<%BANNER%>

Behavioral Consistency and Inconsistency in the Pursuit of Self-Control

Permanent Link: http://ufdc.ufl.edu/UFE0022340/00001

Material Information

Title: Behavioral Consistency and Inconsistency in the Pursuit of Self-Control
Physical Description: 1 online resource (61 p.)
Language: english
Creator: Laran, Juliano
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: goals, inhibition, regulation
Marketing -- Dissertations, Academic -- UF
Genre: Business Administration thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: During the course of a day, consumers encounter many opportunities to engage in self-regulatory behaviors. In some cases, resistance to temptations leads to more resistance and indulgence leads to more indulgence. In other cases, resistance to temptations leads to subsequent indulgence and indulgence leads to subsequent resistance. I posit that regulatory successes and failures can be controlled by a bottom-up goal guidance system. In this system, a behavior that sustains goal activation encourages similar behaviors, whereas a behavior that results in goal achievement encourages dissimilar behaviors. Six experiments demonstrate factors that influence goal activation and goal achievement. A pilot experiment shows that some behaviors lead to consistent behavior whereas other behaviors lead to inconsistent behavior. Experiment 1 shows that when an initial behavior is insufficient (sufficient) for goal achievement, it encourages consistent (inconsistent) subsequent behaviors. Experiment 1A provides support for a bottom-up goal guidance system that relies on goal activation and inhibition. Experiments 2 and 3 force people to use a bottom-up or a top-down guidance system and show that each system leads to a different pattern of behavior. Experiment 4 shows that a bottom-up guidance system can influence behaviors that a top-down guidance system deems irrelevant.
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 Juliano Laran.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Janiszewski, Chris A.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-02-28

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022340:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022340/00001

Material Information

Title: Behavioral Consistency and Inconsistency in the Pursuit of Self-Control
Physical Description: 1 online resource (61 p.)
Language: english
Creator: Laran, Juliano
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: goals, inhibition, regulation
Marketing -- Dissertations, Academic -- UF
Genre: Business Administration thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: During the course of a day, consumers encounter many opportunities to engage in self-regulatory behaviors. In some cases, resistance to temptations leads to more resistance and indulgence leads to more indulgence. In other cases, resistance to temptations leads to subsequent indulgence and indulgence leads to subsequent resistance. I posit that regulatory successes and failures can be controlled by a bottom-up goal guidance system. In this system, a behavior that sustains goal activation encourages similar behaviors, whereas a behavior that results in goal achievement encourages dissimilar behaviors. Six experiments demonstrate factors that influence goal activation and goal achievement. A pilot experiment shows that some behaviors lead to consistent behavior whereas other behaviors lead to inconsistent behavior. Experiment 1 shows that when an initial behavior is insufficient (sufficient) for goal achievement, it encourages consistent (inconsistent) subsequent behaviors. Experiment 1A provides support for a bottom-up goal guidance system that relies on goal activation and inhibition. Experiments 2 and 3 force people to use a bottom-up or a top-down guidance system and show that each system leads to a different pattern of behavior. Experiment 4 shows that a bottom-up guidance system can influence behaviors that a top-down guidance system deems irrelevant.
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 Juliano Laran.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Janiszewski, Chris A.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-02-28

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022340:00001


This item has the following downloads:


Full Text

PAGE 1

1 BEHAVIORAL CONSISTENCY AND INCONSISTENCY IN THE PURSUIT OF SELF-CONTROL By JULIANO LARAN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008

PAGE 2

2 2008 Juliano Laran

PAGE 3

3 ACKNOWLEDGMENTS I would first like to thank Chris Janiszewski fo r his relentless dedica tion to m y training and for shaping who I am going to be as a researcher for the rest of my life. I would certainly not be here without his constant support and encouragemen t. I would also like to thank Alan Sawyer and Alan Cooke for all the things I learned from working with them in the past few years. I thank Robyn LeBoeuf, Joe Alba, Jason Colquitt, and Joel Cohen for their valuable contribution to this dissertation. I would also like to thank Marcus Cunha, Jr for his friendship and for teaching me so many things through our several research projects. My colleagues made this a wonderful experience, and I am especially indebted to Da n Rice, Julia Belyavsky, Jesse Itzkowitz, Baler Bilgin, Wouter Vanhouche, Melissa Minor, JoAn drea Hoegg, Tim Silk, and Eduardo Andrade for their friendship and support. My parents, my sisters, and my grandpa rents are the reason why I continue to work harder and harder so I can keep making them proud. Finally, I want to thank Juliana for being the most wonderful person I could have ever met.

PAGE 4

4 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................3LIST OF FIGURES.........................................................................................................................6ABSTRACT.....................................................................................................................................7 CHAP TER 1 INTRODUCTION....................................................................................................................82 TOP-DOWN AND BOTTO M-UP GOAL GUI DANCE....................................................... 10Top-Down Guidance.............................................................................................................. 10Bottom-up Guidance............................................................................................................. ..13Passive Goal Guidance...........................................................................................................163 PILOT EXPERIMENT........................................................................................................... 22Method....................................................................................................................................23Participants and Design................................................................................................... 23Procedure.........................................................................................................................23Stimuli........................................................................................................................ .....24Results.....................................................................................................................................24Manipulation Check and Control Tests........................................................................... 24Analysis....................................................................................................................... ....25Supporting Evidence............................................................................................................ ...25Discussion...............................................................................................................................274 EXPERIMENT 1....................................................................................................................30Method....................................................................................................................................31Participants and Design................................................................................................... 31Procedure.........................................................................................................................31Stimuli........................................................................................................................ .....31Results.....................................................................................................................................32Control Tests...................................................................................................................32Analysis....................................................................................................................... ....32Discussion...............................................................................................................................335 EXPERIMENT 1A.................................................................................................................36Method....................................................................................................................................37Participants and Design................................................................................................... 37Procedure and Stimuli..................................................................................................... 37

PAGE 5

5 Results.....................................................................................................................................38Emotions....................................................................................................................... ...38Reaction Times................................................................................................................38Discussion...............................................................................................................................396 EXPERIMENT 2....................................................................................................................42Method....................................................................................................................................42Participants and Design................................................................................................... 42Procedure and Stimuli..................................................................................................... 42Predictions..............................................................................................................................43Results.....................................................................................................................................44Commitment and Progress.............................................................................................. 44Analysis....................................................................................................................... ....44Discussion...............................................................................................................................457 EXPERIMENT 3....................................................................................................................47Method....................................................................................................................................47Participants and Design................................................................................................... 47Procedure and Stimuli..................................................................................................... 47Pretest..............................................................................................................................48Results.....................................................................................................................................48Discussion...............................................................................................................................498 EXPERIMENT 4....................................................................................................................51Method....................................................................................................................................51Participants and Design................................................................................................... 51Procedure and Stimuli..................................................................................................... 51Results.....................................................................................................................................52Discussion...............................................................................................................................529 GENERAL DISCUSSION..................................................................................................... 54Theoretical Implications and Future Research.......................................................................54Marketing Implications...........................................................................................................56Limitations.................................................................................................................... ..........57LIST OF REFERENCES...............................................................................................................58BIOGRAPHICAL SKETCH.........................................................................................................61

PAGE 6

6 LIST OF FIGURES Figure page 3-1 Influence of activating and achieving an indulgence goal on the accessib ility of goal concepts..............................................................................................................................294-1 Influence of indulging / regulating an d subjectively activating / achieving a goal at time 1 on the subsequent desire for food........................................................................... 355-1 Influence of indulging / regulating and subjectively activating / achieving a goal at time 1 on the accessibility of goal concepts....................................................................... 416-1 Influence of full / partia l regulation and monitoring on the subsequent desire for food... 467-1 Influence of indulging / regulating and e xperiencing / being told about a behavior at time 1 on the subsequent valuation of food....................................................................... 508-1 Influence of indulging / regulating at time 1 on the subs equent valuation of products in different contexts.......................................................................................................... .53

PAGE 7

7 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree OF Doctor of Philosophy BEHAVIORAL CONSISTENCY AND INCONSIS TENCY IN THE PURSUIT OF SELFCONTROL By Juliano Laran August 2008 Chair: Chris Janiszewski Major: Business Administration During the course of a day, consumers encount er many opportunities to engage in selfregulatory behaviors. In some cas es, resistance to temptations leads to more resistance and indulgence leads to more indulgence. In othe r cases, resistance to temptations leads to subsequent indulgence and indulge nce leads to subsequent resist ance. I posit that regulatory successes and failures can be controlled by a bottom-up goal guidan ce system. In this system, a behavior that sustains goal ac tivation encourages similar behaviors, whereas a behavior that results in goal achievement encourages dissimilar behaviors. Six experiments demonstrate f actors that influence goal ac tivation and goal achievement. A pilot experiment shows that some behaviors lead to consistent behavior whereas other behaviors lead to inconsistent behavior. Experiment 1 shows that when an initial behavior is insufficient (sufficient) for goal achievement, it encourages consistent (inconsistent) subsequent behaviors. Experiment 1A provides support for a bottom-up goal guidance system that relies on goal activation and inhibition. E xperiments 2 and 3 force people to use a bottom-up or a topdown guidance system and show that each system leads to a different pattern of behavior. Experiment 4 shows that a bottom-up guidance sy stem can influence beha viors that a top-down guidance system deems irrelevant.

PAGE 8

8 CHAPTER 1 INTRODUCTION A consum er walks into a supermarket to shop. Early in the shopping trip, she passes through the bakery department. Sh e considers buying a pastry, but exerts self-cont rol and decides not to make the purchase. Later in the shopping trip, the consumer encounters many other tempting products (e.g., potato chips, cookies, ca ndy). How will her decision to resist purchasing the pastry influence her subsequent behaviors? Will her initial act of regulation encourage additional regulation (i.e., behavioral consistency) or will it encourage indulgence (i.e., behavioral inconsistency)? More importantly, how might we explain either course of action? Goal theory is a popular framework for unders tanding the relationship between the series of decisions faced by our supermarket shopper. For example, consider goal theories that rely primarily on a top-down goal guidance system (e.g., Dhar and Simonson 1999; Fishbach and Dhar 2005). A top-down goal guidance system activ ely manages the pursuit of competing goals. In a top-down goal guidance system, people (1) se t performance standards for important goals, (2) monitor their progress toward these goals, an d (3) engage in and disengage from the pursuit of these goals over a period of time so as to ma intain a balance in their lives (i.e., the singleminded pursuit of one goal would be unhealthy). The top-down goal guidance system has been used to explain menu choices (e.g., Dhar a nd Simonson 1999), the balance between academic and entertainment activi ties (e.g., Fishbach and Dhar 2005), th e relationship between exercise and diet (Fishbach, Dhar, and Zhang 2006), and choices between virtues and vices (Khan and Dhar 2007). Thus, a top-down goal guidance model would explain our supermarket consumers behavior as a series of well-reasoned decisions. There is evidence that the top-down goal guidan ce system is not sufficient to explain all goal-directed behavior. For exampl e, people often exhibit regulato ry failure when their top-down

PAGE 9

9 system fails to maintain vigilance over their be haviors (Baumeister et al. 1998). This implies that there is a second system that has the potential to influence behavi oral consistency and inconsistency. A system that is suggesting of a bottom-up influen ce on behavior is proposed in Goal Systems Theory (Kruglanski et al. 2002). Goal Systems Theory argues that a goal can be represented as a cognitive structur e and that people will select means that are most strongly associated with activated goals. Thus, Goal Sy stems Theory provides a framework for explaining how consumers might engage in a behavior without the participation of a top-down goal guidance system. I seek to provide an account of how a bo ttom-up goal guidance system might manage goal pursuit and influence behavioral consistency an d inconsistency. Although Goal Systems Theory can explain behavior in the ab sence of top-down goal guidance, it does so without making any assumptions about guidance. I will provide evid ence for a bottom-up goal ac tivation system that has a guidance component. I will show that this guidance system is passive, that it relies on the activation and inhibition of competing goals, and that it is sensitive to the framing of a means (i.e., the meaning of the behavior). More specifically, I will show that behavioral consistency is a consequence of a goal remaining active, and competing goals remaining inhibited, across a series of behaviors. In contrast, beha vioral inconsistency is a conse quence of goal achievement (i.e., completion), the release of the activated goal, and the rebound of inhibited goals. I seek to provide insight into how a passive goal guidance system can choos e behaviors from an available array of behaviors and use these behaviors to manage the pursuit of competing goals.

PAGE 10

10 CHAPTER 2 TOP-DOWN AND BOTTO M-UP GOAL GUI DANCE There has been considerable research on self -regulatory success and failure (Baumeister and Heatherton 1996; Carver and Scheier 1998). For example, Mann et al.s (2007) review of 31 dieting studies shows that people manage to achieve a weight loss of at least five-10% in the first six months of dieting, yet they gain an average of 11 pounds over their original weight after five years. Similarly, an estimated 2.5 million alcoholic s in the United States have sought treatment. After receiving treatment, 51% of the patients experience at least one relapse by the end of the first month, half of the remaining patients relaps e by the end of the third month, and half of the remaining patients relapse by the end of the sixth month (Jones and McMahon 1994). Comparable rates of relapse exist for drug use, adultery, spousal abuse, over-shopping, and gambling. Previous literature suggests that two types of systems may be used to guide the pursuit of multiple goals and self-control (Kruglanski et al. 2002; Locke and Latham 1990). In a top-down behavioral guidance system, consum ers actively monitor their behavi or in relation to previously established goal standards (i.e., self-regulatory goals) and choose which behaviors to pursue based on their current goal state and these standa rds. In a bottom-up behavioral guidance system, the performance of behaviors incr eases the level of activation of the goals these behaviors serve and determine subsequent behaviors, indepe ndent of pre-determined goal standards. Top-Down Guidance Top-down goal guidance m odels seek to explai n goal pursuit (i.e., behavioral consistency) and goal disengagement (i.e., beha vioral inconsistency). For exam ple, Fishbach and Dhar (2005) propose that inferences of goal commitment encourage behavioral consistency, whereas inferences of goal progress encourage behavior al inconsistency. Dhar and Simonson (1999)

PAGE 11

11 argue that a tradeoff between a goal (e.g., pleasure) and a resource (e.g., money) encourages behavioral consistency, whereas a tradeoff between two competing goals (e.g., pleasure and health) encourages behavioral inconsistency. Kh an and Dhar (2006) show that people who make an initial choice that boosts a positive self-concept feel licensed to make subsequent choices that are more self-indulgent (i.e., regulatory success encourages inconsistency). Thus, top-down goal guidance models are active goal monitoring systems. I provide more details of these models next. Fishbach and Dhar (2005; see also Fishbach, Dhar, and Zhan 2006) propose that the determinant of consistency vs. inconsistency in two consecutive behavi oral decisions is the extent to which an initial behavioral decision signals goal progress or commitment. They posit that when people perceive an initial act as goa l progress, they become less likely to pursue the same goal, and end up pursuing opposing goals, when making the second decision. When people perceive an initial act as commitment to a goal, they become more likely to pursue the same goal when making the second decision. For exampl e, in Fishbach and Dhar (2005, Study 3), participants were asked to indi cate how much progress they fe el they have made (vs. how committed they feel they are) toward their academic success goal after having studied all day. They were then asked how interested they were in hanging out with th eir friends that night. Those who indicated how much progress they had made towards their academic success goal were more interested in hanging out with their friends than those who indicated how committed they were toward their academic success goal. Th erefore, people make an inference of how far they are from achieving a certain goal and, depending on whether this is an inference of progress or commitment, choose thei r subsequent behaviors. Dhar and Simonson (1999) propose that behavioral consistency and inconsistency may be the result of a different dynamic. They propose th at when there is a tradeoff between pursuing a

PAGE 12

12 goal (e.g., pleasure) and spending a resource (e.g. money), people tend to focus on either goal fulfillment or resource conservation by behaving consistently in each of two consumption episodes (e.g., focus on maximizing pleasure in on e episode and on saving money in the second episode). For example, someone who has two baseba ll tickets, one for a good seat and one for an average seat, will tend to get a more expensive b eer (as opposed to an average one) in the game where the good seat is being used. The consumer highlights the goal of enjoyment (i.e., good seat plus good beer) in one of the consumption epis odes and focuses on the goal of saving money (i.e., average seat plus average beer) in the second consumption episodes. A second proposition is that when there is a tradeoff between two goals (e.g., pleasure and health), people tend to pursue each of these goals within a single c onsumption episode. For example, someone who wants to enjoy their meal and be healthy will tend to skip desert on the day they had a tasty, fatty meal. The consumer tries to achieve balan ce by pursuing competing goals across multiple decisions. Apparent in this con ceptualization is consumers activ e monitoring of their behaviors in order to achieve a desire end state. Novemsky and Dhar (2005) propose that choices in a sequence depend on the level of goal fulfillment resulting from a previous choice. If a previous choice leads to more fulfillment (vs. less fulfillment) of a consumers goal, subsequent choices tend to be riskier. For example, someone who has had an enjoyable meal (vs. a less enjoyable meal) will be more likely to choose a wine that sometimes is good and sometimes is not (i.e., risky option) than a wine that is always good (i.e., riskless option). The authors argue that a good first experience produces an upward shift in the goal standard and leads to c hoices that can be more capable of fulfilling a goal given the new standard. Thus, people mon itor and shift their goal standards within a consumption episode, making choices to reach these standards.

PAGE 13

13 In addition to the conceptualizations pr esented above, there is evidence for goal monitoring, and deliberation, in a number of other consumer behaviors. For example, Kivetz, Urminsky, and Zheng (2006) show that people who are closer to achieving their goal in a rewards program (e.g., earning a free coffee) accelerate their purchases. With respect to deliberation, Okada (2005) found that a hedonic product option, as compar ed to a utilitarian option, is preferred when each option is presen ted by itself (i.e., people may have a default pleasure goal), but the utilitarian option is chosen when they ar e presented together. Similarly, Kivetz and Zheng (2006) find that people who perceive they have made great effort in an initial task are more likely to subsequently choose vi ces over virtues (e.g., choc olate cake vs. a fruit salad). These monitoring and deliber ation processes are also appare nt in optimistic expectations about future choices (Zhang, Fi shbach, and Dhar 2007) and lay theories of self-control (Mukhopadhyay and Johar 2005). Although top-down models differ with respect to the specifics of the monitoring process, the models do have some common assumptions. First, the tendency for a consumer to engage in consistent or inconsistent behavior is a function of an act ive monitoring system. Second, the active monitoring system has to manage the pu rsuit of competing goals. Third, managing the pursuit of competing goals requires that goal targets (i.e., performa nce standards) be established and that progress toward the targets be assess ed. Fourth, a person must be vigilant about competing goals if sufficient progress is going to be made toward all goals. Together, these commonalities explain how a top-down goal guidance system can promote behavioral consistency and behavi oral inconsistency. Bottom-up Guidance Bottom -up goal guidance models seek to expl ain how goal activation influences behavior. In these models, there are no goal standards. Consumers choose behaviors based on which goals

PAGE 14

14 are currently most activated. Goal Systems Theory (GST; Kruglanski et al. 2002) is an example of a bottom-up goal pursuit model. For example, Fishbach, Friedman, and Kruglanski (2003, Study 5) found that priming a temptation (e.g., fatte ning food) activates the goal of dieting in restricted eaters. Part icipants were either exposed to fattening food (chocolate bars, food magazines) or diet-related objects (diet magazi nes). Participants primed with fattening food estimated an even higher likelihood to avoid a number of fatty foods than di d participants primed with diet-related objects. These results are evid ence that goals can be pr imed by both means that share a facilitative relationship with them and me ans that share an inhibitory relationship with them. However, a word of caution about these re sults is important. They were obtained in a context where all participants were women c oncerned about their weight. Thus, these were participants who had activated th e goal of avoiding fattening f ood many times in the past, a goal that became more activated due to the priming manipulation. If participants were prone to activating the goal of satiating their hunger or of eating so mething sweet, seeing a chocolate bar could have activated these goals to a larger exte nt and led to opposite re sults from those obtained by Fishbach et al. (2003). Shah, Friedman and Kruglanski (2002) presen ted evidence of goal shielding. The authors had participants describe a current goal (e.g., an attribute they want ed to possess) and list alternative goals. In Study 2, they found that pa rticipants who were primed with one of the attribute goals they listed (e.g., intelligent) were slower to recognize another attribute (e.g., happy) as compared to those who were primed with presentation of control words. However, when priming participants were primed with an attribute that was considered to have a facilitative relation with a ta rget goal attribute (Study 4), Sh ah et al. (2002) found that the priming actually facilitated recognition as comp ared to priming with a control word. Goal

PAGE 15

15 shielding is an example of how current goal pursuits may inhib it the activation of means that share an inhibitory relati onship with these goals. Finally, the multifinality principle of GST al so sheds light on how consumers evaluate means and choose behaviors. More specifically, goa ls become activated, of ten without awareness by the consumer, and are capable of determini ng subsequent behavior. According to recent findings, multifinal means (i.e., means that can serve more than one goal) are more valued than unifinal means. Chun et al. (2005; Study 3) presented participants with the task of choosing the best tasting soda among products labeled Pepsi, Coke, and generi c cola. The sodas labeled Pepsi and Coke were actually the same soda (m ade by mixing equal proportions of shoppers-brand cola and water) while the generic cola had a better taste (no water was added) in one condition and a worse taste (a quarter of water was added) in another condition. Part icipants were primed without their awareness with one of two background goals: id entification or disidentification with the United States of America. When the gene ric cola had the best ta ste of all colas, it was chosen by most participants regardless of what the background goal was. When the generic cola had the worst taste of all colas a nd Pepsi and Coke had exactly the same taste, most participants chose Coke (a brand that was indi cated in pretests as being more American) when they were in the identification with the United States of Am erica condition, but chose Pepsi when they were in the disidentification with the United States of America condition. Thes e results indicate that when means have the same value in terms of one goal, other active goals will influence means evaluation and choice even when people are not aware of those goals. In another series of studies, Kpetz, Fishbach, and Kruglanski (2005) show that activation of multiple goals may narrow the set of acceptable means to a focal goal a nd constrain the consideration set of activities

PAGE 16

16 to ones that benefit all the active goals. These fi ndings show that activated goals may not shift goal standards, but guide behavior by changing perceptions of means efficacy. Although GST explains how activated goals can influence subsequent behavior, it does not explain behavioral consistency and inconsistency. GST provide s the basis for understanding goal pursuit in the absence of top-down guidance, but it is not a goal management model. A bottomup guidance model should be able to predict how consumers ma nage the pursuit of multiple goals and when they will behave consistently vs. inconsistently. Passive Goal Guidance The evidence presen ted above suggests that goa ls are represented in a cognitive structure. If so, goals should have many of the properties of other cognitive representations, including the potential for activation/inhibition via direct and indirect primes. Recent evidence confirms this prediction. First, goals can be activated by exposure to situatio ns (e.g., Aarts and Dijksterhuis 2000), other peoples goals (e.g., Aarts, Gollwitzer, and Hassin 2004), or means that are effective or ineffective for goal achievement (e.g., Fi shbach, Friedman, and Kruglanski 2003). Second, goals can form associations to other goals and non-goal cues (e.g., contextual stimuli) (Bargh et al. 2001). Third, these associations can be facilitatory or inhibito ry (Kruglanski et al. 2002). The passive goal guidance (PGG) model uses the cogn itive properties of goal structures to predict behavioral consistency and inconsistency. The passive goal guidance (PGG) model propos es a system that complements the topdown goal guidance system. The PGG model differs from a top-down model in important ways. First, there is no active monitoring of goal pursuit (e.g., Aarts, Custers, and Holland 2007; Bargh et al. 2001). The system responds to the environm ent based on the state of the system and the appeal of available means. Second, the PGG syst em can disengage from goals, and engage in other goals, without active de liberation (e.g., Aarts et al. 2007; Bargh, Chen, and Burroughs

PAGE 17

17 1996; Custers and Aarts 2007). This implies that goal pursuit and disengagement will depend on the interpretation of the means behavior (i.e., is th e behavior perceived to be in pursuit of a goal or in completion of a goal) as well as envir onmental cues that activate competing goals. Third, the system responds to goal activators within the environment (e.g., direct goal primes, means primes, observed behavior primes, prior behavi or) without being consciously aware of their influence (e.g., Aarts, Gollwitzer, and Hassi n 2004; Bargh and Gollwitzer 1994; Shah and Kruglanski 2003). Finally, the system is not simply an automated version of the well-practiced components of a top-down guidance system. If we assume that the top-down guidance system is rational and benevolent (i.e., it was meant to explain the pursuit of self-regulation goals), then the top-down system should have corrected maladaptive behaviors (e.g., compulsions, addictions, anti-social behaviors) before ceding them to bottom-up control. Maladaptive behavior is evidence of an independent bottom-up guidance system. The PGG model is grounded by the conceptualization of goals as knowledge structures (Dijksterhuis, Chartrand, and Aarts 2007; Kruglanski 1996). The PGG model describes how goals, behavioral plans, means, and behaviors interact to influe nce behavior over time. In this system, goals are represented using two informati on features: the semantic representation of the goal and the affective-motivational property of the goal (i.e., whether a goal is positive or negative). Hence, goals can be self-motivating without top-dow n guidance (Custers and Aarts 2007; Dijksterhuis et al. 2007). Beha vioral plans are the lower-order goals people pursue in order to achieve superordinate goals. Behavioral plans have also been called task goals or subgoals in order to differentiate them from superordin ate goals (e.g., Kruglanski et al. 2002). Means are opportunities to execute a behavior that will satisfy a goal. Behavi ors are the act of implementing the means.

PAGE 18

18 The PGG model addresses the processes by whic h people passively engage in goal pursuit. The model has two fundamental assumptions. First, the value of a means is a function of both its usefulness for achieving a goal and the activation of a goal. All thi ngs being equal, relative goal activation drives behavior. The work of Bargh is evidence of the role of activation of goal-related information on goal pursuit. Bargh et al. (2001) found that priming of achievement words (such as win, compete, succeed, strive,) in a word-search puzzle had a positive influence on participants performances on subsequent word -search puzzles as indicated by an increased number of words found. In a second experiment a similar priming manipulation involving the concept of cooperation led participants to sh are a common resource in a resource-management game. Therefore, once a goal is activated, encoun tering a proper means for goal achievement (in this case, the word-search puzzles) may lead to th e use of this means without an assessment of a goal standard. Second, the impact of a behavior, in terms of subsequent goal activa tion, is a function of the perceived meaning of the behavior. That is, if perceptions of the current state of a goal are cognitive representations, goal states should be sensitive to framing: the same behavior (means) can be perceived as aiding in the pursuit of a goa l(s) or completing a goal (s) (i.e., goal states are subjective experiences, not objec tive experiences). This assumption is necessary because a bottom-up guidance system does not set an object ive performance standard there is no topdown monitoring. For example, a diner can achieve the goal of reducing hunger by consuming the portion that is on their plate. As long as enviro nmental features can be used to activate goals, they can also be used to frame goal pursuit as ach ieving a goal. If the behavior is framed as an isolated act, the goal remains active. If the beha vior is framed as an end-state, the goal is achieved. Thus, if consuming the food on a plate is framed as eating one course, the reduce

PAGE 19

19 hunger goal will remain active. If consuming the f ood on a plate is framed as eating dinner, the reduce hunger goal will be achieved.1 Note that achievement occurs as a function of the framing of the behavior (i.e., what the goal means in the context of the task). If a person is dining out with the goal of enjoying a w onderful meal by a renowned chef, framing the main course as sampling the chefs cooking keeps the goal activ e, whereas framing the main course as the signature dish may achieve the goal. Even tho ugh the person is aware that there is a dessert course coming up, signature dish framing is more likely to be pe rceived as a behavior that leads to goal achievement. Previous theories featuri ng some bottom-up behavioral guidance properties (e.g., GST; Kruglanski et al. 2002) do not offer a ssumptions of how behaviors achieve goals in the goal system. These two assumptions lead to propositions 1 and 2. P1 : As goal activation increases (decreases), m eans that are relevant to the goal will gain (lose) value. P2 : The perception of the efficacy of a beha vior influences its impact on continued goal activation or goal release (i.e ., goal deactivation owing to goal achievement or frustration). In order to pursue goals, a passive goal gui dance system must have a process for (1) encouraging goal pursuit and (2) for maintaining balance in the pursuit of goals. Although these system objectives are similar to those found in a top-down guidance system, they are achieved differently owing to the constraints of a botto m-up system. For example, a bottom-up system cannot use focused attention to force goal pursuit or to coordinate goal balancing. Instead, the system must rely on the factors that are generally adaptive. With respect to goal pursuit, the PGG system can inhibit the activation of competing go als to ensure that the active goal is pursued 1 Note that physiological factors (e.g., satiation) will also influence the motivation to eat.

PAGE 20

20 appropriately. The inhibition of competing concepts is a property of semantic networks (Anderson, Green, and McCulloch 2000), hence th is property should be available to the cognitive representation of a goal system. For ex ample, the research on negative priming (Tipper 1985) offers evidence that concepts that are not relevant to goal pursuit may be inhibited. Negative priming is observed in priming experi ments where at time 1 a certain object is a distractor and its internal representation needs to be inhibited. When the tasks at time 2 involve those previously inhibited repr esentations, performance in th ese tasks are impaired. Tipper, Weaver, and Houghton (1994) show that negative pr iming is driven by the behavioral goals of the task, and that only properties of the distractor that direct compete with the target in terms of the goals to be achieved are inhibited. Therefore, in order to achieve goa ls, the system not only increases the activation of facilitative means, but also inhibits irrelevant means. With respect to goal balancing, goal achievement results in the release of the recently active goal and the activation of recently inhibi ted goals, similar to a rebound effect (for supporting evidence, see Macrae et al. 1994). Although there is not much research on the influence of post-achievement goal release and competing goal rebound on means evaluation, there is evidence that a goal becomes less ac cessible after it has been achieved. Frster, Liberman, and Higgins (2005) present a series of experiments where participants had their goals either unfulfilled or fulfilled at time one. In subsequent lexica l decision and Stroop-like tasks, goal-related words were more accessible if the goal was still unfulfilled and less accessible if the goal had been fulfilled. In addition, research on the rebound effect (Macrae et al. 1994; Wegner and Pennebaker 1993) suggests that the inhibitio n of thoughts related to a certain object may result in increased accessibility of this type of thought once the inhib itory process ends. The PGG model proposes that the activ ation of the recently inhibited goal must be sufficient to

PAGE 21

21 encourage goal pursuit, suggesting that the activ ation of the inhibited goal rebounds to a level greater than the level of activation in a stea dy state. In a steady, re sting state, means (e.g., different food items) related to a certain goal (e.g., eating) would ha ve the same value after initial goal pursuit (e.g., eating a tasty de sert) and goal achievement. If previously inhibited goals rebound, initial goal pursuit (e.g., eating a tasty desert ) and goal achievement should result in the rebound of inhibited goals and inconsistent behavior (e.g., willingness to eat healthy food). Thus, the PGG system engages in balancing as a natu ral consequence of the inhibition of competing goals during the pursuit of a mo st active goal. This leads to two process propositions: P3 : Goal activation will result in the inhibition of competing goals. P4 : Goal achievement will result in release of a target goal (goal deactivation) and rebound (i.e., activation) of recently inhibited goals. These propositions are tested in six experiments. A pilot experiment shows that some behaviors lead to consistent behavior whereas other behaviors l ead to inconsistent behavior. Experiment 1 shows that when an initial be havior is insufficient (sufficient) for goal achievement, it encourages consistent (inconsis tent) subsequent behavi ors. Experiment 1A provides support for a bottom-up goal guidance system that relies on goal activation and inhibition. Experiment 2 forces people to use a bottom-up or a top-down guidance system and shows that each system leads to a different pattern of behavior. Experiment 3 shows that recent behavior engages the bottom-up guidance system, whereas a prediction based on hypothetical behavior engages the top-down gui dance system. Experiment 4 shows that a bottom-up guidance system can influence behaviors that a t op-down guidance system deems irrelevant.

PAGE 22

22 CHAPTER 3 PILOT EXPERIMENT The PGG model m akes a number of assump tions about how a bottom-up goal guidance system influences behavior differently from a top-down goal guidance system. I begin with a demonstration that bottom-up guidance leads to different behaviors than might be expected from a top-down goal guidance system. Subsequent to the pilot study, I will conduct a more rigorous investigation of the propos ed PGG model and its assumptions about process. The pilot experiment consisted of three conditions. Participants were invited to eat one chocolate, as many chocolates as they wished, or were unaware of the chocolates (control condition). After devoting 25 minutes to completing two filler studies, the participant was invited to take an indulgent or healthy chocolate snack as a thank you fo r completing the studies. If the choice of snacks is driven by a bottom-up goal gui dance system, then prior consumption of a single chocolate should encourag e the choice of the indulgent treat. One act of indulgence activates goals associated with indulgence and encourages subsequent acts of indulgence (proposition 1). In contrast, prior consumption of multiple chocolates should encourage the choice of a healthy treat if the chocolates were consumed until the indulgence goal was achieved. Goal achievement results in goal release and the activation of competing goals such as regulate caloric intake (proposi tion 4). In order to investigate goal activation resulting from our manipulations, a separate pilot study measured reaction times to words related to indulgence and regulation after the chocolate c onsumption episode. Measurements were performed immediately, after 20 minutes, and after 40 minutes (within-subj ects), which allowed us to look at the duration of goal activation af ter a behavior has been performed. If the choice of snacks is driven by a topdown goal guidance system then eating one vs. several chocolates should have no influence on th e choice of a treat. Top-down goal guidance is

PAGE 23

23 not used to guide or encourage indulgent behavior, at least as an initial act. In top-down regulatory models, indulgent behavi or is either a reward for vi rtuous behavior or a relapse when virtuous behavior depletes resources. Method Participants and Design Participants were 164 undergraduate students who participated in the experim ent for extra credit. The design was a one-fact or (goal state), three-level (activated, achieved, control) between-subjects design. Procedure Participan ts were invited into a lab and seated in a private carrel. Participants in the two treatment conditions were told th at the first study investigated th e influence of eating chocolate on their mood. They were asked to assume th at they had come home and found a chocolate truffle in the kitchen. They were then told to gr ab a bag from the left-ha nd side of their carrel, open it, remove the chocolate truffle, and eat it. All participants ate the truffle. Participants in the activated goal condition were then asked about their mood and performed two filler experiments. Participants in the achieved goa l condition were told that they had just realized there were additional truffles at home. They were then told to grab an additional bag of truffles located on the right-hand side of their carrel and eat as many truffles as they wanted. When these participants were done eating, they were aske d questions about their mood and performed the filler experiments. Participants in the control condition did not participate in the truffle study. These participants were drawn from the same subject pool and were run independently. These control condition sessions were always run immediately before or after the tr eatment session in an effort

PAGE 24

24 to limit time of day effects on snack choice. All treatment participants were given water during the experiment. After completing the filler studies, participants were told that they would receive a gift for their participation in the series of experiment s. Participants were asked to choose between a Chips Ahoy chocolate chip granola bar and a Chips Ahoy packet of ch ocolate chip cookies. These two snacks were selected because they both contained chocolate (i.e ., there would not be a bias against one snack owing to satiation with the taste of chocolate) and because both snacks had similar calories (100 calorie s), although participants were not made aware of this fact. Stimuli The truf fles that participants ate were Lindt milk chocolate truffles. The two snacks were chosen based on their healthine ss. It was important to choose two snacks that had a similar description and taste but that va ried in their perceived degree of healthiness. A pretest conducted with a separate sample from the same partic ipant population (n = 58) asked participants to indicate how healthy they perceived each snack to be using a scale ranging from 1 (Very unhealthy) to 9 (Very healthy). The chocolat e chip granola bar was considered healthier ( M = 5.66) than the chocolate chip cookies ( M = 2.98, t (57) = 11.24, p < .01). Results Manipulation Check and Control Tests Participan ts in the ac hieved goal condition ( M = 3.20) ate significantly more truffles than participants in the activated goal condition ( M = 1.00; t (100) = 19.15, p < .01). No measurement of mood was conducted for the control condition. There were no mood differences across the goal activated ( M = 6.22) and the goal achieved conditions ( M = 6.62, t (101) = 1.89, p = .17).

PAGE 25

25 Analysis There was an effect of goal state on the choice of snack (2(2) = 10.39, p < .01). Participants in the activated goa l condition (38.7%) were less likel y to choose a granola bar than participants in the control condition (51.7%, 2(1) = 3.52, p < .05, 1-tailed). Part icipants in the achieved goal condition (70.5%) were more likely to choose a granola bar than participants in the control condition ( 2(1) = 3.65, p < .05, 1-tailed). Post-session debriefing indicated that no participant guessed the purpose of the experime nt or made a connecti on between the initial chocolate tasting and the choice of a gift at the en d of the session. Supporting Evidence I used the sam e manipulations of the main pilot experiment with a separate group of participants (n = 166) and measur ed implicit goal activati on in order to suppor t the claims of the passive goal guidance model. After the chocolate consumption episode, participants were told that they would perform an atte ntion task involving decisions a bout whether certain letter strings were or were not words. This task allowed me to assess the goal activation and inhibition mechanism underlying the pattern of means desira bility. I told particip ants to focus their attention on a fixation point (an X) on the center of the comput er screen. The fixation point disappeared in an interval varying between 0 and 2 seconds (randomly determined) and was replaced by a letter strin g. Participants had to press if th e letter string was a word and if it was not, responding as quickly and accurately as possible. Participants performed ten practice trials and then responded to te n indulgence-related words (delicious, enjoy, gourmet, delight, pleasure, savory, good, desire, indulge, juicy) and ten regulation-related wo rds (calories, slim, regimen, fresh, weight, control, fit, health, ex ercise, workout). These 20 target trials were presented in random order along with 10 irreleva nt-word trials (aimed at avoiding suspicion about the real goals of the task), and 30 nonword trials. In order to exam ine the duration of goal

PAGE 26

26 activation, the procedure was repeated three times using the same words each time: immediately after the consumption episode, after 20 minutes, and after 40 mi nutes. The dependent measure was the latency for recognition of each word (see chapter 5 fo r a more detailed procedure). The results are presented in figure 3-1. A repeated-measures ANOVA revealed a significant 3-way interaction of goa l state, delay, and word type ( F (4, 326) = 3.31, p = .01). The goal state by word type interaction was significant immediately after consumption ( F (2, 163) = 5.41, p < .01). Participants were fast er to identify words related to indulgence than words related to regulation in the ac tivate goal state condition ( MIndulgence = 672 ms, MRegulation = 702 ms; F (1, 163) = 4.55, p < .05). Participants were faster to identi fy words related to regulation than words related to indulgence in the achieved goal state condition ( MIndulgence = 691 ms, MRegulation = 656 ms; F (1, 163) = 5.63, p < .05). There was no significant difference in the control condition ( MIndulgence = 656 ms, MRegulation = 640 ms; F (1, 163) = 1.09, p = .30). The goal state by word type interaction wa s significant 20 minutes after consumption ( F (2, 163) = 7.99, p < .01). Participants were faster to identi fy words related to indulgence than words related to regulation in the activate goal state condition ( MIndulgence = 604 ms, MRegulation = 632 ms; F (1, 163) = 10.11, p < .01). Participants were faster to identify words related to regulation than words related to indulgence in the achieved goal state condition ( MIndulgence = 625 ms, MRegulation = 604 ms; F (1, 163) = 5.54, p < .05). There was no significant di fference in the control condition ( MIndulgence = 618 ms, MRegulation = 632 ms; F (1, 163) = 2.13, p = .15). The goal state by word type interaction was not significant 40 minutes after consumption ( F (2, 163) = 1.27, p = .28). Participants were marginally slower to identify words related to indulgence than words related to regulati on in the activate goal state condition ( MIndulgence = 639 ms, MRegulation = 609 ms; F (1, 163) = 3.61, p = .06). There was no significant difference in the

PAGE 27

27 achieved goal state condition ( MIndulgence = 607 ms, MRegulation = 592 ms; F (1, 163) = .85, p = .36) and in the control condition ( MIndulgence = 601 ms, MRegulation = 608 ms; F (1, 163) = .18, p = .67). Discussion Results of the pilot experiment are consistent with the existence of a bottom-up goal guidance system. I argue that the consumption of one truffle sustained the activation of the indulgence goal, which in turn enco uraged a person to select an indulgent treat 25 minutes later. I also argue that the consumption of multiple tr uffles allowed the participant to release the indulgence goal. As a consequence, the compe ting regulation goal rebounde d and the participant became more likely to select a healthy trea t 25 minutes later. The reaction time evidence provides initial support for my claim of goal activation and inhibition. Words related to an indulgence goal concept were more accessible than words related to a regulation goal concept in the activated goal condition, but the opposite was true in the achieved goal condition. This pattern of activation was still present after 20 mi nutes, but not after 40 minutes. This is evidence of the temporary nature of the goal pursuit pro cess that the passive goal guidance model seeks to explain. Overtime these goals tend to be released even if they are not achieved (i.e., goal frustration). Interestingly, there seems to be a rebound effect in the goal activated condition after 40 minutes (i.e., marginally faster reaction times for words related to regulation). The nature and extension of this rebound effect will be inve stigated further in the next experiments. The results are not consistent with top-dow n goal guidance. When queried, participants could not articulate a relationshi p between the eating of the truffles and their choice of snacks. There were no implied performance standards, no monitoring of goals, no attempts to balance food choices to allow for healthy living. In fact, participants became more willing to select an indulgent snack after eating one truffle. If participants were using a top-down goal guidance system, they should have balanced their behavi ors by selecting a healthy snack or by selecting

PAGE 28

28 according to their preference (i.e., matched the choice share of the c ontrol group). In this context, it does not make sense to discuss progress towa rd indulgence. Top-down goal guidance models were not designed to explain this type of behavior.

PAGE 29

29 A 672 691 656 702 656 640600 620 640 660 680 700 720 ActivatedAchievedControlReaction Time (ms) Indulgence Words Regulation Words 604 625 618 632 604 632600 620 640 660 ActivatedAchievedControlReaction Time (ms) Indulgence Words Regulation Words B 639 607 601 609 592 608580 600 620 640 ActivatedAchievedControlReaction Time (ms) Indulgence Words Regulation Words C Figure 3-1. Influence of activating and achieving an indulgence goal on the accessibility of goal concepts. A) measured immediately after c onsumption, B) measured 20 minutes after consumption, C) measured 40 minutes after consumption.

PAGE 30

30 CHAPTER 4 EXPERIMENT 1 The results of the pilot experim ent provide evidence that is consistent with a bottom-up goal guidance system. Experiment 1 will extend this evidence in two important ways. First, I will use an experimental design that allows me to concurrently investigate all the propositions of the PGG model. I will show that goal activation e nhances the value of means associated with pursuing the activated goal (pr oposition 1), that goal activatio n inhibits competing goals and degrades the value of means associated with pur suing these goals (proposition 3), that behaviors can be framed to be goal activating or goal ac hieving (proposition 2), and that goal achieving behaviors allow previously inhibi ted goals to rebound (activate) and enhance the value of means associated with pursuing these goals (propos ition 4). Second, I will measure goal activation (experiment 1A) in addition to the va luation of means (experiment 1). Experiment 1 modified the pilot experiment in three important ways. First, it had a symmetric manipulation of goal-act ivating behavior. Participants were encouraged to indulge (i.e., eat a chocolate) or to re gulate (i.e., avoid a chocolate) prior to engaging in the means evaluation task. If performing a behavior increases the level of ac tivation of an associated goal, these two behaviors should act ivate indulgence and regulation, respectively. Second, the goal achievement manipulation was changed so that the indulgence and regulation goals could be achieved subjectively, as opposed to objectively (pilot experiment). Third, after the initial behavior, participants were aske d to judge how much they want ed tasty, healthy, and neutral food items, which allowed me to assess patterns of desirability for each type of means separately.

PAGE 31

31 Method Participants and Design One hundred sixty-two undergraduate students pa rticipated in the experim ent for extra credit. The design was a 2 (behav ior: indulge vs. regulate) by 2 (subjective goal state: activated vs. achieved) by 3 (means type: tasty food, h ealthy food, neutral food) mixed design. The behavior and subjective goal stat e factors were manipulated betw een-subjects, while means type was manipulated within-subjects. Procedure Participan ts followed a truffle tasting procedure similar to that of the pilot experiment. In the activated goal state condition, pa rticipants were exposed to a chocolate truffle and were told that they decided to eat (not to eat) it. Pa rticipants in the achieved goal state condition were exposed to the same chocolate truffle and were to ld that they allow themselves to eat one treat per day (indulge condition) or that not eating the truffle al lowed them to accomplish their healthy eating goal (regulate condi tion). Participants were told that they would now perform a second task, a means evaluation task. The means eval uation task instructions told participants to assume that they were looking for something to eat and asked them to indicate how much they wanted tasty, healthy, and neutral food items. Pa rticipants saw pictures of 30 food items and rated how much they wanted each food item on a s cale ranging from 0 (I do not want it) to 100 (I really want it). After this task, all participants responded to questions about their perceptions of the purpose of the experiment. Stimuli The chocolate truffle was the sam e Lindt tr uffle used in the pilot experiment. The evaluation task stimuli were 30 pictures of food accompanied by a word description. Ten items were healthy (i.e., salad, apple, carrot, whole wheat bread, low-fat yogurt, cereal, wheat thins,

PAGE 32

32 rice cake, rice, and pear), ten were tasty but fa tty (i.e., hot dog, ice cream, cheese sticks, potato chips, Doritos, pizza, baked potato skins, French fries, doughnuts, and apple pie), and ten were neutral items (i.e., ham, beans, crackers, omel et, bread, cheese, Jell-O pickles, bologna, and waffle). These items were chosen based on a pr etest with 35 participants from the same participant population as that of the main expe riments, who were asked to rate each food on a scale ranging from 0 (Very unhealthy) to 100 ( Very healthy). The healthy items were rated significantly healthier ( M = 65.0) than the neutral items ( M = 48.4; F (1, 33) = 130.0, p < .01) and the neutral items were rated significan tly healthier than the tasty items ( M = 18.7; F (1, 33) = 210.25, p < .01). A second pretest indicated that th e healthy and the tasty items were equally liked (MHealthy = 51.4 vs. MTasty = 50.5; F < 1), which ensures that any differences in preferences for healthy or tasty items are a function of th e experimental manipulations. The neutral items were less liked ( MNeutral = 42.0) than the tasty or healthy items but this should not impact our results. Neutral items were only included to conf irm that certain conditions were not leading to satiation or mood states th at were influencing the ove rall preference for food. Results Control Tests There was no effect of behavior ( MIndulge = 6.49, MRegulate = 6.90; F (1, 107) = 2.39, p > .10) or subjective goal state ( MActivated = 6.70, MAchieved = 6.69, F < 1) on the participants mood level. Analysis Means are presented in f igure 4-1. A repeat ed-measures ANOVA revealed a significant 3way interaction of behavior, subjec tive goal state, and means type ( F (2, 316) = 23.14, p < .01). The behavior by means type interaction was significant in the activ ated goal condition (F (2, 318) = 10.27, p < .01). Participants assigned higher ratings to the tasty food items than to the healthy food items in the indulge condition ( MTasty = 61.1, MHealthy = 50.4; F (1, 158) = 6.12, p = .01).

PAGE 33

33 Participants assigned higher rati ngs to the healthy food items than to the tasty food items in the regulate condition ( MTasty = 49.7, MHealthy = 61.0; F (1, 158) = 6.49, p = .01). The manipulation did not influence the ratings of the neutral food items ( MActivated = 44.1, MAchieved = 43.3; F < 1). The behavior by means type interaction was significant in the ach ieved goal condition ( F (2, 318) = 13.41, p < .01). Participants assigned lower ratin gs to the tasty food items than to the healthy food items in the indulge condition (MTasty = 44.2, MHealthy = 56.8; F (1, 158) = 10.26, p < .01). Participants assigned lower ratings to th e healthy food items than to the tasty food items in the regulate condition ( MTasty = 58.9, MHealthy = 49.2; F (1, 158) = 6.26, p = .01). The manipulation did not influence the means of the neutral food items ( MActivated = 42.9, MAchieved = 44.0; F < 1). Discussion The results of experim ent 1 show that when people perform a goal-activating behavior, they value means associated with the target goa l to a greater extent (proposition 1) and means associated with the opposing goal to a lesser extent (pr oposition 3). When this same behavior is framed as goal achieving, means associated with th e target goal are valued to a lesser extent and means associated with the opposing goals are favored to a greater extent (proposition 2 and 4). The results of experiment 1 are inconsistent with the results of a top-down goal guidance system for many of the same reasons as the pilo t study. No participant ar ticulated a relationship between the act of eating or resisting the truffle and the evaluation of the food items. The only coherent hypotheses were about the mood state induced by the ini tial behavior and the rating of all food items. This is a plausible inference on the part of the particip ants, although acting on the hypothesis could not have been responsible fo r our results. Also, t op-down guidance is not designed to promote commitment to indulgences, thus there is no reason to assume that participants in the indulge-goal activated condition were committed to indulgence. Yet, a

PAGE 34

34 proponent of top-down goal guidance systems could adjust assumptions about the model to help it account for these data. For example, it could be argued that participants have implicit standards for the consumption of fattening and healthy food a nd that they monitor beha vior with respect to these standards on an ongoing basis. This monito ring, although implicit, allows them to achieve a balance in their food consumption. I test this possibility in experiment 1A.

PAGE 35

35 A 61.1 49.7 50.4 61.0 44.1 43.330 40 50 60 70 80 IndulgeRegulateValue at Time 2 Tasty Healthy Neutral 44.2 58.9 56.8 49.2 42.9 44.030 40 50 60 70 80 Indulge RegulateValue at Time 2 Tasty Healthy Neutral B Figure 4-1. Influence of indulging / regulating and subjectively ac tivating / achieving a goal at time 1 on the subsequent desire for food. A) goal activated at time 1, B) goal achieved at time 1.

PAGE 36

36 CHAPTER 5 EXPERIMENT 1A The PPG model m akes specific predictions ab out goal activation and how it relates to means activation. Proposition 1 posits that a more (less) activated goal leads to higher (lower) evaluation of associated means. If this is so, the pattern of results observed in experiment 1 should have corresponding levels of goal activation. For example, when a behavior is framed as an act of indulgence (regulation), an indulgence (r egulation) goal should be more activated than a regulation (indulgence) goal (see propositions 2 and 3). When a behavior is framed as achieving regulation (indulgence), an indulgence (re gulation) goal should be more activated than a regulation (indulgence) goal (see propositions 2 and 4). In general, top-down models of goal guidance do not make predictions about patterns of goal activation. This is reasonable because top-do wn models are meant to discuss goal balancing that occurs over the course of a day, month or year (i.e., predictions a bout goal activation are too molecular). One top-down model that could make a prediction about goal activation is the goal commitment goal progress model of Fishbach and Dhar (2005). If people are monitoring competing goals, then both goals should be e qually active during assessments of progress (e.g., I made progress toward goal X so now I can pursue goal Y). Experiment 1A collected two additional meas ures. Owing to the difficulty of inferring goal activation predictions from current specificati ons of top-down models, inferences of goal commitment and goal progress toward healthy eat ing were collected. Owing to a concern that specific emotional states might be influencing the evaluation of specific classes of food (e.g., Ramanathan and Williams 2007), a more detail ed set of measures was administered.

PAGE 37

37 Method Participants and Design One hundred twenty-six undergradu ate students participated in the experim ent for extra credit. The design was a 2 (behav ior: indulge vs. regulate) by 2 (subjective goal state: activated vs. achieved) by 3 (word type: indulgence vs. regulation) mi xed design. The behavior and subjective goal state factors were manipulat ed between-subjects, while word type was manipulated within-subjects. Procedure and Stimuli I used the sam e procedure and stimuli as those of experiment 1 with th ree exceptions. First, instead of providing food desirabili ty ratings, participants were to ld that they would perform an attention task involving decisions about whether certain letter stri ngs were or were not words. This reaction time task followed the exact procedure as that of the additional evidence provided in the pilot experiment. Second, I measured specific em otions after participants pe rformed the lexical decision task. Following a procedure used by Ramanathan and Williams (2007), I asked participants to which degree (1 Not at all to 9 Very much ) they were experiencing a series of emotions in response to having eaten (not eaten) the choc olate truffle. Participants were asked about positive (e.g., happy) and negative (e.g., frustrated) hedonic emotions, and positive (e.g., proud) and negative (e.g., guilty) self-conscious emotions. Third, after measuring emotional states, I asked participants in the activat ed goal conditions how committed they were to healthy eating after performing the initial behavior (1 Not committed at all, 9 Very committed). I asked participants in the achieved goal conditions how much progress they thought they had made toward healthy eating after performi ng the initial behavior (1 No progress at all, 9 A lot of

PAGE 38

38 progress). This last measure was designed to provide an additional test of the differences between the bottom-up and the top-down goal guidance systems. Results Emotions In order for specific em otions to account of th e results of experiment 1, there must be a behavior by subjective goal state interaction on one of the four m easures or an interaction of these two variables with the hedoni c contrast (e.g., positive versus negative) or the self-conscious contrast (e.g., positive versus negative). These three-way and two-way interactions were not significant ( F < 1). Interestingly, positive hedonic emotions (e.g., happiness) were higher in the indulge condition ( M = 5.13) than in the regulate condition ( M = 4.47; F (1, 122) = 6.03, p < .05) and negative self-conscious emotions (e.g., gui lt) were higher in th e indulge condition ( M = 3.35) than in the regulate condition ( M = 1.89; F (1, 122) = 16.60, p < .01). These results are consistent with previous findings about th e role of mixed emotions in consumer behavior (Aaker and Williams 2002). Reaction Times Only reaction tim es of correct identifications of a letter st ring as a word or a non-word were included in the analysis. I performed a na tural log transformation of all reaction times and those that exceeded 3 standard deviations from their cell mean were eliminated from the analysis (Bargh and Chartrand 2000; Fazio 1990). Similar to the food ratings in experiment 1, reaction times were averaged to generate one score for each type of word for each participant. Means are presented in figure 5-1. A repeat ed-measures ANOVA revealed a significant 3way interaction of behavior, subjective goal state, and word type ( F (1, 122) = 25.42, p < .01). The behavior by word type interaction was significant in the goal activated condition ( F (1, 123) = 11.63, p < .01). Participants were faster to iden tify words related to indulgence than words

PAGE 39

39 related to regulation in the indulge condition ( MIndulgence = 637 ms, MRegulation = 662 ms; F (1, 122) = 4.24, p < .05). Participants were fa ster to identify words rela ted to regulation than words related to indulgence in the regulate condition ( MIndulgence = 647 ms, MRegulation = 613 ms; F (1, 122) = 7.65, p < .01). The behavior by word type interaction was significant in the goal achieved condition ( F (1, 123) = 13.74, p < .01). Participants were slower to id entify words related to indulgence than words related to regulation in the indulge condition ( MIndulgence = 656 ms, MRegulation = 633 ms; F (1, 122) = 3.99, p < .05). Participants were slower to id entify words related to regulation than words related to indulgence in the regulate condition ( MIndulgence = 622 ms, MRegulation = 661 ms; F (1, 122) = 10.61, p < .01). These patterns of activation are not related to participants inferences of goal commitment or progress. In the activated goal condition, partic ipants did not indicate to be more committed to healthy eating after regulating ( M = 6.23) than after indulging ( M = 6.97; F (1, 122) = 2.09, p > .15). In the achieved goal condition, participants did not indicate to have made more progress toward healthy eating after regulating ( M = 4.78) than after indulging ( M = 4.32; F (1, 122) = 1.45, p > .15). Discussion The reaction time evidence of experiment 1A supports the PPG m odel of bottom-up goal guidance. When participants were engaged in initial goal pursuit, the corresponding goals were more activated. When the same behaviors were framed as goal achievement, the competing goals were more activated. Note that the procedure made no mention of the competing goal. Thus, it should not be surprising that self-reports of goal commitment and goal progress did not vary across the regulate and the indulge conditions during goal activation and goal achievement, respectively. The procedure does not encourag e the creation and mon itoring of performance

PAGE 40

40 standards, or the balancing of behaviors with respect to competing goals. Instead, the procedure takes advantage of the preexisting association between indulgent and regulatory behavior. Behavior, and inhibitory associa tions between goals, influence the patterns of activation between these goals and the valua tion of behaviors associat ed with these goals.

PAGE 41

41 A 637 647 662 613600 620 640 660 680 700 IndulgeRegulateReaction Time (ms) Indulgence Words Regulation Words 656 622 633 661600 620 640 660 680 700 IndulgeRegulateReaction Time (ms) Indulgence Words Regulation Words B Figure 5-1. Influence of indulgi ng / regulating and subjectively activating / achieving a goal at time 1 on the accessibility of goal concepts. A) goal activated at time 1, B) goal achieved at time 1.

PAGE 42

42 CHAPTER 6 EXPERIMENT 2 Experim ent 2 further explores the differen ces between the bottom-up PGG model and topdown guidance models. More specifically, I fo cus on the different processes proposed by the PGG model and the model proposed by Fishbach and Dhar (2005; Fishbach, Dhar, and Zhang 2006). The experiment investigat es goal pursuit when people are encouraged to monitor goal commitment or progress (e.g., top-down goal guida nce) versus when people are not encouraged to monitor goal commitment or progress (e.g., bottom-up goal guidance). When people monitor how an initial behavior impacts goal commitment or progress, commitment and progress will influence subsequent behavior (i .e., behavior is guided by a topdown system). When there is no monitoring of how an initial behavior impact s goal commitment or progress, subsequent behavior will be guided by a bottom-up goal guidance system (see pr edictions below). Method Participants and Design Participants were 160 undergraduate students who participated in the experim ent for extra credit. The experiment investig ated regulatory behavior. The design was a 2 (behavior: full regulation vs. partial regulation) by 3 (monito ring: none, commitment, progress) by 3 (means type: tasty food, healthy food, neutral food) mixe d design. The behavior and monitoring factors were manipulated between-subjects, while mean s type was manipulated within-subjects. Procedure and Stimuli The procedu re was similar to the goal-activated-regulate condition of experiment 1. The only differences were that (1) the bag contained three truffl es, (2) participants in the monitored commitment condition were asked to indicate their commitment to healthy eating (1 Not committed at all, 9 Very committed) prior to evaluating the 30 food items, (3) participants

PAGE 43

43 in the monitored progress condition were asked to indicate their progress toward healthy eating (1 No progress at all, 9 A lot of progress) prior to evaluating the 30 food items, and (4) participants in the no monitoring condition were asked to indicat e their progress toward healthy eating after evaluating the 30 food items by referring back to the chocolate episode. This measure was added because I was curious as to whether participants would make the same inference of progress without using this inferenc e in indicating food desira bility (i.e., the bottomup system does not monitor goal progress). Participants in the full regulation condition ate no truffles and participants in the partial regulation condition ate one truffle. Predictions The key predictions concerned a comparison of the monitored progress and monitored commitment conditions to the no monitoring condition. In the full regulation condition, the no monitoring condition should be equivale nt to a goal progress condi tion. As posited by Fishbach and Dhar (2005, p. 376), in a top-down system people have a general motivation to balance between goals. If there is top-down guidance in the no monitoring condition, participants should prefer tasty food to healthy food. In contra st, a bottom-up goal guidance system predicts participants in the no monitoring condition should remain consiste nt with their initial behavior and prefer healthy food to tasty food. In the partial regulation condition, there should be less inferred commitment or progress in the monitoring conditions, hence it is difficult to anticipate the degree of preference for either type of food. If there is topdown guidance, all that can be reasonably predicted is that the results of the no monitoring condition should match the results of one of the monitoring conditions. If there is bottom-up guidance in the no monitoring condition, participants should remain consiste nt with their initial behavior and prefer tasty food to healthy food.

PAGE 44

44 Results Commitment and Progress The m ean responses to the goal commitment a nd progress questions are shown in figure 61. There are three cri tical tests. First, participants in the monitored commitment condition were more committed to healthy eating in the full re gulation condition than in the partial regulation condition ( MFull = 6.96, MPartial = 4.17; F (1, 154) = 18.44, p < .01). Second, participants in the monitored progress condition perceived more progress toward healthy eating in the full regulation condition than in the partial regulation condition ( MFull = 5.90, MPartial = 2.60; F (1, 154) = 27.81, p < .01). Finally, participants in the no monitoring condition perceived more progress toward healthy eating in the full regulation condition than in the partial regulation condition ( MFull = 5.25, MPartial = 3.54; F (1, 154) = 7.35, p < .01). Interestingly, while in experiment 1, participants who did not eat the truffle did not indicate more progress toward healthy eating than those who ate the truffle, th ey did here. The fact that there were three chocolate truffles in this experiment may have influenced the accuracy of peoples inferences. Analysis Means are presented in f igure 6-1. A repeat ed-measures ANOVA revealed a significant 3way interaction of behavior, monitoring, and means type ( F (4, 308) = 10.32, p < .01). The monitoring by means type in teraction was significant ( F (4, 310) = 6.96, p < .01) in the full regulation condition. Participants assigned lower ratings to the tasty food items than to the healthy food items in the commitment condition ( MTasty = 47.1, MHealthy = 61.5; F (1, 154) = 8.07, p < .01). Participants assigned hi gher ratings to the tasty food ite ms than to the healthy food items in the progress condition (MTasty = 61.7, MHealthy = 51.9; F (1, 154) = 5.85, p = .01). Participants assigned lower ratings to the tasty food items than to the healthy food items in the no monitoring condition ( MTasty = 48.6, MHealthy = 62.7; F (1, 154) = 7.44, p < .01). There was no

PAGE 45

45 influence of monitoring on the ratin gs of the neutral food items ( MCommitment = 50.3, MProgress = 44.1, MNo_Monitoring = 46.8; F (1, 154) = 1.72, p > .10). The monitoring by means type interaction was significant ( F (4, 310) = 8.13, p < .01) in the partial regulation condition. Participants assigne d equal ratings to the tasty and healthy food items in the commitment condition ( MTasty = 54.9, MHealthy = 56.7; F (1, 154) = .12, p = .73). Participants assigned equal rati ngs to the tasty and healthy food items in the progress condition ( MTasty = 56.2, MHealthy = 59.1; F (1, 154) = .27, p = .615). Participants as signed higher ratings to the tasty food items than to the healthy food items in the no monitoring condition ( MTasty = 68.8, MHealthy = 45.8; F (1, 154) = 22.99, p < .01). There was no influence of monitoring on the ratings of the neutral food items ( MCommitment = 49.1, MProgress = 46.8, MNo_Monitoring = 42.0; F (2, 154) = 2.04, p = .12). Discussion Results support our claim of a bottom -up goa l guidance system. In the full regulation condition, participants in the no monitoring condition remained consistent with their initial behavior and preferred healthy food to tasty food. This result was in opposition to the prediction of the top-down goal guidance system. In the pa rtial regulation condition, participants in the no monitoring condition remained consistent with their in itial behavior and pr eferred tasty food to healthy food. This preference was inconsistent wi th the results of either of the goal monitoring conditions, suggesting that goal monito ring was not instrumental in the no monitoring condition results. Finally, no monitoring condition participants indicated that they made more progress toward healthy eating in the full regulation condition than in the partial regulation condition, yet preferred healthy food in the full regulation c ondition and tasty food in the partial regulation condition. These results are in direct opposition to the prediction of the top-down goal guidance system proposed by Fishbach and Dhar (2005).

PAGE 46

46 A 47.1 61.7 48.6 61.5 51.9 62.7 50.344.1 46.830 40 50 60 70 80 CommitmentProgressNone/ProgressValue at Time 2Monitoring Tasty Healthy Neutral 54.9 56.2 68.8 56.7 59.1 45.8 49.1 46.8 42.030 40 50 60 70 80 CommitmentProgressNone/ProgressValue at Time 2Monitoring Tasty Healthy Neutral B 6.96 5.90 5.25 4.17 2.60 3.5402 03 04 05 06 07 08 CommitmentProgressNone/ProgressAmount of Commitment/ProgressMonitoring Full Regulation Partial Regulation C Figure 6-1. Influence of full / partial regulation and monitoring on the subsequent desire for food. A) full regulation at time 1, B) partial regulation at time 1, C) inferences based on initial behavior.

PAGE 47

47 CHAPTER 7 EXPERIMENT 3 The objectiv e of experiment 3 is to investigate a boundary condition of the bottom-up and top-down goal guidance systems. The critical manipulation was whether people experienced a behavior or were told about the behavior of another person. I have shown that when people perform a goal-activating behavior, they inhibit competing goal s and tend to behave consistently. However, when people are told about the behavior of another person, I expect that this will allow for an inference that the behavior achieved th e goal and people will use a top-down guidance system to balance behaviors. Method Participants and Design Participants were 95 undergraduate students who participated in the experim ent for extra credit. The design was a 2 (behav ior: indulge vs. regulate) by 2 (type of behavior: experienced vs. hypothetical) by 3 (means type: tasty food, healthy food, neutral food) mixed design. Behavior and type of behavior were manipulated between-subjects, while the means type factor was manipulated within-subjects. Procedure and Stimuli In the exp erienced behavior conditions, the procedure replicated the goal-activated conditions in experiment 1. Participants in th e hypothetical behavior condition read the same scenario with two exceptions. First, the scenario described Mr. A ra ther than you (i.e., participants were told about the behavior of Mr. A). Second, they we re not exposed to the truffle. Except for these two changes, the hypothetical be havior conditions had the same procedure as the experienced behavior conditions. The means ev aluation task was slightly modified in the hypothetical behavior conditions so that ques tions were about Mr. As food desires.

PAGE 48

48 Pretest A pretest was conducted to confirm that the Mr A scenario encouraged participants to infer goal achievement. I presented the Mr. A scen arios (indulge vs. regulat e) to a separate group of participants (n = 24 ) and asked them to indicate to wh ich extent they thought Mr. A. had achieved his goal to indulge (eat healthy food) in the indulge (regulate) condition (1 Did not achieve it at all to 9 T otally achieved it). The mean response in the indulge ( M = 7.50) and the regulate (M = 7.75) conditions was significantly differe nt from the mid-point of the scale ( t (11) = 4.38, p < .01; t (11) = 6.42, p < .01). Results Means are presented in f igure 7-1. A repeat ed-measures ANOVA revealed a significant 3way interaction of behavior, type of behavior, and means type ( F (2, 182) = 32.48, p < .01). The behavior by means type interaction was signifi cant in the experience d behavior condition (F (2, 184) = 7.39, p < .01). Participants assigned higher rati ngs to the tasty food items than to the healthy food items in the indulge condition ( MTasty = 59.0, MHealthy = 40.7; F (1, 91) = 7.28, p < .01). Participants assigned equal ratings to the tasty and the healthy f ood items in the regulate condition ( MTasty = 45.0, MHealthy = 54.9; F (1, 91) = 2.11, p = .15). The difference, however, was in the expected direction. There was no influe nce of the manipulation on the ratings of the neutral food items ( MIndulge = 41.7, MRegulate = 38.8; F (1, 91) = .72, p = .40). The behavior by means type interaction was significant in the hypothetical behavior condition (F (2, 184) = 28.39, p < .01). Participants assigned hi gher ratings to the healthy food items than to the tasty food ite ms in the indulge condition ( MTasty = 32.7, MHealthy = 66.7; F (1, 91) = 28.41, p < .01). Participants assigned higher ratings to the tasty food items than to the healthy food items in the regulate condition ( MTasty = 63.0, MHealthy = 45.1; F (1, 91) = 8.25, p < .01).

PAGE 49

49 There was no influence of the manipulation on the ratings of the neutral food items ( MIndulge = 45.1, MRegulate = 48.5; F (1, 91) = 1.10, p = .30). Discussion The results of experim ent 3 show that when participants experience a goal-activating behavior, they behave consisten tly with the goal and the initial behavior. When participants are told about a hypothetical behavior they anticipate inconsistent behavior. I argue that a hypothetical behavior does not allow a person to subjectively experience a goal state, hence the person assumes goal achievement and predicts a person will value c ountervailing behaviors more. This prediction is consistent with the be havior that would be encouraged by a top-down goal guidance system. It is also important to note that the hypothetical behavior result replicates Dhar and Simonson (1999). Dhar and Simonson show that a prediction task involving a goal tradeoff results in inconsistent behavior. When there is a conflict between health and pleasure goals, people predict that others will behave inconsistently af ter pursuing one of the goals (e.g., choose a tasty dessert after having eaten a healthy main course).

PAGE 50

50 A 59.0 45.0 40.7 54.9 41.7 38.830 40 50 60 70 80 IndulgeRegulateValue at Time 2 Tasty Healthy Neutral 32.7 63.0 66.7 45.1 45.1 48.530 40 50 60 70 80 IndulgeRegulateValue at Time 2 Tasty Healthy Neutral B Figure 7-1. Influence of indulging / regulating and experiencing / be ing told about a behavior at time 1 on the subsequent valuation of food. A) experienced behavi or at time 1, B) hypothetical behavior at time 1.

PAGE 51

51 CHAPTER 8 EXPERIMENT 4 The bottom -up goal guidance model predicts that a behavior influences goal activation, which in turn influences the valuation of associat ed means. If this is tr ue, resisting a chocolate truffle, as compared to eating it, should not only increase desirability fo r healthy foods but also decrease desirability for any means that ar e ineffective in achievi ng a regulation goal (e.g., expensive products that bring status to peopl e). The opposite pattern is expected among those who eat a chocolate truffle. In contrast, the top-down goal guidance model should deem behaviors that are unrelated to the initial behavior irrelevant. For example, if a person avoids eating a chocolate, the person should make an inference of commitment or progress toward healthy eating. This inference s hould not influence how much the person want s status products. Method Participants and Design Participants were 170 undergraduate students who participated in the experim ent for extra credit. The design was a 2 (behav ior: indulge vs. regulate) by 2 (means type: healthy food vs. status products) between -subjects design. Means type was made a between-subjects factor owing to the different contexts in which these two types of items would be evaluated, in this experiment, and purchased, in a real-life situation. Procedure and Stimuli The procedure replicated that of the activated goal conditions of the previous experim ents with one difference. Participants in the status products condition were told to assume that they went shopping and to indicate how much they wanted each of ten products: expensive new watch, dressy designer shirt, A pple laptop, expensive jewelry, GPS system for your car, cruise on a high-end ship, new digital camera, name brand pair of jeans, home theater, and 49-inch

PAGE 52

52 plasma TV. A pretest indicated that these produ cts were considered symbols of status. The stimuli were the same as those of the previous experiments in the healthy food condition (10 pictures of healthy food items). Results Means are presented in f igure 7-1. A 2 (b ehavior) x 2 (means type) ANOVA revealed a significant interaction of these factors ( F (1, 166) = 20.41, p < .01). There is a significant effect of behavior in the healthy items condition. Part icipants in the regul ate condition assigned significantly higher ratings than part icipants in the indulge condition ( MIndulge = 44.5, MRegulate = 54.0; F (1, 166) = 7.67, p = .01). There is also a significant effect of be havior in the status products condition. Participants in the indulge condition assigned signi ficantly higher ratings than those in the regulate condition ( MIndulge = 57.8, MRegulate = 46.0; F (1, 166) = 13.33, p = .01). Discussion Experim ent 4 provides evidence that the perfor mance of a behavior l eads to goal activation and (de)valuation of an array of behaviors. Resisting a chocolate truffle not only led to more desire for healthy food, but also to less desire for status products. These results are not consistent with a top-down goal guidance system. If participants made inferences of commitment toward a goal based on their initial behavior, this would on ly influence desirability ratings if participants believed that eating/avoiding a c hocolate truffle were connected to status products. Posttest questioning of the participants indicated that 80.4 % of the participants in the status product condition saw no connection between their initial behavior and thei r product ratings. Most of the remaining participants mentioned a relations hip between their mood state and the product ratings, but this was usually in the opposite direction (i.e., not eating the truffle made me want those products to compensate for it).

PAGE 53

53 57.8 46.0 44.5 54.030 40 50 60 70 80 IndulgeRegulateValue at Time 2Means Type Status Healthy Figure 8-1. Influence of indulging / regulating at time 1 on the s ubsequent valuation of products in different contexts.

PAGE 54

54 CHAPTER 9 GENERAL DISCUSSION Past res earch has identified factors that infl uence behavioral consistency and inconsistency in the pursuit of self-control (e.g., Baumeister et al. 1998; Fi shbach and Dhar 2005; Kahn and Dhar 2006; Dhar and Simonson 1999; Kivetz and Zheng 2006). Evidence from six experiments supports a bottom-up, passive goal guidance model of goal pursuit. An initial behavior increases the level of activation of the goal it serves and has a positive impact on the value of related means. Yet, if the initial behavior results in ob jective or subjective goal achievement, the goal is released and competing goals rebound, leading to an increased valuation of competing means and a decreased valuation of previously valued means (pilot experiment and experiment 1). Behavioral consistency and incons istency are the result of goal concept activation and inhibition processes (experiment 1A), and can be guided by a bottom-up goal guidance system rather than a top-down goal guidance system (experiment 2). When people predict behavior based on a hypothetical behavior rather than perform it, they pr edict that the behavior will lead to behavioral inconsistency (experiment 3). This suggests th at hypothetical behaviors may be perceived as achieving a goal. Finally, the activ ation of a goal (e.g., indulgence) through an initial behavior may lead to an increase in the desire for mean s (e.g., status products) asso ciated with a related behavioral plan (experiment 4). Notably, the enti re pattern of results is explained by the bottomup PGG model. Theoretical Implications and Future Research These findings integrate and qualify several prev ious f indings. First, the results in the activated goal conditions of all experiments are at odds with an ego depletion hypothesis (Baumeister et al. 1998). I show that an ini tial act of regul ation may actually improve selfregulation. It is only when the goa l of regulation is ach ieved that inconsistent behavior occurs.

PAGE 55

55 This suggests that some of the behavioral inconsistency that has been attributed to ego depletion may be a consequence of goal achievement. S econd, the licensing hypothesis advanced by Khan and Dhar (2006) implies inconsistent behavior (e.g., donating less money to charity) after an initial behavior (e.g., committing to helping a stud ent) boosts a positive se lf-concept. The present results suggest that licensing may not occur un til an initial behavior moves the cognitive representation of a goal state fr om activated to achieved. The findings also indicate that the pairi ng of a goal and a behavior may moderate behavioral consistency and inconsistency by a pr ocess other than inferences of goal commitment and goal progress (Fishbach and Dhar 2005; Fish bach, Dhar, and Zhang 2006). Goal activation and goal achievement vs. goal commitment a nd goal progress seem to be complementary processes moderated by the extent to which peopl e actively monitor their goal state, compare it to a desired goal state, and make an inference ab out this state based on pa st behavior. As evident in experiment 2, the presence of such top-down, inferential activities may help regulation when people perceive that past (indul gent) behavior has hindered th eir progress toward regulation. When behavior guides goal activation in a botto m-up fashion, the absence of inferences about goal commitment and goal progress may lead to more indulgent behavior after an act of indulgence than after an act of re gulation. Future research is needed to investigate situations that motivate top-down monitoring of the goal stat e vs. a bottom-up goal pursuit process. For instance, contexts involving so cial comparisons (e.g., how much do I study as compared to others?) may be more sensitive to percepti ons of goal progress and goal commitment, while contexts that do not involve comparisons a nd planning are more sensitive to bottom-up influences.

PAGE 56

56 The findings attest a rebound e ffect after a goal is repr esented as achieved. Shah, Friedman, and Kruglanski (2002, p. 1278) argue that continued inhibition of competing goals may render them less salient and less likely to be subsequently pursued once the focal goal is attained or abandoned. However, Shah et al. do state that the rebound effect (Macrae et al. 1994; Wegner and Pennebaker 1993) suggests that the completion or termination of a specific goal pursuit may cause previously inhibited alternatives to flood back into consciousness, increasing the likelihood that th ey will be pursuedthe inhibiti on of alternatives through goal shielding may actually increase the likelihood of their eventual pursuit by increasing their salience after completion. Our results help in the resolution of this conund rum. The goal system does not seem to go back to a steady, resting stat e after the achievement of a goal, as evidenced by the finding that right after pe ople perceive that a goal has been achieved they show desire for behaving inconsistently. This property of the PGG system may be a consequence of evolutionary forces. To the extent competing goals are not contradictory (e.g., ge t food, get water, find shelter), temporarily inhibited goals should be pursued once active goals are achieved. Marketing Implications From a consumer point of view, these findi ngs demonstrate that the performance of a behavior can increase goal activa tion. This activation can result in c onsistent behaviors, an effect that people do not seem to be able to predict. In fact, people seem to believe that the performance of a behavior liberates people for pursuing opposing goals. An implication of this finding is that prediction can be misleading. For example, people ma y predict that the best way to control what they eat is to have a bit of an indulgence whic h will lead to cessation of indulgent behaviors. Such beliefs may lead to the stockpiling of sma ll snacks. Our findings indi cate that this strategy may result in more indulgence unless the eating beha vior is represented as achievement of a goal.

PAGE 57

57 The findings of this research have implications for marketers as well. If an act of regulation is framed as achieving a self-control goal, peopl e may let loose and be more open to hedonic appeals. Also, having consumers perform an act of indulgence may push them toward more indulgence. As indicated in e xperiment 4, stores may motivate consumers to purchase expensive products by having them perform an initial act of indulgence. Small act s of resistance and indulgence may have a quite powerful impact on peoples resolution of a self-control conflict. The current results offer insights on how these acts can be associated with certain goals in order to motivate consumers to perform c onsistent or inconsistent actions. Limitations This research has a number of lim itations. Firs t, an undergrad population was used for data collection, which limits the degree to which the findings can be generalized. Undergrad students do not represent the average American consumer, which might imply deviations from the current findings if the same behaviors were investigated in a br oader population. Second, the same procedure, which involved eating and resisting a chocolate truffl e, was used for every study in this research. Although I believe th at the findings speak to a lot of consumption situations (e.g., someone gets a fine wine to impress others at a dinner and this increases the level of accessibility of an impressing others goal and directs subsequent behavior), using alternative manipulations and dependent measures would certainly provide stronger evidence for the operations of the passive goal guidance system. Finally, although this research studies behavi oral consistency and inconsistency, only the pilot expe riment involved a real behavi or at time 2 (i.e., choosing a snack). The means desirability task I used in mo st experiments has the advantage that it allows me to look at both means valuation and devalu ation patterns, but thes e evaluations may not always match up to actual behavior.

PAGE 58

58 LIST OF REFERENCES Aaker, Jennifer and Patti W illiams (2002), C an Mixed Emotions Peacefully Co-Exist?, Journal of Consumer Research 28 (March), 636-49. Aarts, Henk, Ruud Custers, and Rob W. Holland (2007), The Nonconscious Cessation of Goal Pursuit: When Goals and Negative Affect are Coactivated, Journal of Personality and Social Psychology 92 (February), 165-78. Aarts, Henk, Peter M. Gollwitzer, and Ran R. Ha ssin (2004), Goal Contag ion: Perceiving Is for Pursuing, Journal of Personality and Social Psychology 87 (July), 23-37. Anderson, Michael C., Collin Green, and Kathleen McCulloch (2000), Similarity and Inhibition in Long-Term Memory: Evidence for a Two-Factor Theory, Journal of Experimental Psychology: Learning, Memory and Cognition 26 (September), 1141-59. Bargh, John A. and Tanya L. Chartrand (2000), T he Mind in the Middle: A Practical Guide to Priming and Automaticity Research, in Handbook of Research Methods in Social and Personality Psychology ed. Harry T. Reis and Charles M. Judd, New York: Cambridge University Press, 253-85. Bargh, John A., Mark Chen, and Lara Burrows (1996) Automaticity of Social Behavior: Direct Effects of Trait Construct and Stereotype Activation on Action, Journal of Personality and Social Psychology 71 (August), 230-44. Bargh, John A. and Peter M. Gollwitzer (1994), Environmental Control of Goal-directed Action: Automatic and Strategi c Contingencies between Situations and Behavior, in Integrative Views of Motivation, Cognition, and Emotion ed. William D. Spaulding, Lincoln, NE: University of Nebraska Press, 71-124. Bargh, John A., Peter M. Gollwitzer, Annette Lee-Chai, Kimberly Barndollar, and Roman Trtschel (2001), The Automated Will: N onconscious Activation and Pursuit of Behavioral Goals, Journal of Personality and Social Psychology 81 (December), 101427. Baumeister, Roy F., Elen Bratslavsky, Ma rk Muraven, and Diane M. Tice (1998), Ego Depletion: Is the Active Self a Limited Resource?, Journal of Personality and Social Psychology, 74 (May), 1252-65. Custers, Ruud and Henk Aarts (2007), In Search of the Nonconscious Sources of Goal Pursuit: Accessibility and Positive Affective Valence of the Goal State, Journal of Experimental Social Psychology 43 (March), 312-18. Dhar, Ravi and Itamar Simonson (1999), Mak ing Complementary Choices in Consumption Episodes: Highlighting versus Balancing, Journal of Marketing Research, 36 (February), 29-44.

PAGE 59

59 Dijksterhuis, Ap, Tanya Chartrand, and Henk Aart s (2007), Effects of Priming and Perception on Social Behavior and Goal Pursuit, in Social Psychology and the Unconscious: The Automaticity of Highe r Mental Processes ed. John A. Bargh, New York, NY: Psychology Press, 51-131. Fazio, Russell H. (1990). A Practical Guide to the Use of Response Latencies in Social Psychological Research, in Review of Personality and Social Psychology ed. Clyde A. Hendrick and Margaret S. Clark, Vo l. 11, Newbury Park, CA: Sage, 74. Fishbach, Ayelet and Ravi Dhar (2005), Goals as Excuses or Guides: The Libera ting Effect of Perceived Goal Progress on Choice, Journal of Consumer Research 32 (December), 370-77. Fishbach, Ayelet, Ravi Dhar, and Ying Zhang ( 2006), Subgoals as Substitutes or Complements: The Role of Goal Accessibility, Journal of Personality and Social Psychology 91 (August), 232. Khan, Uzma and Ravi Dhar (2006), Li censing Effect in Consumer Choice, Journal of Marketing Research 43 (May), 259-66. Khan, Uzma and Ravi Dhar (2007), Where Ther e Is a Way, Is There a Will? The Effect of Future Choices on Self-Control, Journal of Experiment al Psychology: General 136 (May), 277-88. Kivetz, Ran, Oleg Urminsky, and Yuhuang Zheng (2006), The Goal-Gradient Hypothesis Resurrected: Purchase A cceleration, Illusionary Goal Progress, and Customer Retention, Journal of Marketing Research 43 (February), 39-58. Kivetz, Ran and Yuhuang Zheng (2006), Determi nants of Justificati on and Self-Control, Journal of Experiment al Psychology: General 135 (November), 572-87. Kruglanski, Arie W. (1996), Goals as Knowledge Structures, in The Psychology of Action: Linking Cognition and Motivation to Behavior ed. Peter M. Gollw itzer. New York, NY: Guilford Press, 599-618. Kruglanski, Arie W., James Y. Shah, Ayelet Fi shbach, Ronald S. Friedman, Woo Y. Chun, and David Sleeth-Keppler (2002), A Th eory of Goal Systems, in Advances in Experimental Social Psychology ed. Mark P. Zanna, San Diego, CA: Academic Press, 331-78. Locke, Edwin and Gary P. Latham (1990), A Theory of Goal Setting and Task Performance, Englewood Cliffs, NJ: Prentice Hall. Macrae, C. Neil, Gralen V. Bodenhausen, Alan B. Milne, and Jolanda Jetten (1994), Out of Mind but Back in Sight: St ereotypes on the Rebound, Journal of Personality and Social Psychology, 67 (November), 808-17.

PAGE 60

60 Mukhopadhyay, Anirban and Gita V. Johar (2005), Where There Is a Will, Is There a Way? Effects of Lay Theories of Self-Contro l on Setting and Keeping Resolutions, Journal of Consumer Research, 31 (March), 779-86. Novemsky, Nathan and Ravi Dhar (2005), Goal Fulfillment and Goal Targets in Sequential Choice, Journal of Consumer Research 32 (December) 396. Okada, Erica (2005), Justifica tion Effects on Consumer Choi ce of Hedonic and Utilitarian Goods, Journal of Marketing Research 42 (February), 43-53. Ramanathan, Suresh and Patti Williams (2007), Immediate and Delayed Emotional Consequences of Indulgence: The Moderating Influence of Personality Type on Mixed Emotions, Journal of Consumer Research 34 (August), 212-23. Shah, James Y., Ron S. Friedman, Arie W. Kr uglanski (2002), Forgetting All Else: On the Antecedents and Consequences of Goal Shielding, Journal of Personality and Social Psychology, 83 (December), 1261-80. Shah, James Y. and Arie W. Kruglanski (2003), When Opportunity Knocks: Bottom-up Priming of Goals by Means and It s Effects on Self -regulation, Journal of Personality and Social Psychology 84 (June), 1109-22. Wegner, Daniel M. and Ja mes W. Pennebaker (1993), Handbook of Mental Control Englewood Cliffs, NJ: Prentice Hall. Zhang, Ying, Ayelet Fishbach, a nd Ravi Dhar (2007), When Thi nking Beats Doing: The Role of Optimistic Expectations in Goal-Based Choices. Journal of Consumer Research 34 (December), 567-78.

PAGE 61

61 BIOGRAPHICAL SKETCH Juliano Laran was born in Porto Alegre, Rio Grande do Sul, Brazil. He attended Unisinos University in Sao Leopoldo, where he earned a Bachelor of Arts degree in business adm inistration. He then attended Universidade Federal do Rio Grande do Sul in Porto Alegre, where he earned a Master of Science degree in business administration/marketing. In 2003 he moved to the United States to ente r the Ph.D. program in marketing at the University of Florida. He has now accepted a position as an assistant pr ofessor of marketing at the University of Miami.