row shows data fr om participants with the condition order Certain Uncertain Certain and the bottom row shows data from participants with the condition order Uncertain Certain Uncertain. Proportion Free Choice Selections Condition All Stimuli Single Stimulus

PAGE 54

54 CHAPTER 4 EXPERIMENT 3 : AMOUNT OF REINFORCEMENT Researchers have hypothesized that a hi story of differential outcomes associate d with free choice can lead to preference for free choice even when outcomes between free and forced choice options are the same. However, few studies have directly tested this hypothesis. Skowronski & Carlston (19 82) did so to a limited extent. They designed a pill dispenser that was automatically or ostensibly manually controlled. When the dispenser was on auto, either sweet or bitter edible chewable pills were dispensed to participants from two bottles without control by the participants When the machine was switched to manual, a left and a right switch allow ed participants to the left and right bottles contained a mixture of pills and the pil l type that was dispensed was predetermined. Participants were exposed to both auto (no choice) and manual (choice) conditions but the outcomes of those conditions were manipulated across participant. When asked whether they wanted auto or manual contro l in two final choice trials, the participants whose experience with choice was associated with better outcomes reported higher preference for the manual (choice) option. Karsina, Thompson, and Rodriguez (2011) arranged a concurrent chains schedule in w hich free choice terminal link selections were differentially reinforced Participants experiencing this condition were more likely to choose the free choice link in subsequent conditions in which free and forced choice links produced the same reinforceme nt. Both of these studies suggest that preference for free choice may be a result of a history of better outcomes as a result of choosing a free choice option.

PAGE 55

5 5 Skowronski & Carlston (1982) and Karsina et al. (2011) also included conditions in which the forced choice option provided a better outcome and they found that this experience produced a decrease in preference for the free choice option ( though it should be noted that the carry over from the condition in Karsina et al. in which selection of the f orced choice option was differentially reinforced was less robust than the carry over from the condition in which free choice selection was differentially reinforced). Additionally, Fisher, Thompson, Piazza, Crosland, and Gotjen (1998) found that their p articipant deferred a choice of reinforcers when a no choice option was associated with the delivery of more highly preferred reinforcers. These results and those cited above indicate that preference for a free or forced option can be altered by presentin g better outcomes in either the fre e or forced choice option. They also suggest that a history of better outcomes for either free or forced choice may influence preference for those options in subsequent conditions in which the outcomes provided in both o ptions is the same. The present experiment extends this previous research by using the amount of reinforcement available in the free and forced choice options as the independent variable. Although Skowronski & Carlston (1982) show ed outcomes, they did not provide their subjects with repeated choices between free and and they did not manipulate the outcomes in the free and forced choice options within subje ct. A within subject comparison m ay better demonstrate control by the independent variable as shifts in preferences within the same participant may be a more compelling measure of the effect of the independent variable. Karsina et al. (2011) use d a with in subject

PAGE 56

56 design but they also had a few methodological problems First the condition in which free choice selection was differentially reinforced was the first condition following baseline for all participants This prior exposure may have contributed to the less robust preference observed for forced choice following the conditions in which forced choice selection was differentially reinforced. Furthermore, a subtle procedural point may make a large difference when extrapolating the results of this st udy. Specifically, the actual dimensions of reinforcement delivered in the free and forced choice options were always the same (1 point). The experimenters manipulated the schedule of reinforcement, but not the outcome of selecting free or forced choice per se. That is, though reinforcement was delivered on a denser schedule when either free or forced choice selections were differentially reinforced, the features of the reinforcer delivered in both options were identical. Therefore it c ould be argued th at neither the free or forced choice option provided a differential outcome; that is, when it was delivered, reinforcement in both options was of the same quality. The present experiment was designed to address some of these procedural issues. First, al l participants were exposed to conditions in which the outcomes delivered in the free and forced choice options were the same, a condition in which outcomes were better in the free choice option and a condition in which outcomes were better in the forced choice option. Additionally, the order of exposure to these conditions was manipulated across participants such that some were exposed to the condition first. The present experiment was designed to observe carry over effects resulting from exposure to better outcomes in both the free and forced choice options.

PAGE 57

57 The outcomes provided in the free and forced choice options were determined by manipulating the amount of reinforcement that was available contingent on responding to stimuli in the free and forced choice options. Therefore, choosing either the free or forced choice option in these conditions resulted in a qualitatively different reinforcer, not the same rei nforcer delivered on a different schedule. This manipulation may be a more direct analogy to the types of histories that are proposed to influ ence preference for free choice that is, that free choice often allows for a selection of a qualitatively bette r or more preferred alternative An additional between subjects manipulation was used in the conditions in which the outcomes in the free and forced choice options were the same. For one group of participants, the alternatives available in the free and forced choice options in this condition were all identical. For the ot her group, the alternatives in the free choice option differed with respect to amount of reinforcement available, but the highest amount of reinforcement was available in both the free and forced choice options. This manipulation was conducted in order to determine whether the alternatives from which a selection is made would have any influence on preference for free versus f orced choice when neither of those options produces a better o utcome than the other Method Participants, Setting, and Apparatus Forty students (21 females and 19 males) at the University of Fl orida served as participants. Participants were recruited from a pool comprised of students enrolled in General Psycholog y and they received course credit in exchange for participation. experimental task using the mouse and keyboard of a PC in one of two rooms

PAGE 58

58 containing the PC a table, and a chair, and were alone in the room for the duration of the task. T he task was presented via a program designed in Microsoft Visual Basic 6.0. Experimental Design concurrent chains procedure. An ABACA reversal design was used to evaluate the effects of reinforcer amount on preference for free versus forced choice options. Each participant was exposed to three conditions in which the amount of reinforce ment delivered in the free an d forced choice options was equal, higher in the f ree choice option, or higher in the f orced choice option; the Baseline, Free Better, and Forced Better conditions, respectively. P articipants were also divided into two groups (n=20 /group) in which the con figuration of stimuli in the free and forced choice options in the Baseline condition s differed. Experimental events and conditions are described in detail below. Before completing any trials, participants were given a set of brief instructions describ ing the general structure of the experiment and telling them how to make responses. As a part of these instructions, they read the following statement: Clicking on the rectangles(s) left on the screen may result in the delivery of points. The rectangl es will be different colors (red, green, blue, yellow) and each color corresponds to the number of points you can earn by clicking on the rectangles. The total number of points you have earned will be displayed on the screen. Your goal should be to try to earn as many points as you can. PLEASE NOTE: You may have to click a rectangle more than once before a point is earned. At the end of the session the total number of points you earned will be recorded by the researcher. Once all data have been collected if you have earned the highest total number of points you will win a $50 gift card to the store of your choice. This statement was included to ensure that participants were sufficiently motivated to attend to the task and respond to earn points.

PAGE 59

59 Concur rent C hains Procedure A concurrent chains schedule was in place in the choice trials of all conditions of the present experiment (as depicted in Figure 4 1 and described in Experiments 1 and 2 ). In the initial links of the schedule, two arrays of small, 0. rectangles were shown on the computer screen. One array contained one rectangle and the other contained three rectangles The colors of these rectangles were yellow, blue, green, or red and the method by which these colors were determined is desc ribed below. The rectangle arrays were placed above two clickable buttons labeled A and B. Button A was always on the left and b utton B was always on the right, but the positions of the rectangle arrays were quasi randomized across trials. That is, somet imes one rectangl e appeared above b utton A and three rectangles appeared above b utton B and vice versa; neither array appeared above the same button more than two times in a gles were constantly visible along the top of the screen during the initial links. These rectangles number of points shown depending on the color next to which the la bel was positioned: yellow = 5 pts, blue = 4 pts, green = 3 pts, and red = 2 pts. A fixed ratio (FR) 1 response schedule was in place in the initial links: a single mouse click to either button A or B removed both buttons and the array that had appeared above the non chosen button, and constituted as a response for either the array with one rectangle or for the array with three rectangles A click to the button associated with one rectangle initiated the forced choice terminal link and a click to the bu tton associated with three rectangles initiated the free choice terminal link. In the forced choice terminal link, one rectangle was positioned randomly on the screen and a

PAGE 60

60 fixed interval (FI) 5s schedule was in place. T he first click to the rectangle af ter 5 seconds had elapsed removed the rectangle from the screen and ended the terminal link. In the free choice terminal link, three rectangles were positioned randomly on the screen (without overlapping) and an FI 5s schedule was in place All three of t he rectangles remained visible for the entirety of the FI 5s schedule and the first click on any of the three rectangles after the elapse of the FI schedule removed all rectangles from the screen and terminated the link. At the end of every terminal lin k points were delivered with a probability of 1.0 When point s w ere del earned [ X ] point s the screen and the amount of point s earned The number of points earned depended on the color rectangle that was chosen as described below. The point total was always displayed on the screen during the initial link of every trial and continued to accumulate across conditions An inter trial interval (ITI) occurred after eve ry point delivery During the ITI the screen remained blank for 3 seconds. At the end of the ITI, a new trial began. When a condition ended, a new set of brief instructions appeared on the screen to remind the participant how to complete the task and to begin the next condition. When all five conditions (either Baseline Free Better Baseline 2 Forced Better Baseline 3 or Baseline Forced Better Baseline 2 Free Better Baseline 3 ) were completed and before they left the room participants responded to a brie f questionnaire consisting of four questions. These questions are listed in Table 4 1. Within S ubject Conditions Baseline condition All participants completed three replications of the Baseline condition. Each Baseline condition consisted of 24 choice trials. In all of these trials,

PAGE 61

61 the highest number of points available in the free and forced choice options was held constant. The method by which the stimuli in the Baseline condition s were arranged in order to keep amount of reinforcement the same in both options differed across group and these methods are described in more detail below. Free Better condition The Free Better condition consisted of 24 choice trials. In this condition, only the free choice option contained the alternative associa ted with the highest number of points That is, of the three rectangles in the free choice option, one was always yellow (worth five points) and the other two rectangles were two different colors: either blue (four points), green (three points), or red (t wo points). The colors of the non yellow rectangles were programmed such that each participant was exposed to each possible combination of yellow plus two other colors eight times. The same color did not always appear in the same position on the screen i n the free choice initial link arrays and terminal links. Forced Better condition The Forced Better condition consisted of 24 choice trials. In this condition, only the forced choice option contained the alternative associated with the highest number of points That is, the rectangle in the forced choice option was always yellow (worth five points). The three rectangles in the free choice option were three different colors: one each blue (four points), green (three p oints), and red (two points). The same color did not always appear in the same position on the screen in the free choice initial link arrays and terminal links. Between S ubjects Manipulation In the Baseline conditions, the highest amount of points available in the free and forced choice options was held constant, but the method by which the stimuli in the Baseline condition s were arranged differed across group. For one group (the Identical

PAGE 62

62 Colors group) all the rectangles in both the forced and free choice options were the same color. Therefore, the amount of points available was exactly the same no matter which alternative was chosen in either the free or forced choice option. Each color (yellow, blue, green, and red) was presented six times in quasi randomized order. For the other g roup (the Mixed Colors group) a yellow rectangle was present in both the forced choice and free choice option in all trials. Therefore, both options contained an alternative associated with the hig hest amount of points (five). The other two rectangles i n the free choice option were two different colors: either blue, green, or red (four, three, or two points, respectively), and each participant was exposed to each possible combination of yellow plus two other colors eight times. The same color did not al ways appear in the same position on the screen in the free choice initial link arrays and terminal links. Data Analysis Data from individual participants and data averaged across participants were analyzed in the same manner as in Experiments 1 and 2. Re sults The results of Experiment 3 are presented in Figure 4 2. Proportion of free choice selections is graphed as a function of condition. This proportion was calculated by dividing the sum of free and forced choice selections by the total number of free choice selections. Bars indicate the mean proportion s of free choice selections for all participants in each condition and individual proportions are plotted as data points. The left panels show data from participants in the Identical Colors group and the right panels show data from the Mixed Colors group; data in the top row are from participants with the order Baseline Forced Better Baseline Free Better Baseline and data in the bottom

PAGE 63

63 row are from participants with the order Baseline Free Better Basel ine Forced Better Baseline Visual analysis of these graphs indicated that selection of the free or forced choice option was directly affected by the presence of the alternative that provided the highest amount of points: when the highest amount of point s was only available in the free choice option, participants selected free choice more and when it was only available in the forced choice option, participants selected force d choice more. Mean proportion of free choice selections varied when the highest amount of points was available in both the free and forced choice options. There did not appear to be any carry over effects in the Baseline conditions that followed the Free Better and Forced Better conditions. Overall, mean proportions were higher in t he initial Baseline conditions compared to the second and third Baseline conditions. Mean proportions were also higher in the Forced Better and initial Baseline conditions in the Identical Colors group compared to the Mixed Colors group. Mean proportions of free choice selections were near or below 0.5 for all but one Baseline replication and all Forced Better conditions in both baseline arrangement groups. Mean proportions were higher than 0.5 for one initial Baseline condition in the Identical Colors g roup and all Free Better conditions in both groups. Analyses were also conducted to reveal statistically significant results. C ondition means and standard deviations are presented in Table 4 2. A univariate ANOVA indicated significant main effects of c ondition, F (1, 148) = 523.20, p < .001 and b aseline arrangement, F (1, 36) = 4.145, p = .049. As evident in Figure 4 2, mean proportion of free choice selection s was directly driven by the availability of the highest amount of points ( looking at the Free Better and Forced Bette r conditions). Mean

PAGE 64

64 proportion of free choice selection s did not differ from Baseline 2 to Baseline 3 in either condition order in either b aseline arrangement group. M ean proportion of free choice selection s in each initial Baselin e condition was higher than the mean proportion collapsed across each subsequent Baseline 2 and 3 condition Independent samples t tests indicated that this difference was significant at the p = 0. 0 5 level in the following case s: the Identical Colors gro up who had Forced Better first ( M = .958 SD = .271 and M = 556 SD = 381 top left panel in Figure 4 2 ), t (28) = 2 .971 p = .006; the Mixed Colors group who had Forced Better first ( M = .678, SD = .210 and M = .404, SD = .355, top right panel in Figure 4 2), t (27) = 2.649 p = .013; and the Mixed Colors group who had Free Better first ( M = .709, SD = .194 and M = .459, SD = .350, bottom right panel in Figure 4 2), t (27.6) = 2.512 p = .018. The main effect of baseline arrangement was driven by differen ces in the Forced Better and initial Baseline conditions across baseline arrangement. No differences between initial Baseline conditions were observed within baseline arrangement group an d so these proportions were averaged and those means were compared across baseline arrangement. The mean proportion of free choice selections was significantly higher in the initial Baseline conditions in the Identical Color group ( M = 861 SD = .2 97 ) compared to the Mixed Colors group ( M = 693 SD = 1 97 ), t (38) = 2.1 08 p = .0 42 No differences were observed between the Forced Better conditions within baseline arrangement and so these proportions were averaged and those means were compared across baseline arrange ment. The mean proportion of free choice selections wa s significantly higher in the Forced Better conditions in the Identical Color group ( M = 280 SD = 275 ) compared to the Mixed Colors group ( M = 052 SD = 115 ), t ( 25.4 ) =

PAGE 65

65 3.43 p = .0 02 No differences were observed between Free Better conditions withi n or across baseline arrangement. A two tailed t test using non transformed data was conducted to determine whether mean proportions of free choice selections were significantly different from 0.5 in any conditions. The results of this test are p resented in Table 4 2 and indicate that several of these differences were statistically significant. Discussion The results of the present experiment indicate that amount of reinforcement has a direct e ffect on preference for free or forced choice. When the hig hest amount of reinfor cement was available in only the free or forced choice option preference shifted to that option. These results support those of previous studies in which differential outcomes for selecting a f ree or f orced choice option produced sh ifts in preference toward that option (Fisher et al., 1998; Karsina et al., 2010; Skowronski & Carlston, 1982; Suzuki, 1997). When amount of reinforcement was controlled (in the Baseline conditions) there was considerable between subject variability in p reference for free choice. Only 3 out of 20 participants in the Identical Colors group and 5 out of 20 participants in the Mixed Colors group showed a pattern in which preference for free choice increased in the Baseline 2 or 3 condition that followed Fr ee Better and decreased in the Baseline 2 or 3 condition that followed Forced Better. On average, p reference for free choice either stayed the sa me as or decreased from Baseline in both baseline arrangements groups in the Baseline 2 conditions that follow ed Forced Better. However, this same pattern was also observed in the Baseline 2 conditions that followed Free Better, indicating that the decreases following Forced Better were likely

PAGE 66

66 not due to carry over from the Forced Better condition. In addition, there were only 10 out of 40 participants for whom preference for free choic e increased compared to Baseline in the Baseline 2 or 3 conditions that followed Free Better. All of these results indicate that brief exposure to the Free Better and Forced Bette r conditions did not produce consistent effect s on free choice preference in the Baseline 2 and 3 conditions. The present experiment was designed to test the hypothesis that preference for free choice results from a history in which free choice options pro vided better outcomes. Such a history may result in the availability of multiple alternatives coming to function as a conditioned reinforcer F ree choice options may then be preferred even when they do not currently provide a better outcome than forced o r no choice options. The Free Better and Forced Better conditions of the present experiment were designed to simulate conditions under which a free or forced choice option would come to function as a conditioned reinforcer. The Baseline 2 and 3 condition s were implemented to test whether either option would continue to be preferred if it had been in a preceding Free Better or Forced Better condition, but there was no carry over of preference It is possible that the exposure to the Free Better and Forced Better conditions in this experiment was not long enough to allow the free or forced option s to come to function as conditioned reinforcer s ; i f the option s do not function as conditioned reinforcer s, then shifts in preference toward those options in condi tions in which outcomes are equal might not occur Overall, mean proportion s of free choice selections during all Baseline conditions w ere either near or below 0.5 indicating either indifference between free and forced choice or preference for forced cho ice. However, on average, there were differences

PAGE 67

67 between the initial Baseline conditions and Baselines 2 and 3 in all condition order subgroups in both b aseline arrangements. S ome carry over effects from the Free Better and Forced Better conditions were observed in Baseline conditions for some participants but 24 out of 40 participants showed an overall decrease in preference for free choice in Baselines 2 and 3 compared to their initial Baseline condition. These results differ from those of Skowronski and Carlston (1982) who found that participants preferred the option in which they had had success when they were told that past on later choice trials S ome partici pants in that study were told that past successes would not have an effect on later choice trials, and under these conditions, preference for the free and forced options were not different from each other. These results are more similar to those of the pr esent experiment. Skowronski and Carlston (1982) noted that perceptions of past control and future expectancies reduces the effects of past p. 700). In the present experiment, reinforcement was certain and outcomes were the same for free and forced choice selection s during all Baseline conditions. Though participants experienced conditions in which the free and forced choice options produced better outcomes, these conditions were easily dis criminable from those in which they provided the same outcomes (e.g. free choice stimuli were different). Participants also had initial exposure to the Baseline condition in which outcomes between free and forced choice were t he same. For these reasons, participants may have quickly learned that neither the f ree or f orced choice option produced a better outcome during Baseline conditions This experience would account

PAGE 68

68 for indifference between the free and forced choice option s C onducting this experime nt using uncertain reinforcement might make the discrimination between Baseline and the other two conditions less clear and might lead to some carry over from the Free Better and Forced Better conditions and/or higher preference for free choice during Bas eline conditions Reponses to p os t session questionnaires reveal some other reasons why preferences were near or below 0.5 In the Mixed Colors group, the forced choice option always contained the yellow rectangle that was worth the highest amount of po ints (except in the Free Better condition). Responses by some participants in this group suggest that seeing the yellow rectangle on its own in the forced choice option so often may have given an impression that the forced choice option was always the bet ter one, even during Baseline conditions in which the yellow rectangle was also pres ent in the free choice option. In addition, several participants reported that they preferred the forced choice option because having the yellow rectangle in isolatio n mad e it easier to identify the best alternative In the Identical Colors group, many participants reported that they preferred the forced choice option because having only one rectangle in the terminal link reduced the effort required to produce points. The se responses suggest previous exposure to the Free Better and Forced Better conditions but rather other features of the procedure Statistical tests indicated a main effect of baseline arrangement that was driven by the differences between the initial Baseline and Forced Better conditions across baseline arrangement groups. These results are similar to those of Suzuki (1997) who

PAGE 69

69 compared preferences between free and forced choice options using conditions in which the free choice option contained two or three identical stimuli (all worth 10 points) and in which the free choice option contained one stimulus worth 10 points and two or three other stimuli with a lower poi nt value. Suzuki found that preference for free choice was higher in the conditions in which identical stimuli were available in the free choice are support ed by the findings in the present experiment. In the present experiment, t he free choice options in the Baseline conditions in the Mixed Colors group contained one rectangle worth the highest amount of points and two other stimuli with lower values. The free choice option in the Baseli ne conditions in the Identical C olors group contained three identical rectangles. Based on findings of their own study, Hayes, Kapust, Leonard, Rosenfarb (1981) observed about equal then prob ability of selecting A given choice of A and B is nearly zero then the organism will p. 6) In the context of the present experiment, p articipants in the Mixed Colors group preferred the yellow rectangle alone (forced choice) to a choice between the yellow rectangle and two lower valued rectangles (free choice) Responses to the post session questionnaires also provide reasons why the Mi xed Colors group may have preferre d the forced choice option in the Baseline and Forced Better conditions more than the Identical Colors group As stated above, some Mixed Colors more exposure to the yellow rectangle in isolation may have given an illusion that the forced option was always the

PAGE 70

70 better choice Participants i n the Identical Colors group were only exposed to one condition in which the forced choice option alone contained the yellow rectangle and in which lower valued rectangles appeared in the free choice option; participants in the Mixed Colors group were exposed to four conditions in which the yellow rectangle was in the forced choice option and at least two lower valued rectangles were in the free choi ce option. However, this amount of exposure was not present for the Forced Better conditions that directly followed the initial Baseline conditions It is possible that for the Mixed Colors group, the initial Baseline exposure to the single yellow rectan gle in forced choice and lower valued rectangles in free choice was enough to shift preference more strongly toward forced choice compared to the Identical Colors group who did not have this initial exposure. As in Experiments 1 and 2, a major finding of the present experiment was the great deal of between subject variability with regard to free choice preference, particularly in the Baseline conditions. The range of initial Baseline prefere nce was wide, suggesting that participants likely entered the stu dy with different pre experimental preferences for free choice. Just as was the case in Experiments 1 and 2 the present experiment used arbitrary stimuli and i t is possible that any relevant history that may otherwise contribute to free choice preferenc e was not generalizable to these experimental conditions. Data and responses on post session questionnaires also suggest that the arbitrary stimuli and certain reinforcement may have caused behavior to come under the control of variables other than those being controlled in the experiment. Some participants provid ed responses that suggested their behavior was controlled by perceived Other participants indicated

PAGE 71

71 that they were indifferent between the free and forced choic e options because they (accurately) perceived that neither option was better or that they chose arbitrarily or in an alternating pattern because they were bored with the task. Future attempts to evaluate the effects of exposure to differential outcomes in free and forced choice options can be improved in several ways. It may be more likely to rather than the arbitrary stimu li. F uture studies might also find more robu st results if choices are made among reinforcers rather than response alternatives. Fenerty and Tiger (2010) conducted a study in which four participants were given choices between a choice of tasks, a choice of reinforcers, no choice of either, and a con trol (in which reinforcement was not delivered). The y found that non e of the participants preferred choice of tasks over no choice of tasks, but three of four exhibited their highest prefere nce for choice of reinforcers. In addition, the fact that reinfo rcement was certain in the present experiment may have made it eas ier to learn that neither free nor forced P reference for free choice and/or carry over from experience with better outcomes c ould be more likely if reinforcement is uncertain and it is not as easy to learn that free and forced choice outcomes are the same. Finally, if the goal of an experiment is to evaluate carry over effects of exposure to conditions in which free or forced choice o ptions provide better outcomes, longer exposure to those conditions may be necessary.

PAGE 72

72 Table 4 1. Experiment 3: Post session questionnaire contents Question # Question text 1 When you had to choose between the two arrays of rectangles, which of the arr ays did you choose more often, the one with one rectangle or the one with three rectangles? Why? 2 In the case of the array of three rectangles, did you think that three rectangles were enough to choose from? Why or why not? 3 Did you enjoy earning poin ts? 4 Is there anything else you would like to say about this study? Table 4 2 Experiment 3: Results of two tailed t test evaluating whether mean proportions of free choice selections (non transformed) are different from 0.5. Group Condition M t Sig. Identical Colors Baseline .638 (.198) 2.192 .056 Free Choice Forced Better .179 (.199) 5.089 .001* Baseline2 .288 (.252) 2.667 .026* Free Better .975 (.079) 19.000 .000* Baseline3 .396 (.308) 1.071 .312 Baseline 488 (.269) 0.147 .887 Free Better .963 (.060) 24.222 .000* Baseline2 .304 (.304) 2.040 .072 Forced Better .071 (.168) 8.084 .000* Baseline 3 .333 (.354) 1.491 .170 Mixed Colors Baseline .404 (.189) 1.599 .144 Free Choice Forced Bet ter .025 (.053) 28.500 .000* Baseline2 .183 (.201) 4.994 .001* Free Better 1.00 (.000) --** Baseline3 .271 (.292) 2.480 .035* Baseline .429 (.184) 1.215 .255 Free Better .963 (.093) 15.715 .000* Baseline2 .258 (.249) 3.068 .013* Forced Better .004 (.013) 119.000 .000* Baseline3 .258 (.272) 2.809 .020* Note: Standard deviations are in parentheses. Degrees of freedom were 9 for every condition. *T he mean p roportion is significantly higher or lower than 0.5 (p < = .05). **A statistical test could not be performed.

PAGE 73

73 Figure 4 1. Schematic of the concurrent chains schedule in place during choice trials for in the Mixed Color s Baseline conditions in Experiment 3. In t he initial links, only buttons A or B could be clicked and the positions of the arrays of 1 or 3 rectangles were quasi randomized. In this example, Button A represents the forced choice option and Button B represents the free choice option. In the termin al links, clicks on the circles following the elapse of the FI 5s schedule produced points at a probability of 1.0. The number of points earned depended on the color of the rectangle that was selected in the terminal link. Initial link Terminal link Forced Choice Free Choice FI 5s 5, 4, 3, or 2 pts delivered at 1. 0 probability FI 5s 3s ITI FR 1 FR 1 A B

PAGE 74

74 F igure 4 2. Proportion of free choice selections in the Baseline, Forced Better and Free Better conditions in Experiment 3 Bars indicate proportions averaged across participant; data points indicate proportions from individual participants Data from th e Identical Colors and Mixed Colors group s are in the top and bottom left and top and bottom right panels, respectively The top row shows data from partic ipants with the order Baseline Forced Better Baseline Free Better Baseline and the bottom row shows data from participants with the condition order Baseline Free Better Baseline Forced Better Baseline Proportion Free Choice Selections Condition Identical Colors Mixed Colors

PAGE 75

75 CHAPTER 5 EXPERIMENT 4 : PROBABILITY OF REINFORCEMENT This experiment was conducted as a companion to Experiment 3 and therefore its design had, in general the same rationale and design difference between these two experiments was the dimension of reinforcement along choice options provided e ither the same or a better probability of reinforcement delivery. Method Participants, Setting, and Apparatus Forty students ( 32 females and 8 males) at the University of Florida ser ved as participants. Participants we re recruited from a pool comprised of students enrolled in General Psychology and from two sections of Applied Behavior Analysis and they 18 to 22 years. All participants completed the experimental task using the mouse and keyboard of a PC in one of two rooms containing the PC a table, and a chair and were alone in the room for the duration of the task T he task was presented via a program designed in Microsoft Visual Basic 6.0. Experimental Design Pa concurrent chains procedure. An ABACA reversal design was used to evaluate the effects of reinforcer probability on preference for free versus forced choice options. Each participa nt was exposed to three conditions in which the probability of r einforcement delivered in the free and forced choice options was equal, higher in the free choice option, or higher in the forced choice option; the Baseline, Free Better, and

PAGE 76

76 Forced Better co nditions, respectively. P articipants were also divided into two groups (n=20/group) in which the configuration of stimuli in the free and forced choice options in the Baseline condition s differed. Experimental events and conditions a re described in detai l below. Before completing any trials, participants were given a set of brief instructions describing the general structure of the experiment and telling them how to make responses. As a part of these instructions, they read the following statement: Cl icking on the rectangles(s) left on the screen may result in the delivery of points. The rectangles will be different colors (red, green, blue, yellow) and each color corresponds to the probability of earning 5 points by clicking on the rectangles. For e xample, one of the colors may correspond to a 0.75 probability of earning points. This means that there is a 75% chance that clicking that rectangle will result in the delivery of points. The total number of points you have earned will be displayed on the screen. Your goal should be to try to earn as many points as you can. PLEASE NOTE: You may have to click a rectangle more than once before a point is earned. At the end of the session the total number of points you earned will be recorded by the researc her. Once all data have been collected, if you have earned the highest total number of points you will win a $50 gift card to the store of your choice. This statement was included to ensure that participants were sufficiently motivated to attend to the t ask and respond to earn points. Concurrent C hains Procedure A concurrent chains schedule was in place in the choice trials of all conditions of the present experiment (as depicted in Figure 5 1 and described in Experiments 1 2, and 3 ). In the initial li nks of the schedule, two arrays of small, rectangles were shown on the computer screen. One array contained one rectangle and the other contained three rectangles. The colors of these rectangles were yellow, blue, green or red and the method by which these colors were determined is described below. The rectangle arrays were placed above two clickable buttons labeled A and B. Button A

PAGE 77

77 was always on the left and b utton B was always on the right, but the positions of the rectangle arrays were qu asi randomized across trials. That is, sometimes one rectangle appeared above b utton A and t hree rectangles appeared above b utton B and vice versa; neither array appeared above the same button more than two times in a row. Additionally, four smaller, 0.4 blue, green and red rectangles were constantly visible along the top of the screen during the initial links. These rectangles probability of earning p oints shown depending on the color next to which the label was positioned: yellow = 0.8, blue = 0.6, green = 0.4, and red = 0.2. A fixed ratio (FR) 1 response schedule was in place in the initial links: a single mouse click to either button A or B remov ed both buttons and the array that had appeared above the non chosen button, and constituted as a response for either the array with one rectangle or for the array with three rectangles. A click to the button associated with one rectangle initiated the fo rced choice terminal link and a click to the button associated with three rectangles initiated the free choice terminal link. In the forced choice terminal link, one rectangle was positioned randomly on the screen and a fixed interval (FI) 5s schedule was in place. The first click to the rectangle after 5 seconds had elapsed removed the rectangle from the screen and ended the terminal link. In th e free choice terminal link, three rectangles were positioned randomly on the screen (without overlapping) and an FI 5s schedule was in place. All three of the rectangles remained visible for the entirety of the FI 5s schedule and the first click on any of the three rectangles after the elapse of the FI schedule removed all rectangles from the screen and terminat ed the link.

PAGE 78

78 At the end of every terminal link five p oints were delivered with a probability that depended on the color of the rectangle that had been selected (as described below) When points were delivered, the statement earned 5 red briefly on the screen and five points were add point total. The point total was always shown during the initial link of every trial and continued to accumulate across conditions. An inter trial interval (ITI) occurred after eve ry terminal link, whether or not points were delivered. During the ITI the screen remained blank for 3 seconds. At the end of the ITI, a new trial began. Each condition ended when 24 trials were completed. When a condition ended, a new set of brief i nstructions appeared on the screen to remind the participant how to complete the task and to begin the next condition. When all five conditions (either Baseline Free Better Baseline 2 Forced Better Baseline 3 or Baseline Forced Better Baseline 2 Free Better Baseline 3 ) were completed and before they left the room, participants responded to a brief questionnaire consisting of four questions These questions are listed in Table 5 1. Within S ubject Conditions Baseline condition. All participants completed thre e replications of the Baseline condition. Each Baseline condition consisted of 24 choice trials. In all of t hese trials, the highest probability of earning points in the free and forced choice options was held constant. The method by which the stimuli i n the Baseline condition s were arranged in order to keep probability of reinforcement the same in both options differed across group and these methods are described in more detail below. Free Better condition. The Free Better condition consisted of 24 choice trials. In this condition only the free choice option contained the alternative associated with

PAGE 79

79 the highest probability of earning points. That is, of the three rectangles available in the free choice option, one was always yellow ( 0.8 probabilit y of reinforcement ) and the other two recta ngles were two different colors: either blue ( 0.6 probability of reinforcement ) green ( 0.4 probability of reinforcement ) or red ( 0.2 probability of reinforcement ) The colors of the non yellow rectangles were p rogrammed such that each participant was exposed to each possible combination of yellow plus two other colors eight times. The same color did not always appear in the same position on the screen i n the free choice initial link arrays and terminal links Forced Better condition. The Forced Better condition consisted of 24 choice trials. In this condition, only the forced choice option contained the alternative associated with the highest probability of earning points That is, the rectangle in the forc ed choice option was always yellow ( 0.8 probability of reinforcement ) The three rectangles in the free choice option were three different colors: one each blue (0.6 probability of reinforcement) green (0.4 probability of reinforcement), and red (0.2 pro bability of reinforcement). The same color did not always appear in the same position on the screen in the free choice initial link arrays and terminal links Between S ubjects Manipulation In the Baseline conditions, the highest probability of earnin g points in the free and forced choice options was held constant, but the method by which the stimuli in the Baseline condition s were arranged differed across group. For one group (the Identical Colors group) all the rectangles in both the forced and fre e choice options were the same color. Therefore, probability of earning points was exactly the same no matter which alternative was chosen in either the free or forced choice option. Each color (yellow, blue, green, and red) was presented six times in qu asi randomized order.

PAGE 80

80 For the other group (the Mixed Colors group) a yellow rectangle was present in both the forced choice and fre e choice option in all trials. Therefore, both options contained an alternative associated with the highest probability of reinforcement (0.8). The other two rectangles in the free choice optio n w ere two different colors: blue green, or red (0.6, 0.4, or 0.2 probability of reinforcement, respectively), and each participant was exposed to each possible combination of yellow plus two other colors eight times. The same color did not always appear in the same position on the screen in the free choice initial link arrays and terminal links Data Analysis Data from individual participants and data averaged across participants w ere analyzed in the same manner as in Experiments 1 3. Results The results of Experiment 4 are presented in Figure 5 2. Proportion of free choice selections is graphed as a function of condition. This proportion was calculated by dividing the sum of free and forced choice selections by the total number of free choice selections. Bars indicate the mean proportion s of free choice selections for all participants in each condition and individual proportions are plotted as data points. The left panels show data from participants in the Identical Colors group and the right panels show data from the Mixed Colors group; data in the top row are from participants with the order Baseline Forced Better Baseline Free Better Baseline and data in the bottom row are fr om participants with the order Baseline Free Better Baseline Forced Better Baseline. Visual analysis of these graphs indicate d that selection of the free or forced choice option was directly affected by the presence of the alternative that provided the highest

PAGE 81

81 probability of earning points: when the highest probability of earning points was only available in the free choice option, participants selected free choice more and when it was only available in the forced choice option, participants selected fo rced choice more. However, the difference in mean proportion of free choice selections between the Free Better and Forced Better conditions depended on baseline arrangement : differences between these conditions were greater in the Mixed Colors group comp ared t o the Identical C olors group Mean proportions varied when the highest probability of earning points was available in both the fr ee and forced choice options. W ith one exception (bottom right panel) mean proportions of free choice selections did n ot differ between the three Baseline conditions in both condition orders in bot h baseline arrangement groups. There did not appear to be any carry over effects in the Baseline conditions that followed the Free Better and Forced Better conditions. Overall mean proportion of free choice selections was near or below 0.5 for all but two Baseline replications and all Forced Better conditions in both groups. Mean proportions were higher than 0.5 for two Baseline conditions in the Identical Colors group and al l Free Better conditions in both groups. Analyses were also conducted to reveal statistically significant results. C ondition means and standard deviations are presented in Table 4 2. A univariate ANOVA indicated significant main effects of condition, F (1, 148 ) = 14.533 p < .001 and baseline arrangement, F (1, 36) = 1.866 p = .0 04. This test also indicated a significant interaction between condition and baseline arrangement, F (4, 148) = 5.256, p = .001. As evident in Figure 4 2, mean proportion of fre e choice selections was directly driven by the availability of the highest probability of earning points (the Free Better and

PAGE 82

82 Forced Better conditions), but the degree of difference between the Free Better and Forced Better condition was larger in the Mixe d Colors group. Mean proportion of free choice selections in the Forced Better conditions did not differ wi thin baseline arrangement and so these proportio ns were averaged and those means were compared across baseline arrangement groups The mean proport ion of free choice selections was significantly higher in the Forced Better conditions in the Identical Color group ( M = 562 SD = 423 ) compared to the Mixed Colors group ( M = 225 SD = 240 ), t ( 30.1 ) = 3. 10 7, p = .00 4 No statistically significant dif ferences were observed between Free Better conditions within or across baseline arrangement. The main effect of baseline arrangement and the interaction between condition and baseline arrangement was also driven by differences between the Baseline condit ions within and across baseline arrangement group. Overall, no statistically significant differences were observed between the initial Baseline conditions in any condition order in either baseline ar rangement In addition, no statistically significant di fferences were observed between Baseline 2 and Baseline 3 in any condition order in either baseline ar rangement With one exception, the mean proportion of free choice selection s did not differ between any of the Baseline replications in eith er group. Th e difference between the initial Baseline ( M = 812 SD = .355 ) and the collapsed average of Baseline s 2 and 3 ( M = 426 SD = 372 ) for participants in the Mixed Colors group who had Free Better first (bottom right panel of Figure 5 2) was statistically s ignificant t ( 28) = 2.715 p = .01 1. Finally, comparisons were made betwee n the collapsed mean proportion of free choice selections of the initial Baselines an d the collapsed mean proportion of Baselines 2 and 3 across baseline arrangement The differenc e between

PAGE 83

83 initial Baseline c onditions was not significant, but the collapsed mean proportion of free choice selections in Baselines 2 and 3 was higher in the Identical Colors group ( M = 864 SD = 390 ) compared to the Mixed Colors group ( M = 514 SD = 3 46 ), t (78) = 4.250 p < .001. A two tailed t test using non transformed data was conducted to determine whether mean proportion of free choice selections was significantly different from 0.5 in any conditions. The results of this test are presented in Ta ble 5 2 and indicate that several of these differences were statistically significant. Discussion The results of the present experiment indicate that probability of reinforcement has a direct effect on preference for free or forced choice. When the high est probability of reinforcement was available in only free or forced choice, preference shifted to that option. These results support those of previous studies in which differential outcomes for selecting either forced or free choice resulted in shifts i n preference toward that option (Fisher et al., 1998; Karsina et al., 2010; Skowronski & Carlston, 1982; Suzuki, 1997). However, the difference in preference between t he Free Better and Forced Better conditions was affected by baseline arrangement. When probability of reinforcement was controlled (in the Baseline conditions) there was a great deal of between subject variability with regard to preference for free choice. Only a few participants (mostly in the Mixed Colors group) showed a p attern in which preferences increased in the Baseline condition that followed Free Better and decreased in the Baseline condition that followed Forced Better. In the Identical Colors group, no consistent effects of exposure to the Free Better or Forced Better condition s were observed. For 7 of 10 participants, p reference for free

PAGE 84

84 choice was higher in Baseline 2 than Baseline when Baseline 2 followed Free Better but the same number of participants also had higher preference for free choice in Baseline 3 when it followe d Forced Better. Only 5 of 10 participants showed lower preference for free choice in Baseline 2 following Forced Better and none of them had higher preference than Baseline in Baseline 3 following Free Better. In the Mixed Colors group, there were only 2 of 20 cases in which preferences increased from B aseline following Free Better and decreased from Baseline following Forced Better. All of these results suggest that brief exposure to the Free Better and Forced Better conditions did not produce any cons istent effect on free choice preference in the Baseline 2 and 3 conditions As was the case with Experiment 3, t he present experiment was designed to test the hypothesis that preference for free choice results from a history in which free choice options provided better outcomes. Such a history may result in the availability of multiple alternatives coming to functio n as a conditioned reinforcer. The Free Better and Forced Better conditions of the present experiment were designed to simulate conditions u nder which a free or forced choice option would come to function as a conditioned reinforcer. The Baseline 2 and 3 conditions were implemented to test whether either option would continue to be preferred if it had been in a preceding Free Better or Forced Better condition but there was no carry over of preference It is possible that the exposure to the Free Better and Forced Better conditions in this experiment was not long enough to allow either the free or forced option to come to function as a condit ioned reinforcer ; if the options do not function as conditioned

PAGE 85

85 reinforcers, then shifts in preference toward those options in conditions in which outcomes are equal might not occur Statistical tests indicated a significant main effect of baseline arran gement that was driven by the differences between the Forced Better and Baseline 2 and 3 conditions across baseline arrangement groups. Based on findings of their own study, Hayes et al. (1981) observed ice of A and B is about equal then the organism will prefer the choice of A and B over either ability of selecting A given choice of A and B is nearly zero then the p. 6) In the c ontext of the Baseline conditions in the present experiment, participants in the Mixed Colors group preferred the yellow rectangle alone (forced choice) to a choice between the yellow rectangle and two lower valued rectangles (free choice) Similarly, a few participants in the Mixed Colors group indicated on post session questionnaires that their preference for forced choice during Baseline was driven by avoiding the lower valued rectangles in the free choice option; the forced choice option was viewed as the same in the Identical Colors condition, neither the free or forced choice option was clearly the better source of reinforcement. Additional r esp onses to the post session questionnaires suggest why participants in the Mixed Colors group may have preferred the forced choice option more during the Forced Better and Baseline 2 and 3 conditions It seems that more exposure to the yellow rectangle in i solation may h ave given an impression that the forced option was always the better choice: participants in the Identical Colors group were only exposed

PAGE 86

86 to one condition in which the forced choice option alone contained the yellow rectangle and in which l ower valued rectangles appeared in the free choice option; participants in the M ixed Colors group were exposed to four conditions in which the yellow rectangle was in the forced choice option and at least two lower valued rectangles were in the free choice option. However, this amount of exposure was not present for the Forced Better conditions that directly followed initial Baseline. It is possible that for the Mixed Colors group, the initial Baseline exposure to the single yellow rectangle in forced cho ice and lower valued rectangles in free choice was enough to shift preference more strongly toward forced choice compared to the Identical Colors group who did not have this initial exposure. In general, i n the Identical Colors group, mean proportion of f ree choice selections did not differ across Baseline conditions. This was also the case in the subgroup of the Mixed Colors group that had Forced Better first (t op right panel of Figure 5 2). However, t here was a difference from Baseline to Baselines 2 a nd 3 in the other Mixed Colors subgroup, and, on average, mean proportion of free choice selections w as higher in Baseline in the Identical Colors group. These results can be compared to those of Experiment 3 in which mean proportion of free choice select ions decreased from Baseline to Baselines 2 and 3 in all condition order groups in both baseline arrangements. Those decreases were discussed as being partly caused by the certainty of reinforcement that allowed participants to quickly learn that neither the free nor forced option provided a better outcome in the Baseline conditions In the present experiment, this decrease was only observed in for one group of participants. Importantly in both subgroups of the Identical Colors group, mean proportion of free

PAGE 87

87 choice selections remained near or above 0.5. These data suggest that the quick discrimination and shift in preference toward forced choice in all groups in Experiment 3 and in one group in the present experiment did not occur with many of the parti cipants in the present experiment. It is likely that this was due to the uncertain ty of reinforcement. Because reinforcement was uncertain, it was not as easy for participants to learn that neither the free or forced choice option provided a better outco me, especially when all stimuli were identical and there was no extended exposure to a single and clearly best alternative (as in the Mixed Colors group as discussed above). The univariate ANOVA indicated a significant interaction between baseline arrang ement and condition in the present experiment. This result is partly due to the differences in Baseline preference just discussed. However, individual data indicate that the preferences of several participants in the Identical Colors group did not change across any conditions. Though on average preferences shifted toward free and forced choice in the Free Better and Forced Better conditions, for six out of twenty participants in the Identical Colors group, preferences did not differ by more than 0.08 b etween these conditions. There were no participants in the Mixed Colors group or in either baseline arrangement group in Experiment 3 that showed this pattern of responding. It is likely that the uncertainty of reinforcement in the present experiment ca used participants to be less sensitive to the availability of the best alternative Several participants in the Identical Colors group reported that they preferred the free choice option because they felt that having three rectangles available made it mor e likely that even though only one rectangle could be chosen. In reality, there

PAGE 88

88 was no better chance of a participant earning points whether they chose the free or forced choice option This misinterpretation of the task c ould accoun t for higher preferences for free choice, on average, in the Forced Better and Baseline conditions. It does not account for the lower, more variable responding in the Free Better condition (compared to the same group in Experiment 3). It appears that bec ause reinforcement during the F ree Better and F orced Better conditions rather than simply choosing the option with the highest value. Alt hough individual proportions of free choice selections were more variable than in Experiment 3, mean proportions in the Free Better and Forced Better conditions were closer to expected values in the Mixed Colors group compared to the Identical Colors group. As discussed above, particip ants in the Mixed Colors group were given initial exposure to a condition in which a single yellow rectangle was pitted against a yellow rectangle plus two lower valued rectangles. This exposure may have shifted preference toward an isolated yellow rectan gle because it was easier to identify the best stimulus. Therefore, when that stimulus was available only in the free or forced choice option, participants selected the option that contained that rectangle. The participants in the Identical Colors group had Baseline exposure in which both the free and forced choice options contained only yel low rectangles and most of them chose both the free and forced choice options during Baseline. Some participants continued to sample both options even when the yello w rectangle was available only in the free or forced choice option, though the reasons for this behavior are not clear.

PAGE 89

89 As in the previous three e xperiments there was great deal of between subject variability with regard to free choice preference, parti cularly in the Identical Colors group and in the Baseline conditions in the Mixed Colors group The range of initial Baseline preference was wide, suggesting that participants entered the study with different pre experimental preferences for free choice. Just as in the previous experiments the present experiment used arbitrary stimuli and i t is possible that any relevant history that may otherwise contribute to free choice preference was not generalizable to these experimental conditions. Data and resp onses on post session questionnaires also suggested that some participants misinterpreted the probability of reinforcement, generated their own rules about how best to earn points, and then followed those rules. Some participants indicated that they chose arbitrarily or in an alternating pattern between the free and forced choice options because they were bored with the task Future attempts to evaluate the effects of exposure to differential outcomes in free and forced choice options can be improved in s everal ways. It may be more likely to rather than the arbitrary stimuli. For example, to examine the effect of probability of reinforcement, a gambling simulation c ould be designed in which participants choose multiple levers (free choice). It is also possible as suggested by the results of Fenerty and Tiger (2010), that future studies mi ght find more robust results if choice is among reinforcers rather than response alternatives. Finally, if the goal of an experiment is to evaluate carry over effects of exposure to conditions in which free or forced choice options provide better outcomes longer exposure to those conditions may be necessary.

PAGE 90

90 Table 5 1. Experiment 4: Post session questionnaire contents Question # Question text 1 When you had to choose between the two arrays of rectangles, which of the arrays did you choose more often, the one with one rectangle or the one with three rectangles? Why? 2 In the case of the array of three rectangles, did you think that three rectangles were enough to choose from? Why or why not? 3 Did you enjoy earning points? 4 Is there anything else you would like to say about this study? Table 5 2 Experiment 4 : Results of two tailed t test evaluating whether mean proportions of free choice selections (non transformed) are different from 0.5 Group Condition M t Sig. Identica l Colors Baseline .521 (.223) .297 .773 Free Choice Forced Better .321 (.258) 2.192 .056 Baseline2 .458 (.314) 0.421 .684 Free Better .708 (.323) 2.041 .072 Baseline3 .496 (.299) 0.044 .966 Baseline .533 (.153) 0.689 .508 Free Better .8 54 (.193) 5.812 .000* Baseline2 .667 (.246) 2.143 .061 Forced Better .350 (.346) 1.369 .204 Baseline 3 .625 (.296) 1.336 .214 Mixed Colors Baseline .396 (.251) 1.313 .222 Free Choice Forced Better .108 (.115) 10.783 .000* Baseline2 375 (.208) 1.901 .090 Free Better .908 (.105) 12.251 .000* Baseline3 .342 (.262) 1.912 .088 Baseline .508 (.302) 0.087 .932 Free Better .879 (.132) 9.055 .000* Baseline2 .296 (.216) 2.983 .015* Forced Better .083 (.135) 9.785 .000* Baseline3 .204 (.287) 3.261 .010* Note: Standard deviations are in parentheses. Degrees of freedom were 9 for every condition. A sterisks indicate that the mean p roportion is significantly higher or lower than 0.5 (p < = .05).

PAGE 91

91 Figure 5 1. Schematic of the concurrent chains schedule in place during choice trials for in the Mixed Color s Baseline conditions in Experiment 4 In the initial links, only buttons A or B could be clicked and the positions of the arrays of 1 or 3 rectangles were quasi randomized. In this example, Button A represents the forced choice option and Button B represents the free choice option. In the terminal links, clicks on the circles following the elapse of the FI 5s schedule produced 5 poi nts at a probability that depended on the color of the rectangle that was selected in the terminal link. Initial link Terminal link Forced Choice Free Choice FI 5s 5 pts delivered at 0.8, 0.6, 0.4 or 0.2 probability FI 5s 3s ITI FR 1 FR 1 A B

PAGE 92

92 Figure 5 2. Proportion of free choice selections in the Baseline, Forced Better and Free Better conditions in Experiment 4 Bars indicate proportions averaged across participant; data points indicate proportions from individual participants Data from the Identical Colors and Mixed Colors group s are in the top and bottom left and top and bottom right panels, respectively The top row shows data from partic ipants with the order Baseline Forced Better Baseline Free Better Baseline and the bottom row shows data from participants with the condition order Baseline Free Better Baseline Forced Better Baseline Proportion Free Choice Selections Condition Identical Colors Mixed Colors

PAGE 93

93 CHAPTER 6 EXPERIME NT 5 : AVAILABILITY OF PREF ERRED ALTERNATIVES A common hypothesis regarding the origins of free choice preference is that organisms demonstrating preference for free choice have had experiences in which a free choice option has been differentially associate d with access to more preferred alternatives. This history may then generalize to new choice situations, even when neither a free or forced choice option provides a better alternative The present experiment w as a partial test of this hypothesis. P artic ipants were required to rank a list of cuisine types in order of preference and those rankings were used to manipulate the availability of a more preferred alternative in two conditions. In one, the higher ranked alternative was available only in the fre e choice option and in another the higher ranked alternative was available only in the forced choice option. Therefore, preference for the free versus forced choice option as a function of the availability of a more preferred alternative was directly obs erved. In addition, an initial baseline condition evaluated preference for free versus forced choice when neither option provided access to a more preferred alternative. Th e conditions in this experiment are unique from other studies that have assessed preference for choice of response alternatives in one important way. The stimuli in this study were designed to be used in other studies in which adult humans were participants (including the previous four exper iments in this dissertation). F amiliar stimuli were included in order to assess whether preference for free choice would be affected by the type of stimuli in the free and forced choice options. Organisms p refer free choice even under conditions in which neither a free or forced option provides a better alternative. This may be because

PAGE 94

94 the free choice context itself has been established as a conditioned reinforcer due to history in which free choice has produced better outcomes. If preference for f ree choice is maintained by conditioned reinforcement, then it is likely that contexts that are more similar to those that an organism may have confronted in the natural environment will support higher degre es of free choice preference. It is likely that an undergraduate student has had an extensive history with choosing the type of cuisine that s/he would like to eat and so an experiment that models this choice may be a better test of the effect of past history on current preference for free choice than one using more generic and less meaningful stimuli. Method Participants, Setting, and Apparatus Twenty four students ( 13 females and 11 males) at the University of Fl orida served as participants. Participants w ere recruited from a pool comprised of stud ents enrolled in General Psychology and from two sections of Applied Behavior Analysis and they from 18 to 21 years All participants completed the experimental task using t he mouse and keyboard of a PC in o ne of two rooms containing the PC a table, and a chair and were alone in the room for the duration of the task T he task was presented via a program designed in Microsoft Visual Basic 6.0. Experimental Design At the b eginning of their session, each participant was instructed to imagine that s/he was traveling to a new city for one night and told that s/he would be given a number of scenarios in which s/he would choose to which side of a new town to travel to eat dinner B efore each set of choices participants conduct ed a preference ranking of

PAGE 95

95 nine different types of cuisines. After each preference ranking participants were given a set of choices between the East and West side of town The choices were presented in the context of a concurrent chains procedure An AB C design was used to evaluate the effects of alternative preference on preference for free versus forced choice options. Each participant conducted three preference rankings (one prior to every condition ) and was exposed to three conditions in which the alternative with the highest ranking was present in both the forced and free choice options, the free choice option only, or the forced choice option only; the Baseline, Free Better, and Forced Better c ond itions, respectively. Experimental control was demonstrated by counterbalancing the order of exposure to these conditions across groups such that 1 2 participants were given the order Baseline Free Better Forced Better and the other 1 2 participants were gi ven the order Baseline Forced Better Free Better Experimental events and conditions are described in detail below. Preference Ranking At the beginning of each condition, participants were prompted to complete a preference ranking. During each preferen ce ranking, participants were shown a list of ni ne different types of cuisines with each type named as follows (including parenthetical remarks): American (BBQ/Burgers), Chinese, Indian, Italian (not including pizza), Japanese (including sushi), Mediterran ean (e.g. Greek), Mexican, Pizza, Tapas The cuisines were listed vertically and were presented in the same order (alphabetically) for every ranking for every participant. Participants were given the following instructions: Below is a list of different t ypes of cuisines. Thinking only about the type of food (and not particular restaurants), rank these cuisines in order of your preference of them with 1 being your favorite and 9 being your least favorite. Do not give the same ranking to more than one item (i.e. give each

PAGE 96

96 type of food a different number from 1 9). Click "Finished" when you are done ranking. A text box was p ositioned in front of each cuisine name and participants entered one number between 1 and 9 into each text box to indicate the order of their preferences. If a number was repeated, a text box was left incomplete, or an entry other than a digit between 1 and 9 was made, an error message prompt ed the participant to correctly complete the ranking. Once the ranking was finished, the progra m advanced to the set of choice trials for that condition. The alternatives presented in the choice trials were based on the rankings made immediately prior to that set in order to ensure that the alternatives presented in th e choice trials reflected the to attempt to accommodate any momentary changes in cuisine preference). Concurrent C hains Procedure A concurrent chains schedule was in place in the choice trials, as depicted in Figure 6 1. Participants we re given the following instructions prior to completing the choice trials: You are staying in a new city for one night and you are trying to decide where to go for dinner. All the restaurants in the city are similarly priced and have been reviewed equally well. You also have to travel the same distance to get to both places (you are going by cab and the ride will cost the same no matter what). On one side of town, there is an area with three restaurants all located in close proximity. On the other side o f town, there is only one restaurant. When you click 'START,' you will be given several sets of choices between the East and West side. Each choice between East and West should be made independently, only considering the options in that trial. That is, each choice should be considered the choice you would make for that one night; in other words do not make your choices as if they are consecutive. Another set of brief statements provided general instruction s on how to respond in the concurrent chains sch edule.

PAGE 97

97 In the initial links of the concurrent chains schedule, two arrays of approximately rectangle shaped buttons were shown on the computer screen. One array contained one button and the other contained three buttons These arrays were pla ced above two additional clickable buttons labeled E AST and W EST E AST was always on the right and W EST was always on the left but the positions of the button arrays were quasi randomized across trials. That is, sometimes one button appeared above EAST and three buttons appeared above WEST and vice versa; neither array appeared in the same position more than two times in a row. Each of the buttons in the arrays was labeled with the name of one of the cuisines that had appeared in the preference ranking. The procedures by which each label was determined are described in the description of experimental conditions below. The buttons in the arrays were slightly dimmed during the initial links to indicate that they could not be clicked in that link. A fixe d ratio (FR) 1 response schedule was in place in the initial links: a single mouse click to either EAST or W EST removed the EAST and W EST buttons and the array that had appeared above the non chosen button, and constituted as a response for either the arr ay with one alternative or for the array with three alternatives A click to the button associated with one alternative initiated the forced choice terminal link and a cl ick to the button associated with three alternatives initiated the free choice termin al link. The button(s) that remained on the screen in the terminal links were no longer dimmed, indicating that they could be clicked. In the forced choice terminal link, the singl e alternative remained in its original position on the screen and a second FR 1 schedule was in place One mouse click to the button removed it from the screen and ended the terminal link. In the free choice terminal link, the three alternatives also

PAGE 98

98 remained in their original positions on the screen and an FR 1 schedule was i n place. One mouse click to any of the three buttons removed all three buttons from the screen and terminated the link. An inter trial interval (ITI) occurred after every terminal link. During the ITI the screen remained blank for 3 seconds. At the en d of the ITI, a new trial began. When a condition ended, a new preference ranking was conducted and followed by a set of instructions to remind the participant how to complete the task and to begin the next set of choice trials When all three condit ions (either Baseline Free Better -Forced Better or Baseline Forced Better Free Better) were completed and before they left the room, participants responded to a brief questionnaire consisting of four questions. These questions are listed in Table 6 1. Experimental Conditions Baseline condition. In the Baseline condition, the alternative with the highest ranked preference was available in both the free and forced choice options. In the free choice option each of the three alternatives was labeled as a different cuisine. The configurations of these alternatives were prearranged to be some combination of the cuisines that were ranked 2 5 in the immediately preceding preference ranking. The alternati ve in the forced choice option was matched to the alte rnative in the free choice option that had the highest ranking. For example, if a configuration was programmed such that the alternatives in the free choice option were the cuisines ranked 2, 4, and 5, the alternative in the free choice option would be th e cuisine ranked 2. The configurations were prearranged such that nine included the cuisine ranked 2 as the highest preference alternative and in the other nine, the cuisine ranked 3 was the highest preference alternative. The order of trials drawn from the 18 pre programmed

PAGE 99

99 configurations was randomized for each participant. The Baseline condition end ed when 18 choice trials were completed. Free Better condition. In this condition, the alternative with the highest ranked preference was available only i n the free choice option. In the free choice option, the three alternatives included the cuisine ranked 1 in the immediately preceding preference ranking plus two different cuisines from the rankings 2 t hrough 5. Combinations of these alternatives were p re programmed, but the order in which they were presented was randomized for each participant. The alternative available in the forced choice option was a cuisine ranked between 2 and 5. Each of these cuisines was presented five times with the order of presentation randomized The Free Better condition ended when 20 choice trials were completed. Forced Better condition. In this condition, the alternative with the highest ranked preference was available only in the forced choice option. The alternati ve in the forced choice option was always the cuisine that the participant had ranked 1 in the immediately preceding preference ranking. In the free choice option, each of the three alternatives was labeled as a different cuisine. The configurations of t hese alternatives were prearranged to be some combination of the cuisines ranked 2 through 5. Combinations of these alternatives were pre programmed, but the order in which they were presented was randomized for each participant. The Forced Better condi tion ended when 20 choice trials were completed. Data Analysis Data from individual participants and data averaged across participants were analyzed in the same manner as in Experiments 1 4.

PAGE 100

100 Results The results of Experiment 5 are presented in Figure 6 2. P roportion of free choice selections is graphed as a function of condition. This proportion was calculated by dividing the sum of free and forced choice selections by the total number of free choice selections. Bars indicate the mean proportion s of free choice selections for all participants in each condition and individual proportions are plotted as data points. D ata in the top panel are from participants with the order Bas eline Forced Better Free Better and data in the bottom panel are from participa nts with the order Baseline Free Better Forced Better Visual analysis of these graphs indicated that on average, when preference for cuisin e was controlled (in Baseline) participants chose the option in which three a lternatives were available more than the option in which one alternative was available Selection of the free or forced choice option was directly affected by the presence of the choice option, participa nts selected free choice more and when it was only available in the forced choice option, participants selected forced choice more The mean proportion of free choice selections was higher than 0.5 in all Baseline and Free Better conditions and lower than 0.5 in both Forced Better conditions. Analyses were also conducted to reveal statistically significant results. C ondition m eans and standard deviations are presented in Table 6 2. A univariate ANOVA indicated a main effect of condition F (2, 44) = 42. 42, p < .0 0 1, and a significant interaction between condition and condition order F (2, 44) = 4.25, p = .021 This interaction was driven by the differences in proportion of free choice selections between participants for whom the Forced Better condition followed Baseline and those for whom

PAGE 101

101 it followed the Free Better conditi on. An independent samples t test indicated that t he mean proportion of free choice selections was higher in the Forced Better condition that followed the Free Better condition ( M = 656 SD = 395 ) compared to the mean proportion in the Forced Better condition that followed Baseline ( M = 324 SD = 362 ) at a statistically significant level, t (22) = 2.1 56 p = .04 2 There were no differences between the Baseline or Free Better condi tions depending on condition order group Finally, a two tailed t test using non transformed data indicated that the mean proportion of free choice selections was significantly different from 0.5 in all conditions except the Forced Better that followed F ree Better T he mean proportion of free choice selections was higher than 0.5 in both Baseline conditions, t ( 11 ) = 3.916, p = .002 and t ( 11 ) = 2.968, p = .013, and both Free Better conditions t ( 11 ) = 14.326, p < .001 and t ( 11 ) = 10.018, p < .001. The mea n proportion of free choice selections was lower than 0.5 in the Forced Better condition that followed Baseline t ( 11 ) = 4.774, p = .001. Discussion The results of this experiment provide support for the hypothesis that preference for free choice may be the result of a history in which free choice options better choice were directly driven by the availability of the cuisine that was ranked highest in the preference ranking that immediately preceded the Forced Better and Free Better conditions. All but three participants that preferred free choice during Baseline shifted preference toward the forced choice option when their most preferred alternat ive was only available there These results support those of Fisher et al. (1998) and Suzuki (1997). Fisher et al. found that participants that had previously preferred free choice of reinforcers to experimenter choice of reinforcers allocated more responding to the

PAGE 102

102 experimenter ch oice option when that option resulted in the delivery of a higher quality reinforcer than the free choice option. Suzuki observed that participants who had a choice between free choice alternatives that delivered 10, 15, or 15 points and a forced choice a lternative that delivered 10 points preferred free choice more than participants with the same forced choice alternative and three free choice alternatives that all delivered 10 points. That is, preference for choice was higher in the condition in which t he free choice option provided a better outcome than the forced choice option. The high degree of preference for the free choice option in Baseline may indicate that many participants came int o the experiment with a history in which free choice options a llow ed for more access to better outcomes and/or the accommodation of changing preferences post Did you like having an opti on available in wh ich there was more than one restaurant to choose from? Why or why not? they liked having the opportunity to change their mind. This suggests that participants recognized that the free choice option would be better able to accommodate changes in preference and this self report is likely due to a history in which such changes have been accommodated when multiple alternatives were available. Of course, such a history was not specifically tested for in this experiment so i t is not possible to definitively conclude that it is what contributed to the high degree of Baseline preference for free choice. The results of the present experiment indicate that the presence of a more preferred alternative strongly affected free and forced choice preference However, preference for free choice was higher in the Forced Better condition that followed Free

PAGE 103

103 Better compared to the Forced Better condition that followed Baseline. These results suggest that although preference may shift to a forced choice option when a more preferred alternative is available there, there may still be cases in which the free choice option is preferred. In addition, higher preference in the Forced Better condition that followed Free Better suggests a possibl e carry over effect. That is, having had a brief history in which the free choice option contained the more preferred alternative may have produced higher preference for free choice in the subse quent Forced Better condition. There was no similar carry ov er effect of the Forced Better condition when it was followed by the Free Better condition. These results are similar to those found by Karsina et al. (2011). They found that even when selection of a restricted choice option was differentially reinforce d, the degree of preference for that option was not as high as the degree of preference for a free choice option when its selection was differentially reinforced. In addition, the experimenters found that when free choice selection was differentially rein forced, preference for free choice persisted in a subsequent equal outcome condition. However, preference for restricted choice did not persist in the equal outcome condition that followed differential reinforcement of restricted choice selection. These results and those of the present experiment suggest that preference for free choice is more robust than preference for forced choice, perhaps because of a longer pre experimental history in which free choice has provided better outcomes. This hypothesis c ould be better tested in future studies by repeating the Baseline condition following the Forced Better and Free Better conditions to determine whether preferences from either of these conditions carry over to a subsequent Baseline condition.

PAGE 104

104 A major f inding of this experiment that differs from the other four experiment s in this dissertation is that preference for the free choice option in Baseline was high and between subject variability was low. It is likely that this finding is due to the nature of the procedure used in this experiment. Specifically, whereas the other four experiments in this dissertation included arbitrary and possibly irrelevant stimuli, the present experiment included scenarios and stimuli that are likely to be similar to those e xperienced by participants in their real lives. Catania (1980) observed: If preferences among alternatives change from moment to moment, a free choice preference may develop as the organism learns that momentarily preferred alternatives are always availa ble in free choice terminal links. A preference based on such learning might be manifested even over sessions in which free choice and forced choice reinforcers were perfectly matched. It is reasonable to speculate that such learning, based on history of free choices and forced choices, might generalize across a variety of experimental and extraexperimental settings (138). It is likely that the more realistic nature of the present experimental task allowed for better generalization of pre ex perimental histories and therefore better evaluation of free choice preference. Th e present experiment was designed to test the hypothesis that preference for free choice is due to a history in which free choice options have been differentially associat ed with access to more preferred alternatives. The results of this experiment both the choice data and questionnaire responses support this hypothesis. This hypothesis can continue to be tested by conducting additional experiments in which the indepe ndent variable is some feature of an alternative the value of which can be changed such as amount, quality, duration, preference, etc. Finally, future studies of free choice preference should

PAGE 105

105 bet experimental histories as well as better generalization of experimental results.

PAGE 106

106 Table 6 1. Experiment 5: Post session questionnaire contents Question # Question text 1 How did you decide whether to choose the Eas t or West side of town? 2 Did you like having an option available in which there was more than one restaurant to choose from? Why or why not? 3 Did your preference rankings change at all through the course of the study? Why or why not? 4 Is there anyt hing else you would like to say about this study? Table 6 2. Experiment 5 : Mean proportions of free choice selections (non transformed) in each condition for both condition sequences. Condition M Baseline .764 (.233)* Forced Better .188 (.227) F ree Better .917 (.101)* Baseline .687 (.218)* Free Better .846 (.120)* Forced Better .417 (.299) Note: Standard deviations are in parentheses. Asterisks indicate that the mean proportion is significantly larger than 0.5 (p < = .05).

PAGE 107

107 Figure 6 1. Schematic of the concurrent chains schedule in place during choice trials for all conditions in Experiment 5 In the initial links, only buttons EAST or WEST could be clicked and the positions of the arrays of 1 or 3 cuis ine alternatives were quasi randomized. In this example, EAST represents the forced choice option and WEST represents the free choice option. In the terminal links, one click to an alternative terminated the link. Italian Mexican Indian Indian WE S T T EAST FR 1 Initial link Indian Italian Mexican Indian Terminal link Fo rced Choice Free Choice FR 1 3s ITI FR 1 FR 1

PAGE 108

108 Figure 6 2. Proportion of free choice selections in the Baseline, Forced Better and Free Better conditions in Experiment 5 Bars indicate proportions averaged across participant; data points indicate proportions from individual participants The top panel shows data from partic ipants with the order Baseline Forced Better Free Better and the bottom row shows data from participants with the condition order Baseline Free Better Baseline Condition Proportion Free Choice Selections

PAGE 109

109 C HAPTER 7 GENERAL DISCUSSION The purpose of the present investigation was to better understand the conditions under which preference for free choice is exhibited. All experiments were designed to replicate or add to previous experiments conducted with non humans, preschoolers, individuals with developmental disabilities, and typic ally developing adults. Experiments 1 and 2 extended previous research on the role of number of alternatives and certainty of reinforcement. Results of these experiments did not indicate consistent effects of either variable on preference for free choic e. Results of Experiments 3 and 4 indicated that when outcomes were better for free choice, participants, on average, preferred free choice and when outcomes were better for forced choice, they preferred forced choice In conditions in which outcomes for free and forced choice were the same, only some T here was no strong evidence that brief histories of differential outcomes for free or forced choice influenced preferen ce when outcomes were the same. Results of Experiment 5 indicated a direct effect of the availability of a preferred alternative on free and forced choice preferences. In addition, this is the only experiment in which consistent and relatively high prefe rence for free choice was exhibited in an initial condition in which free and forced choice outcomes were equal. Limitations The procedures of Experiments 1 4 were designed to be comparable to previous non human and human studies that used arbitrary stim uli and provided choices between response alternatives instead of reinforcers. Although these procedures allow for better cross species and cross study comparisons, in the present experiments there

PAGE 110

110 were several drawbacks to these procedure s Because the stimuli were arbitrary, they were relatively meaningless to participants. In many cases it seemed that this resulted T he experimental tasks were simple: click A or B then click on a shape to d eliver points. However, many participants reported in post session questionnaires that they were looking for patterns followed even though they were unrelated to the actual experimental contingencies. Other participants stated that the tasks were boring and tedious and admitted that they tasks that evaluate the same variables studi ed in these experiments would be better suited to study preference for choice with adult humans. Another major limitation of the first four experiments was the large degree of between subject variability. This variability made data analysis difficul t, and challenged the ability to draw general conclusions about the effects of the independent variables. This variability is possibly due to differing pre experimental histories with regard to free choice preference. Some participants reported that they liked fewer options because they required less effort or provided less distraction. It is likely that at least some of these participants also prefer restricted choice in their own lives. However, if we assume that participants do prefer free choice in their ow preference generalize to initial baseline conditions in the present experiments? I believe that th e procedural design contributed to thi s lack of generalization. For instance, participants may prefer multiplexes to sing le screen cinemas or large menus to small ones, etc. but whatever contingencies produce free choice preference under

PAGE 111

111 these sorts of conditions may be absent when choices are between identical shapes on a computer screen. The task in Experiment 5 was more similar to the choices that participants would make in their own lives and baseline preferences for free choice were highest and least variable in this experiment. The results of Experiment 5 support the use of more relevant, familiar, and realistic tas ks to study free choice preference. Future Directions The outcomes of the present experiments indicate that there is more work to be done in the area of preference for choice. Evaluations of the variables under study in each of these experiments are sti ll incomplete but remain relevant. Improvements in the methodologies of the present experiments would help address some lingering problems and questions regarding free choice of response alternatives. Organisms also make ch oices between reinforcers, and little research in this area has been conducted with adult humans. The results found by Fenerty and Tiger (2010) suggest that choice of consequences of responses may be more important than choice of the responses themselves. T he present experiments could easily be designed to evaluate the effect of each of the independent variables on preference for free choice of reinforcers rather than response alternatives. Other problems in the area of preference for choice remain to be studied For example, the rol e of punishment on preference for free choice has been relatively overlooked The results of previous studies and of Experiments 3 and 4 in this dissertation indicate that providing better outcomes for free or forced choice shifts preference in the direct ion of the better option. However, in few studies ha ve free or forced choice been associated with a loss of reinforcement or presentation of an aversive stimulus Many researchers have hypothesized that preference for free choice

PAGE 112

112 is shaped by ontogen et ic histories in which free choice has been associated with better outcomes. A lack of such history could result in a lack of free choice preference However it is also possible that preference for free or forced choice is shaped by experiences in which f r ee or forced choice is associated with negative outcomes As Lockhart (1979) points out, it is possible for organisms p. 27 ). Free choice might be surrendered if choosing wrong is more likely in a free choice option Evaluating the role of a history of punishment would provide fuller picture of the types of ontogen et ic histories that contribute to free choice preference or l ack thereof. Another important issue that remains to be sorted out is the nature of what it means to have (and therefore prefer) free choice Martin, Martin, Yu, and Fazzio (2006) observe that choice can be defined both in terms of the behavior of choos ing as well as in terms of the presence of multiple alternatives (reinforcers or response alternatives). Therefore preference for free choice can be described both as a preference for multiple alternatives and as a preference for the behavior of choosin g itself and the two are likely to be confounded. For example, Schmidt, Hanley, and Layer (200 9) designed an experiment that controlled for the number of alternatives in free choice and no choice reinforcer arrays. This control was implemented to elimin ate the influence of the presence of more reinf orcers in the free choice array. In the choice terminal link an array of five identical reinforcers wa s made available from which participant s could choose one. In the no choice terminal link, the same arra y was shown, but the experimenter chose the reinforcer. P articipants consistently preferred the choice option

PAGE 113

113 and Schmidt et al. (2009) concluded that their results supported previous research in which different numbers of stimuli were available in the f ree and forced/no choice options. However, the distinction between preference for free choice of alternatives and preference for the opportunity to choose are not distinguishable here. Other studies that have observed preference for free choice include s imilar confounds between preference for the availability of multiple alternatives and preference to choose or control the delivery of consequences (Dixon & Tibbetts, 2009; Karsina et al., 2010; Thompson, Fisher & Contrucci, 1998). This is an interesting a spect of research on preference for free choice that remains to be disentangled. Conclusions Choice is a ubiquitous feature i n the existences of most organisms and m any researchers have found that organisms prefer conditions in which they have access to more versus less choice. A number of variables modify this preference including the number of and preference for available alternatives, and various features of reinforcement including amount and probability. Preference for choice is the result of a c ombination of phylogenetic and ontogenetic processes and understanding both is important to understanding how and whether individual organisms come to prefer multiple alternatives and the behavior of choosing itself Identifying the conditions under whic h more choice is preferred to less (and vice versa) has important implications for psychologists, economists, marketers, educators, politicians, and any other group interested in explaining and controlling why and how organisms make choices.

PAGE 114

114 LIST OF REFERENCES Catania, A.C. (1975). Freedom and knowledge: An experimental analysis of preference in pigeons. Journal of Experimental Analysis of Behavior 24 89 106. Catania. A.C. (1980). Freedom of choice: A behavioral analysis. In G.H. Bower (Ed.), The psychology of learning and motivation (Vol. 14, pp. 97 145). New York: Academic Press. Catania, A.C. & Sagvolden, T. (1980). Preference for free choice over forced choice in pigeons. Journal of Experimental Analysis of Behavior 34 77 86. Cerutt i, D.T. & Catania, A.C. (1986). Rapid determinations of preference in multiple concurrent chain schedules. Journal of Experimental Analysis of Behavior 46 211 218. Cerutti, D. & Catania, A.C. (1997). key s versus key area. Journal of Experimental Analysis of Behavior 68 349 356. Dixon, M.R. & Tibbetts, P.A. (2009). The effe cts of choice on self control. Journal of Applied Behavior Analysis 42 243 252. Fenerty, K.A. & Tiger, J.H. (2010). Determining making opportunities: Choice of task versus choice of consequence. Journal of Applied Behavior Analysis 43 503 507. Fisher, W.W., Thompson, R.H., Piazza, C.C., Cro sland, K., & Gotjen, D. (1997). On the relative rei nforcing effects of choice and differential consequences. Journal of Applied Behavior Analysis 30 423 438. Galizio, M. (1979). Cont ingency shaped and rule governed behavior: I nstructional control of human loss avoidance Journal of the Experimental Ana lysis of Behavior 3 1 53 7 0 Hackenberg, T.D. & Joker, V.R. (1994). Instructional versus schedule control of humans' choices in situations of diminishing returns. Journal of the Experimental Analysis of Behavior 62 367 383 Hayes, S. C., Brownstein, A. J., Zettle, R. D., Rosenfarb, I., & Korn, Z. (1986). Rule governed behavior and sensitivity to changing consequences of responding. J ournal of the Experimental Analysis of Behavior 45 237 256. Hayes, S.C., Kapust, J., Leonard, S.R., & Rosenfarb, I. ( 1981). Escape from freedom: Choosing not to choose in pigeons. Journal of the Experimental Analysis of Behavior 36 1 7.

PAGE 115

115 Iyenga r, S.S. & Lepper, M.R. (2000). When choice is demotivating: Can one desire too much of a good thing? Journal of Personali ty and Social Psychology 79 995 1006. Karsina, A., Thompson, R.H., & Rodriguez, N.M. (2011). Effects of a history of differential reinforcement on preference for choice. Journal of the Experimental Analysis of Behavior 95 189 202. Lockhart, K.A. (197 9). Behavioral as sessment of human preference. T he Behavior Analyst 2 20 29. Martin, T.L., Yu, C.T., Mart in, G.L. & Fazzio, D. (2006). On choice, prefere nce, and preference for choice. The Behavior Analyst Today 7 234 241. Ono, K. (2000). Free ch oice preference under uncertainty. Behavioural Processes 49 11 19. Reutsk aja, E. & Hogarth, R.M. (2009). Satisfaction in choice as a function of the Psychology & Marketing 26 197 203. Rosenfarb, I.S., N ewland, M.C., Brannon, S.E., & Howey, D.S. (1992). Effects of self generated rules on the development of schedule controlled behavior. Journal of the Experimental Analysis of Behavior 58 107 121. Schmidt, A.C., Hanle y, G.P., & Layer, S.A. (2009). A furt her analysis of the value of choice: Controlling for illusory discriminative stimuli and evaluating the ef fects of less preferred items. Journal of Applied Behavior Analysis 42 711 716. Sh ah, A.M. & Wolford, G. (2007). Buying behavior as a function of parametric variation of number of choices. Psychological Science 18 369 370. Skowronski J.J. & Carlston, D.E. (1982). Effects of previously experienced outcomes on the desire for choice. Journal of Personality and Social Psychology 43 689 701. Suzu ki, S. (1997). Effects of number of alternatives on choice in humans. Behavioural Processes 39 205 214. Suzuki, S. (1999). Selection of forced and free choice by monkeys (Macaca fascicularis). Perceptual and Motor Skills 88 242 250. Suzuki, S. (20 00). Choice between single response and multichoice tasks in humans. The Psychological Record 50 105 115. Thompson, R.H., Fisher, W.W., & Contrucci, S.A. (1998). Evaluating the reinforcing effects of choice in comparison to reinforcement rate. Resear ch in Developmental Disabilities 19 181 187.

PAGE 116

116 Tiger, J.H., Hanley G.P., & Hernandez, E. (2006). An evaluation of the value of choice with preschool children. Journal of Applied Behavior Analysis 39 1 16. Vo ss, S.C. & Homzie, M.J. (1970). Choice as a value. Psychological Reports 26 912

PAGE 117

117 BIOGRAPHICAL SKETCH Jennifer was born and raised in Pennsylvania and at Lafayette College in Easton, PA in 2004 She majored in philosophy and psychology and took an interest in beha vior analysis from the first day of her first Learning class. Jennifer spent the year after graduation as an AmeriCorps*VISTA volunteer at Mercyhurst College in Erie, PA and got hands on practice using behavior analytic techniques in the tutoring programs she helped coordinate. After her AmeriCo rps year, Jennifer entered the m Maryland, Baltimore County and worked in the Neurobehavioral Unit at Kennedy Krieger Institute (KKI), Baltimore, MD. Wh ile working at KKI and studying at UMBC, Jennifer developed a strong passion for the science of behavior analysis. Applied, clinical work was fulfilling but Jennifer wanted more exposure to the experimental analysis of behavior and the philosophy of beha vior a nalysis. After completing her Master of Arts degree in 2007, Jennifer entered the Ph.D. program at the University of Florida. Under the supervision of Dr s. Timothy Hackenberg and Jesse Dallery, Jennifer conducted research in the area of behavioral economics with both pigeons and undergraduate students. Jennifer also served as instructor for several sections of General Psychology and Principles of Behavior Analysis and worked as an academic advisor in the Psychology department. Following graduation from UF in 2011, Jennifer hopes to continue to teach and mentor at the college level.