Neural substrates of intrinsic motivation: f MRI studies

Transcription

1 University of Iowa Iowa Research Online Theses and Dissertations Fall 2011 Neural substrates of intrinsic motivation: f MRI studies Woogul Lee University of Iowa Copyright 2011 Woogul Lee This dissertation is available at Iowa Research Online: Recommended Citation Lee, Woogul. "Neural substrates of intrinsic motivation: fmri studies." PhD (Doctor of Philosophy) thesis, University of Iowa, Follow this and additional works at: Part of the Educational Psychology Commons

2 NEURAL SUBSTRATES OF INTRINSIC MOTIVATION: FMRI STUDIES by Woogul Lee An Abstract Of a thesis submitted in partial fulfillment of the requirements for the Doctor of Philosophy degree in Psychological and Quantitative Foundations (Educational Psychology) in the Graduate College of The University of Iowa December 2011 Thesis Supervisor: Associate Professor Joyce Moore

3 1 ABSTRACT Numerous social and educational psychologists propose that intrinsic motivation generated by personal interests and spontaneous satisfactions is qualitatively different from extrinsic types of motivation generated by external compensations and also that intrinsic motivation is more beneficial to learning than extrinsic types of motivation. However, in the field of neuroscience, intrinsic motivation has been little studied while extrinsic types of motivation (e.g., incentive motivation) have been thoroughly studied. The purpose of the present studies was to expand the neural understanding of motivation to include intrinsic motivational processes. To do so, a series of three event-related functional magnetic resonance imaging (fmri) studies were conducted. Study 1 and Study 2 compared the neural activities when participants decided to act for intrinsic reasons (i.e., self-determined volitional and agentic behavior) versus when they decided to act for extrinsic reasons (i.e., non-self-determined volitional and agentic behavior). Both studies showed that the anterior insular cortex, known to be related to a sense of agency, was more activated during self-determined behavior associated with intrinsic reasons for acting while the posterior parietal regions (e.g., posterior cingulate cortex, angular gyrus), known to be related to a sense of a loss of agency, were more activated during non-self-determined behavior associated with extrinsic reasons for acting. These findings confirm the existence of neural-based intrinsic motivational processes, differentiate intrinsic motivation from incentive motivation, and document the important neural activities which function for generating self-determined agentic action. Study 3 examined these same neural activities as participants engaged in interesting and

4 2 uninteresting versions of two experimental tasks. Results confirmed the results of the earlier two studies, as the anterior insular cortex was more recruited when participants performed the interesting, but not the uninteresting, version of the tasks. Results also extended the findings from Studies 1 and 2 in an important way in that the ventral striatum, a well-known brain region for reward processing, was more activated when participants performed the interesting, but not the uninteresting, version of the experimental tasks. These findings suggest that intrinsic motivation is generated based on the feeling of intrinsic need satisfaction (from anterior insular cortex activations) and the feeling of reward (from ventral striatum activations). Overall, the present research established three new findings: (1) the neural bases of intrinsic motivation lies largely in increased anterior insular cortical activities; (2) when people made decisions about selfdetermined intrinsically-motivated behavior, they show enhanced insular cortical activities and suppressed posterior parietal cortical activities; and (3) when people engaged in actual self-determined intrinsically-motivated behavior, they show enhanced insular cortical and ventral striatal activities. In establishing these new findings, the paper introduces a new area of study for motivational neuroscience namely, intrinsic motivation. Abstract Approved: Thesis Supervisor Title and Department Date

5 NEURAL SUBSTRATES OF INTRINSIC MOTIVATION: FMRI STUDIES by Woogul Lee A thesis submitted in partial fulfillment of the requirements for the Doctor of Philosophy degree in Psychological and Quantitative Foundations (Educational Psychology) in the Graduate College of The University of Iowa December 2011 Thesis Supervisor: Associate Professor Joyce Moore

6 Copyright by WOOGUL LEE 2011 All Rights Reserved

7 Graduate College The University of Iowa Iowa City, Iowa CERTIFICATE OF APPROVAL PH. D. THESIS This is to certify that the Ph. D. thesis of Woogul Lee has been approved by the Examining Committee for the thesis requirement for the Doctor of Philosophy degree in Psychological and Quantitative Foundations (Educational Psychology) at the December 2011 graduation. Thesis Committee: Joyce Moore, Thesis Supervisor Johnmarshall Reeve Jinhu Xiong Kathy Schuh Walter Vispoel

8 TABLE OF CONTENTS LIST OF TABLES... iv LIST OF FIGURES...v CHAPTER I INTRODUCTION...1 II REVIEW OF LITERATURE...5 Self-Determination Theory...6 Basic Needs Theory...7 Organismic Integration Theory...9 Positive Functioning of Intrinsic Types of Motivation...11 Critics of Intrinsic Motivation...12 Neural Bases of Incentive Motivation...13 Findings from Animal Studies...13 Findings from Patient Studies...16 Neural Similarity between Human and Animal Reward Processing...18 Findings from Neuroimaging Studies...19 Neural Evidence of Intrinsic Types of Motivation...21 Findings from Recent Intrinsic Motivation Studies...22 Findings from Studies on Volition...23 Findings from Studies on Craving and Addiction...24 Research Questions and Hypotheses...25 III STUDY1: NEURAL DIFFERENCES BETWEEN INTRINSIC REASONS FOR DOING VERSUS EXTRINSIC REASONS FOR DOING: AN FMRI STUDY...28 Introduction...28 Method...31 Participants...31 Stimuli...31 Task and Procedure...33 fmri Data Acquisition...34 fmri Data Analysis...35 Results...37 Behavioral Results...37 fmri Results...39 Discussion...43 ii

9 IV STUDY 2: EXPERIENCING SELF-DETERMINED VERSUS NON- SELF-DETERMINED REASONS FOR ACTING: THE NEURAL CORRELATES OF DIFFERENT MANIFESTATIONS OF PERSONAL AGENCY...46 Introduction...46 Method...51 Participants...51 Stimuli...52 Task and Procedure...54 Measure...56 fmri Data Acquisition...56 fmri Data Analysis...57 Results...59 Pilot Test Results...59 Behavioral Results...61 fmri Results...62 Discussion...67 V STUDY 3: NEURAL SUBSTRATES OF INTRINSIC MOTIVATION DURING THE TASK PERFORMANCE: AN FMRI STUDY...74 Introduction...74 Method...76 Participants...76 Tasks...76 Procedure...80 fmri Data Acquisition...82 fmri Data Analysis...82 Results...84 Behavioral Results...84 fmri Results...85 Discussion...93 VI GENERAL DISCUSSION...96 APPENDIX BASIC PSYCHOLOGICAL NEEDS SCALE REFERENCES iii

10 LIST OF TABLES Table 1. Examples of phrases from each of the three experimental conditions used in the experimental task Results of the subtraction analysis between the IM and EM conditions Examples of phrases from each of the three experimental conditions used in the experimental task Results of the subtraction analysis between the IM and EM conditions Results of the regression analysis between participants self-reported intrinsic satisfactions and the neural activities in the IM condition Examples of curious and non-curious questions Examples of challenging and non-challenging anagrams Results of the conjunction analysis of the contrast between curious versus non-curious questions and the contrast between challenging versus nonchallenging anagrams Results of the contrast between curious versus non-curious questions Results of the contrast between challenging versus non-challenging anagrams...90 iv

11 LIST OF FIGURES Figure 1. Task and experimental design Participants mean yes percentage and mean reaction time Insular cortex activities in the subtraction analysis Posterior cingulate cortex in the subtraction analysis Procedure of the pilot test Task and experimental design Results of the pilot test Participants mean behavioral energization rating and mean reaction time Anterior insular cortex activities in the subtraction analysis Angular gyrus activities in the subtraction analysis Anterior insular cortex activities in the intrinsic motivation condition and its correlation with participants perceived intrinsic satisfactions Task and experimental design Participants mean interest rating Neural activations in the conjunction analysis Ventral striatum activities in the subtraction analysis...92 v

12 1 CHAPTER I INTRODUCTION Intrinsic motivation is inherently-created energy and sense of purpose, and it can be distinguished from extrinsic motivation, which is environmentally-created energy and sense of purpose. This distinction between the two types of motivation is important because several different programs of research have shown that intrinsic motivation is more beneficial for learning than is extrinsic motivation, as intrinsic motivation has been shown to enhance many crucial aspects of learning, such as creativity (Amabile, 1985), depth of engagement (Cordova & Lepper, 1996; Reeve, Jang, Carrell, Jeon, & Barch, 2004), conceptual understanding of what one is trying to learn (Vansteenkiste, Simons, Lens, Soenens, & Matos, 2005), psychological well-being (Deci et al., 2001; Kasser & Ryan, 1996; Sheldon, Ryan, & Reis, 1996), and so on. In addition, intrinsic motivation is qualitatively different from extrinsic motivation, and these two different types of motivation are not additive but selective. If learners receive both intrinsic rewards and extrinsic rewards, there are no synergic effects between them; instead, there are detrimental effects of extrinsic rewards on intrinsic motivation (Deci, 1971; Deci, Koestner, & Ryan, 1999a, 1999b). For these reasons, educational researchers have studied how to enhance learners intrinsic motivation, as through the provision of contextualization, personalization, choice (Cordova & Lepper, 1996), and autonomysupportive environments (Reeve, 2009; Reeve & Jang, 2006; Ryan & Deci, 2000). The concept of intrinsic motivation, however, has been questioned by researchers whose academic backgrounds are more rooted in behavioral traditions (Bandura, 1986;

13 2 Locke & Henne, 1986). They believe that intrinsic motivation is a utopian concept and, in the real world, is not significantly different from extrinsic motivation. They have insisted that there is limited evidence to justify or validate a distinction between intrinsic motivation and extrinsic motivation. Among those who study intrinsic motivational processes, theoretical assumptions are made as to the underlying antecedents of intrinsic motivation, and these have been said to be basic psychological needs such as autonomy and competence. But, the critics have argued that the results of intrinsic motivation have limitations because the self-report questionnaires rely on participants subjective perceptions and the free choice behavioral measures that are often used are not direct measures of intrinsic motivation itself but, rather, only indirect measures based on the behavioral consequences of intrinsic motivation. Neuroscientific research methods can be used to inform and resolve this theoretical debate (Berninger & Corina, 1998; Byrnes & Fox, 1998). The current state of affairs in neuroscience, however, is to ignore the empirical study of intrinsic motivation. As there are few studies on intrinsic motivation, there is of course little existing neuroscientific knowledge about intrinsic motivation, but, at the same time, neuroscientific research methods do offer the promise of contributing to a greater understanding of the concept of intrinsic motivation. Even though motivation studies in the neuroscience field have been exclusively focused on extrinsic types of motivation (e.g., incentive motivation), it is important to review neuroscientific knowledge of motivation because some knowledge of these studies is applicable to intrinsic motivation.

14 3 Motivation theories within the field of neuroscience are deeply rooted in animal studies on physiological need satisfaction (e.g., hunger, thirst). In these animal studies, researchers made animals hungry or thirsty and then examined neural activities when the animals later consumed foods or water. Results have consistently demonstrated that extrinsic types of motivation are associated with neural brain activities, such as the striatum regions (e.g., nucleus accumbens, caudate, putamen), the frontal regions (e.g., ventromedial prefrontal cortex, dorsolateral prefrontal cortex, orbitofrontal cortex, anterior and posterior cingulate cortex), the limbic system (e.g., amygdala, hypothalamus), and the basal ganglia system (e.g., subthalamic nucleus) (Cardinal, Parkinson, Hall, & Everitt, 2002; McClure, York, Montague, 2004; Schultz, 2000). In clinical patient studies, numerous studies on extrinsic types of motivation and decision making based on these external types of motivation have been conducted. According to the somatic marker hypothesis, emotions guide people s reward-based decision making processes (Bechara & Damasio, 2005; Bechara, Damasio, Tranel, & Damasio, 2005; Damasio, Everitt, & Bishop, 1996). Damage to the ventromedial prefrontal cortex leads people to make disadvantageous decisions that are mainly guided by immediate benefits without considering future consequences. The ventromedial prefrontal cortex activates the brain regions related to primary emotional triggers (e.g., amygdala) and secondary emotional triggers (e.g., insular cortex), which are important for reward-based decision making. Therefore, the researchers assumed that damage to the ventromedial prefrontal cortex is related to emotional impairments and, as a result, related to inappropriate reward-based decisions even though these patients seem to be cognitively intact.

15 4 In human neuroimaging studies on extrinsic types of motivation, the same brain regions observed during reward processing in animal studies have been similarly activated in human studies (Cardinal et al., 2002; Hampton & O Doherty, 2007; McClure et al., 2004; O Doherty, 2004). In particular, the ventromedial prefrontal cortex including the orbitofrontal cortex, the ventral striatum (e.g., nucleus accumbens), the anterior and posterior cingulate cortex, the dorsolateral prefrontal cortex, and the amygdala have been frequently observed during human reward processing. As the technologies of the neuroimaging methods have been developed, it has become possible to examine the functions of each brain region related to reward processing and the neural similarities and differences among different types of extrinsic types of motivation. In studies on extrinsic rewards, the activities of the ventromedial prefrontal cortex, including the orbitofrontal cortex, have been frequently observed when participants receive or expect to receive extrinsic types of rewards. The ventromedial prefrontal cortex is activated when computing different rewarding values (or expected rewarding values) of various environmental stimuli (Murray, O Doherty, & Schoenbaum, 2007; Rushworth, Behrens, Rudebeck, & Walton, 2007). The main purpose of the current program of research is to identify the neural substrates of intrinsic motivation. To accomplish this research objective, theories of intrinsic motivation from social psychology and from educational psychology were used to formulate hypotheses that were then tested empirically in a series of studies that utilize neuroscientific research methods.

16 5 CHAPTER II REVIEW OF LITERATURE Intrinsic motivation is the inherent propensity to engage one s interests and to exercise one s capacities and, in doing so, to explore, to learn (for its own sake), and to seek out, master, and persist in optimal challenges that can extend one s skills and capacities (Amabile, 1985; Deci & Ryan, 1985; Lepper, Greene, & Nisbett, 1973; Pittman & Heller, 1987). It emerges spontaneously from one s intrinsic psychological needs. When intrinsically motivated, people willingly participate in activities without tangible rewards or external control, which is because the task or activity has its own inherent rewards (interest, enjoyment) and spontaneous satisfactions (feeling autonomous, feeling competent, feeling related; Deci & Ryan, 1985; Ryan, 1995). In contrast, extrinsic motivation is defined as an environmentally-created motivation for beneficial consequences or for positive utility values of a task or activity (Deci & Ryan, 1985; Ryan & Deci, 2000). It is an in order to type of motivation in which the person engages in motivated action to receive some task exogenous benefit or consequences, such as a reward or privilege. Because extrinsic motivation revolves around the strategic pursuit of external contingencies, people focus less on the task or activity itself. When extrinsically motivated, people tend to adjust why they engage in an activity namely, not to experience intrinsic rewards and inherent satisfactions but, rather, to gain an attractive environmental consequence. In general, this adjustment tends to leave people to orient toward their environment in a less proactive (more reactive) and less challenge-seeking (more reward-seeking) way (Deci & Ryan, 1985; Ryan, 1995).

17 6 Self-Determination Theory Self-determination theory has been one of the most influential theories to explain the nature of human motivation (Deci & Ryan, 1985; Ryan & Deci, 2000). Selfdetermination theory was first conceptualized by Edward L. Deci based on studies of the overjustification hypothesis (Deci, 1971). The overjustification hypothesis states that extrinsic rewards exert detrimental effects on intrinsic motivation. This hypothesis assumes that, if extrinsic rewards are given when people perform a task that they would be intrinsically motivated to perform without any extrinsic rewards, their intrinsic motivation to do the task decreases once extrinsic rewards are removed. Even though researchers had previously mentioned the possible negative relation between extrinsic rewards and intrinsic motivation (decharms, 1968), Deci s study was the first empirical study that directly tested and confirmed the overjustification hypothesis. Since then, the overjustification hypothesis has been verified and generalized by numerous other studies (Deci et al., 1999a, 1999b; Lepper et al., 1973). These studies are important to the present program of research for two reasons. First, they validated the idea that types of motivation exist that intrinsic motivation and extrinsic motivation represent qualitatively different types of motivation. Second, they validated the idea that the two types of motivation are not additive or complementary and, in fact, that they are sometimes antagonists as extrinsic motivation can undermine (reduce) intrinsic motivation.

18 7 Basic Needs Theory Based on the results showing that there were detrimental effects of extrinsic rewards on intrinsic motivation, researchers started to suggest that intrinsic motivation and extrinsic motivation are not a single construct but qualitatively different constructs. In addition, researchers addressed the question of what factors make intrinsic motivation different from extrinsic motivation. They suggested that people have three basic psychological needs (i.e., competence, autonomy, relatedness), and also that intrinsic motivation is the motivation that arises from the process of satisfying these needs. This subtheory within the larger self-determination theory framework is called basic needs theory (Deci & Ryan, 1985, 2002; Ryan & Deci, 2000). Competence is defined as a need to be effective in one s interactions with the environment. It is the psychological need for competence that energizes and directs people s willingness to seek out, master, and persist in optimal challenges and, in doing so, seek to exercise and extend their capacities to interact more effectively with their environmental surroundings. Hence, increases in perceived competence underlie increases in intrinsic motivation. People experience perceptions of competence when they have optimal challenge and effectance-promoting feedback. This means that people cannot have perceptions of competence if they perform a task of which the level is too easy or if they receive feedback that communicates a sense of incompetence even during an optimally challenging task (Deci & Ryan, 1985, 2002; Ryan & Deci, 2000). Autonomy is defined as a need to engage in a task or participate in an activity out of internally-locused causalities rather than out of externally-locused causalities. It is the psychological need for autonomy that underlies people s desire to experience self-

19 8 direction and personal endorsement in the initiation and regulation of their behaviors. Perceptions of autonomy are generated when people believe that they have choices, volition, and psychological freedom to initiate and regulate their behavior in an authentic and self-determined way. Therefore, people can show enhanced intrinsic motivation in environments that support their autonomy, which are called autonomy-supportive environments (Deci & Ryan, 1985, 2002; Ryan & Deci, 2000). Relatedness is defined as a need that people want to have close emotional bonds with other people. It is the psychological need for relatedness that leads people to desire close, affectionate bonds or attachments with the important people in their life. This means that people need to be interpersonally involved with others in warm, caring, and reciprocal relationships. Perceptions of relatedness are generated when people involve themselves with relationship partners who express concern for their welfare and take steps to deepen the emotional bond between self and other. Collectively, these three basic needs provide individuals with a proactive energy source, which is intrinsic motivation, for them to initiate and regulate their volitional behavior and, as a result, to facilitate optimal functioning, at least to the extent that environmental conditions support and nurture (rather than neglect and frustrate) these psychological needs (Deci & Ryan, 1985, 2002; Ryan & Deci, 2000). In this regard, the self-determination theory researchers have suggested that intrinsic motivation is qualitatively different from extrinsic motivation because intrinsic motivation is generated by the spontaneous satisfactions they experience during a task or activity (e.g., having fun while completing a project) while extrinsic motivation is generated by the anticipation of beneficial consequences made contingent on task

20 9 engagement (e.g., anticipating money for completing a project). This distinction conceptualizes intrinsic motivation as an inherent and proactive inside-out type of motivation in which motivational resources inside the person energize and direct outwardly expressed behaviors, while it conceptualizes extrinsic motivation as an acquired and reactive outside-in type of motivation in which outside environmental incentives and rewards create acquired motivational states inside the person (Deci & Ryan, 1985, 2002; Ryan & Deci, 2000). Organismic Integration Theory Even though self-determination theorists emphasize the importance of inner motivational resources for generating self-determined motivation, they also insist that extrinsic motivation can be a type of self-determined motivation in the adaptive process of internalization. Self-determination theorists insist that even extrinsic types of motivation can become more self-determined as people internalize the externally-locused causalities of extrinsic motivation into the self-system (Deci & Ryan, 1985; Ryan & Deci, 2000). Under this assumption, self-determination theorists categorize five types of motivational regulatory processes that differ in the extent to which they represent a relatively autonomous type of motivation. External regulation represents the least autonomous but the most controlled type, and extrinsically motivated behaviors are those that are regulated by environmental forces rather than by the self (e.g., reward contingencies). Introjected regulation involves taking in an external regulation but not accepting it as one s own; the reason to act is introjected rather than internalized into the self. Although internally driven, introjected behaviors nevertheless arise mostly out of an

21 10 external perceived locus of causality that is experienced as self-induced pressure rather than as self-endorsed choice (e.g., guilt, shame, obligation). Identified regulation represents a relatively autonomous type of extrinsic motivation. With identified regulation, behaviors are engaged in with high volition and self-endorsement because the person perceives them to be personally important and useful (but not inherently interesting) things to do (e.g., recycling, flossing). Integrated regulation occurs when otherwise separate and psychologically-isolated identified motivations are assimilated (i.e., integrated) into the self-system. Integrated regulation is a highly autonomous type of extrinsic motivation because the person reflects on an action, evaluates it as part of the self, and brings that way of acting into congruence with existing values and one s sense of self and personal identity. The most autonomous type of motivation is intrinsic regulation. Overall, this subtheory of self-determination theory that distinguishes between qualitatively different types of extrinsic motivation is called organismic integration theory (Deci & Ryan, 1985; Ryan & Deci, 2000). Autonomous motivation reflects all those types of motivation that are selfendorsed, volitional, and highly concordant with one s inner motivational resources (decharms, 1968; Deci, 1980; Deci & Ryan, 2000; Ryan & Deci, 2000; Sheldon & Elliot, 1999) namely, intrinsic motivation, identified motivation, and integrated motivation. Controlled motivation reflects all those types of motivation that are environmentallyendorsed, pressure oriented, discordant with one s inner motivational resources, and arise from learning response-outcome environmental and intrapsychic contingencies namely, extrinsic motivation (also called external regulation) and introjected motivation. The

22 11 term controlled is used because the functional purpose of incentives, rewards, and guilt-messages is often to control (i.e., increase, shape) another person s behavior. Positive Functioning of Intrinsic Types of Motivation Dozens of experimental and field studies have examined the correlates and consequences of the autonomous and controlled types of motivation. Consistently, autonomous types of motivation have been associated with more positive functioning than have controlled types of motivation (Reeve, Deci, & Ryan, 2004; Ryan & Connell, 1989), and this has been found to be true across a wide range of outcomes, including greater engagement in activities (Reeve et al., 2004), greater conceptual understanding of the material to be learned (Vansteenkiste, Simons, Lens, Soenens, & Matos, 2005), and optimal functioning and psychological well-being (Deci et al., 2001), among many other positive outcomes (Ryan & Deci, 2000). In addition, over 100 empirical studies have examined the effect that the receipt of extrinsic incentives, such as You will receive a $3 reward at the end of today s session for doing the puzzles (Ryan, Mims, & Koestner, 1983), have on subsequent intrinsic motivation toward an activity, and the results of a meta-analysis showed that, overall, extrinsic incentives decrease intrinsic motivation (Deci et al., 1999a, 1999b). Taken as a whole, these research studies show that (1) different types of human motivation exist, (2) these different types are associated with different qualities of functioning and outcomes, and (3) the experience of one type of motivation can interfere with the experience of the other.

23 12 Critics of Intrinsic Motivation Self-determination theory in general and the concept of intrinsic motivation in particular have been criticized by numerous researchers from organizational psychology and behavioral science (Bandura, 1986; Cameron & Pierce, 1994; Eisenberger & Cameron, 1996; Locke & Henne, 1986, Schwartz, 2000). These critics forward three core criticisms of self-determination theory and intrinsic motivation. First, critics assert that there is little evidence to distinguish intrinsic motivation from other motivation (e.g., achievement motivation, incentive motivation). This means that, even though selfdetermination theory differentiates intrinsic motivation from other types of motivation, there is still insufficient empirical evidence to warrant self-determination theory s theoretical assumptions (e.g., the existence of psychological needs). Second, critics have questioned the validity of the measures of intrinsic motivation. In most studies, intrinsic motivation has been measured by either self-reported questionnaires or free choice behavioral measures after performing tasks, and many studies have utilized both measures. Critics raise a possibility that responses to the self-report questionnaires can be biased by other confounding factors (e.g., social desirability, self-esteem) as these responses are based on participants perceptions. The critics also insist that free choice behavioral measures are not a direct measure of intrinsic motivation and, as a result, able to be affected not only by intrinsic motivation but also by other factors (e.g., protecting self-pride, preparing for the future trials). Third, the critics postulate that extrinsic motivation is not a controlled type of motivation. They insist that there can be many types of motivation out of diverse environmental factors (e.g., competitions, incentives, rewards), which are different from the source of intrinsic motivation, but self-

24 13 determination theory negates positive energy of these types of motivation (Cameron & Pierce, 1994; Eisenberger & Cameron, 1996). That is, under some conditions, the presence of external contingencies can enhance, rather than undermine, constructive outcomes such as creativity and persistence. In this regard, the critics conclude that it is hard to generalize the detrimental effects of incentives or rewards on intrinsic motivation. Neural Bases of Incentive Motivation Much of the early neural understanding of motivation was developed based on findings from animal studies. Similarly, the contemporary neural understanding of motivation continues to make remarkable progress based on the concept of incentive motivation rooted in behavioral tradition (Berridge, 2004). Findings from Animal Studies Mogenson and his colleagues (1980) reviewed the results of early animal studies on the neural circuits of goal-directed behaviors. These authors presented many motivation models at an early stage which explain how emotion and cognitive functions distinctively or interactively guide behaviors. Based on the results of numerous animal studies, the authors proposed interconnected relationships among the limbic system, the striatum (e.g., nucleus accumbens, caudate), and the motor system to initiate biologicallysignificant behaviors. Some studies specified the different mechanisms of animal motivation and hence the functional roles of various brain regions related to each mechanism of animal motivation. Robbins and Everitt (1996) specified associative structures underlying

25 14 motivation into classic (Pavlovian) associations about the knowledge of stimuli reinforce relationships, associations based on the knowledge of the contingency between voluntary actions and the reinforcing outcome, and habit-like automatic stimuli response associations. These authors also classified motivational behaviors as appetitive behaviors and consummatory behaviors. They explained that appetitive behaviors, which are also called preparatory behaviors, are behaviors in anticipation of reinforcers (e.g., locomotor approach responses) while consummatory behaviors are behaviors based on subjective affective states (e.g., ingestion, sexual mounting). Based on the knowledge about the different mechanisms of animal motivation, they specified the associated brain functions. They have shown that the amygdala, the prefrontal cortex, and the brain stem work for positive and negative reinforcement. They also showed that motivated behaviors are generated by interactive functions between the amygdala and the dopamine-dependent functions of the ventral striatum (e.g., nucleus accumbens). The activities of the amygdala are known to be more related to associative information about conditioned stimuli and reinforcement. Therefore, this region involves basic functions for both appetitive and consummatory behaviors. It is also known that appetitive behaviors and consummatory behaviors are separately controlled by distinct neural pathways from the ventral striatum. In addition, there are studies trying to specify the phases of animal reward processing, which is an important notion for generating animal motivation, and related brain functions for each phase (e.g., reward receipt, reward anticipation). Shultz (2000) specified three phases of reward processing (i.e., reward detection and perception, expectation of future rewards, reward-based goal direction) and compared the neural

26 15 activities of addiction with those of reward processing. Reward detection and perception have been known to be related to the activations in the striatum regions (e.g., nucleus accumbens, caudate, putamen), the subthalamic nucleus, the substantia nigra, the dorsolateral prefrontal cortex, the orbitofrontal cortex, the anterior cingulate cortex, the amygdala, and the lateral hypothalamus in numerous animal studies. In particular, striatum activities are known to be related not only to activities of dopamine neurons but also to activities of other neurons. The striatum, the orbitofrontal cortex, and the amygdala are known to be related to expectation of future rewards which is established by prior experiences. Shultz explained that expected rewards could serve as goals for voluntary behaviors. In that case, the information of reward expectation needs to be integrated with behavior for obtaining rewards. The ventral striatum (e.g., nucleus accumbens), the dorsolateral premotor cortex (e.g., cingulate motor area), the dorsolateral prefrontal cortex, the caudate, and the parietal cortex (e.g., posterior cingulate cortex) are known to be related to these processes of goal direction. In particular, the dorsolateral prefrontal cortex works for holding information which is needed during decision making processes; the cingulate regions work for monitoring performances or outcomes and adjusting behaviors for beneficial consequences; and the parietal cortex is activated when animals choose larger or more frequent rewards over smaller or less frequent rewards. In particular, the author reviewed that the nucleus accumbens, which is an important region for reward processing, has been also observed in the studies on addiction. This result proposed a possibility that the neural activities of addiction have similarities to those of reward processing.

27 16 Berridge and his colleagues specified the processes of motivation (Berridge, 2004; Berridge & Robinson, 2003). They suggested that the motivational concepts can be parsed into three components, learning, liking, and wanting. They assumed that people have an innate tendency to learn about the knowledge of associations. They also assumed that, based on this knowledge, people can like something, which is a hedonic (i.e., objective affective) reaction, and want something, which is a conscious or subjective desire. Based on this theory, they examined the neural activities of each component and recognized the neural differences of the subcortical (i.e., liking) and cortical (i.e., wanting) regions among the components in the neuronal level. Findings from Patient Studies Based on the observations of clinical patients neural activities and their behavioral impairments (e.g., the case of Phineas Gage), researchers have tried to extend the neural understanding of human motivation and reward processing. One of the most influential theories about human motivation and reward-related decision making processes is the somatic marker hypothesis (Bechara & Damasio, 2005; Bechara et al., 2005; Damasio et al., 1996). The somatic marker hypothesis proposes that emotional processing greatly influences human decision making processes, including reward-related decision-making. Based on the somatic marker hypothesis, the amygdala, the insular cortex, and the ventromedial prefrontal cortex are important brain regions for human motivation and reward-related decision making. The amygdala is thought to work as a primary inducer for somatic states (i.e., emotional states). Once somatic states from primary inducers are created, people can remember and represent the somatic states from

28 17 primary inducers. Researchers assumed that this information is stored in and retrieved from the somatosensory map. The insular cortex is thought to work for the somatosensory map. Even though somatic states are not triggered by primary inducers, people can have those somatic states in memory that are generated by the somatosensory map. For these reasons, researchers called the function of the insular cortex as-if-bodystates. It is suggested that the ventromedial prefrontal cortex catalyzes the functions of the amygdala and the insular cortex. Therefore, the ventromedial prefrontal cortex has a critical role in the somatic marker hypothesis. Due to the functional roles of the ventromedial prefrontal cortex, patients who have lesions in this region tend to make inadequate decisions even though they are cognitively intact because they have deficits in emotional processing. The somatic marker hypothesis is important to the larger study of reward processing because it shows how emotional processing feeds into and affects reward processing. Some researchers have tried to integrate the motivationally-relevant findings from human clinical patient studies with those from animal studies (Cardinal et al., 2002). These authors tried to specify the functions of each brain region related to reward processing. They suggested that the amygdala plays an important role in Pavlovian conditioning of emotional responses. This means that the amygdala works for associative learning among stimuli and outcomes. They also suggested that the ventral striatum (e.g., nucleus accumbens) mediates the associative knowledge to guide actions or behaviors. The prelimbic cortex, a part of the medial prefrontal cortex in the human brain (e.g., ventromedial prefrontal cortex), was suggested to detect instrumental contingencies. Therefore, these neural activities are essential to inhibit inappropriate actions or

29 18 behaviors. The insular cortex was suggested to be related to storing or retrieving incentive values of the rewards when the rewards are absent. This means that insular cortex activities are not related to incentive values themselves but related to stored memories or feelings of incentive values. The activities of the orbitofrontal cortex were suggested to be related to representing the anticipated incentive values of stimuli. They also suggested the functions of the anterior cingulate cortex which are related to selecting actions based on information of emotionally significant stimuli. In the case of the frontal regions (e.g., ventromedial prefrontal cortex, orbitofrontal cortex, anterior cingulate cortex), they were suggested to be related to more complex associative learning than the amygdala was. Neural Similarity between Human and Animal Reward Processing As studies on reward processing have accumulated, researchers have tried to compare the neural circuits of human reward processing to those of animal reward processing (McClure et al., 2004; O Doherty, 2004). These studies covered not only primary reward (e.g., fruit juice, water) but also conditioned rewards, which are initially neutral but conditioned through experience to predict consumable rewards and social rewards (e.g., beautiful faces, social interactions, affect-laden words, pleasant touch) to propose the important brain regions for human reward processing, including the orbitofrontal cortex, the amygdala, and the ventral striatum (e.g., nucleus accumbens). The activities of the orbitofrontal cortex involve reward valuation. This means that orbitofrontal cortex activities are associated with computing and evaluating the rewarding values of the stimuli. The activities of the amygdala involve positive and negative

30 19 valence of rewarding stimuli. This means that amygdala activities are more related to affective aspects of stimuli while orbitofrontal cortex activities are more related to rewarding magnitudes of stimuli. The activities of the nucleus accumbens involve monitoring the gap between the expected values and the real values, and directing behavior appropriately. In addition, functional reduction or malfunctioning of the nucleus accumbens is associated with motor and cognitive deficits (e.g., Parkinson s disease). Findings from Neuroimaging Studies With the results of recent neuroimaging studies, functional roles of brain regions for human reward processing have become increasingly specified. In particular, the functions of the ventral striatum (e.g., nucleus accumbens) have been differentiated from the functions of the dorsal striatum (e.g., dorsal parts of caudate and putamen). Essentially, the ventral striatum works more as a critic while the dorsal striatum works more as an actor in reward processing (Haruno et al., 2004; O Doherty et al., 2004). This means that the dorsal striatum is recruited when people need to select actions based on reward-related information while the ventral striatum is recruited when people learn the reward-related information regardless of the existence of action selections. However, there are still many disagreements on the specific role of each striatum region. For example, many researchers have agreed that the nucleus accumbens plays an important role in reward processing generally but they have disagreed on the specific roles of this region. Some researchers argue that the nucleus accumbens is associated with hedonic feelings during reward expectation or reward receipt (Knutson, Adams, Fong, & Hommer,

31 ; Knutson & Bossaerts, 2007; Leknes & Tracey, 2008). Other researchers limit the role of the nucleus accumbens during reward expectation and receipt specifically to reinforcement learning by readjusting errors. This position, called the prediction error theory, means that people learn from previous decision errors to make subsequent decisions (Hare, O Doherty, Camerer, Schultz, & Rangel, 2008; McClure et al., 2004; O Doherty et al., 2004). In this regard, to confirm the functional roles of each striatum regions, there should be more accumulated supporting evidence. Some studies have tried to dissociate the functional roles of some brain regions that had been believed to have similar functions for human motivation. For instance, some studies have tried to distinguish the functional roles between the orbitofrontal cortex and the anterior cingulate cortex (Rushworth et al., 2007; Walton, Devlin, & Rushworth, 2004). These studies have suggested that the anterior cingulate cortex is more related to rewarding values of reward-based actions or behaviors while the ventromedial prefrontal cortex, including the orbitofrontal cortex, is more related to rewarding values of stimuli themselves. This means that the anterior cingulate cortex works for recognizing action reward associations while the ventromedial prefrontal cortex works for recognizing stimuli reward associations. Therefore, the anterior cingulate cortex has been frequently observed in studies where participants need to perform a task with effort in order to obtain rewards (Hampton & O Doherty, 2007; Kirsch et al., 2003; Knutson et al., 2001; Knutson, Fong, Bennett, Adams, & Hommer, 2003). In contrast, the ventromedial prefrontal cortex has been generally observed in the studies in which participants do not have choices or behavioral efforts but instead evaluate how much the given stimuli are beneficial (i.e., evaluating the magnitude of

32 21 rewarding stimuli) (Elliott, Friston, & Dolan, 2000; Murray et al., 2007; O Doherty, 2004). Therefore, researchers have assumed that the reward value monitoring of the anterior cingulate cortex is more action-based and effort-based than that of the ventromedial prefrontal cortex and the orbitofrontal cortex. But, the functional dissociations between the orbitofrontal cortex and the anterior cingulate cortex are still unclear. Neural Evidence of Intrinsic Types of Motivation Neuroscientific work on intrinsic motivation has been sparce. One reason is that neuroscience is basically rooted in behavioral traditions (Berridge, 2004; Schultz, 2000). In behavioral traditions, the main focus is whether experimental manipulations cause externally-observable behavioral changes. In the same sense, the experimental manipulations of neuroscientific studies have not been designed to investigate changes in people s inner psychological states. This means that, in the neuroscience field, intrinsic types of motivation have not been accepted as viable scientific concepts. Another reason is that intrinsic types of motivation are more difficult to generate in experimental settings compared to extrinsic types of motivation. In neuroscientific studies, extrinsic types of motivation are generated simply by providing rewarding stimuli (e.g., monetary rewards, food, water) or by making participants expect those rewarding stimuli. In contrast, intrinsic types of motivation are not that simple to generate (e.g., what are the environmental conditions that lead to high perceived autonomy?). For these reasons, the concepts related to motivation in the neuroscience field are exclusively based on the understanding of incentive motivation (i.e., reward processing). Recently, however,

33 22 neuroscientific studies have examined intrinsic types of motivational concepts, such as curiosity and achievement motivation (Kang et al., 2009; Mizuno et al., 2008; Murayama, Matsumoto, Izuma, & Matsumoto, 2010) and also concepts more closely related to intrinsic motivation such as volition (Brass & Haggard, 2007; Haggard, 2008; Nachev, Rees, Praton, Kennard, & Husain, 2005). Findings from Recent Intrinsic Motivation Studies One pioneering study tried to identify the neural substrates of academic achievement motivation, which is a type of intrinsic motivation in learning situations (Mizuno et al., 2008). This study found that participants self-reported academic achievement motivation scores were highly correlated with the magnitude of putamen activities while performing a working memory task. Another pioneering study examined curiosity, one of the concepts related to intrinsic types of motivation (Kang et al., 2009). This study demonstrated that higher curiosity showed greater caudate activities and enhanced memory. More recently, Murayama and his colleagues (2010) sought neural evidence to support the undermining effects of extrinsic rewards on intrinsic motivation and found that neural decreases in caudate activities were related to the undermining effects. From these recent studies on intrinsic types of motivation, there is a common finding showing that intrinsic types of motivation share the neural circuits of reward processing (e.g., caudate, putamen). This finding is consistent with the hypothesis of common neural currency (e.g., striatal activities) for reward processing regardless of whether the nature of rewards is monetary, social, or primary (Aharon et al., 2001; Fehr & Camerer, 2007; Izuma, Saito, & Sadato, 2008). That is, very different environmental

34 23 rewards are understood neurally in similar ways namely, via the extent of striatal activations. However, the studies on intrinsic types of motivation have not examined the neural differences between intrinsic types of motivation and extrinsic types of motivation even though the study of Murayama and his colleagues (2010) raised a possibility that such neural differences exist. Findings from Studies on Volition In the fields of psychology and education, volition is defined as people s cognitive, emotional, and motivational control over the whole processes of learning, performance, and participation (Corno, 1993; Gollwitzer & Brandstätter, 1997; Heckhausen & Gollwitzer, 1987). This means that volition includes cognitive, emotional, and motivational processes after actions are initiated (i.e., how to implement intention), which is different from the concept of motivation mainly focusing on how actions are initiated (e.g., expectancy, values, goals). However, in the neuroscience field, volition means self-guided-action as contrasted with other-guided-action. That is, in neuroscience, the conceptualization of volition is narrowed to action control (Haggard, 2008); in psychology and education, the conceptualization of volition is broadened to include not only action control but also action-related cognitive, emotional, and motivational control. Findings from the neuroscientific studies indicate that the motor-related brain regions (e.g., primary motor cortex, supplementary motor area, pre-supplementary motor area), and the anterior cingulate cortex work for volitional processes (Brass & Haggard, 2007; Haggard, 2008; Nachev et al., 2005).

35 24 Findings from Studies on Craving and Addiction Findings from neuroscientific studies on craving and addiction have implications for understanding the neural bases of intrinsic types of motivation, because craving and addiction are aggressive or excessive forms of intrinsic types of motivation. Results of neuroscientific studies have indicated that the orbitofrontal cortex and the insular cortex are the well-known regions for craving and addiction, such as food craving, cigarette addiction, drug addiction, and so on (Brody et al., 2002; Goldstein et al., 2009; Naqvi & Bechara, 2009; Naqvi, Rudrauf, Damasio, & Bechara, 2007; Pelchat, Johnson, Chan, Valdez, & Ragland, 2004). In particular, Naqvi and his colleagues (2007) suggest that the orbitofrontal cortex and the insular cortex are associated with distinct functions of craving and addiction. That is, the orbitofrontal cortex is more related to conscious and value-based urges while the insular cortex is more related to automatic and subjective urges. This suggestion is based on findings showing that participants cease their addicted behavior after incurring insular cortex damage, while participants could not cease their addicted behavior after incurring orbitofrontal cortex damage. The findings indicate that the orbitofrontal cortex is related to addicted behavior due to its cognitive functions, such as associative learning between stimuli and reward (or hedonic) values while the insular cortex is related to addicted behavior due to its emotional functions, such as subjective and visceral feelings. Based on these findings, it can be assumed that there are different types of energy that motivate people to move forward, as one type of energy comes from inherent inner sources while another type of energy is learned and acquired from outer sources.

36 25 Research Questions and Hypotheses The present research pursued three research questions and tested three core hypotheses through a series of three fmri studies. The purpose of Study 1 was to address Research Question 1 and to test Hypothesis 1. The purpose of Study 2 was to address Research Question 2, to provide a second test of Hypothesis 1, and to test Hypothesis 2. The purpose of Study 3 was to address Research Question 3, to provide a third test of Hypothesis 1, and to test Hypothesis 3. Research Question 1 Are there neural differences between intrinsic motivation and extrinsic motivation? If so, what are those neural differences? Hypothesis 1 Based on the neuroscientific studies of volition, craving, and addiction, it can be hypothesized that the neural activities of the insular cortex, especially those of the anterior insular cortex, would be potentially related to intrinsic motivation. This is because the neural activities of the insular cortex are known to be related to intrinsic need satisfaction, and this self-satisfaction is assumed to be an important antecedent of intrinsic motivation. In contrast, the neural activities related to incentive motivation are well-known, and include the striatum regions (e.g., nucleus accumbens, caudate), the prefrontal regions (e.g., ventromedial prefrontal cortex, dorsolateral prefrontal cortex), the anterior and posterior cingulate cortex, and the limbic system (e.g., amygdala). Hypothesis 1 was that the neural activities of intrinsic motivation (i.e., anterior insular

37 26 cortex) would be different from the neural activities of extrinsic motivation (i.e., striatum regions, prefrontal regions, limbic system). More specifically, anterior insular cortex activities would be associated with intrinsic motivation. Research Question 2 Are the neural activities of intrinsic motivation associated with participants selfreported intrinsic need satisfaction? Hypothesis 2 As anterior insular activities are known to be related to intrinsic need satisfaction (i.e., intrinsic urges), it can be hypothesized that the neural activities of the anterior insular cortex would be correlated with participants general tendency to self-report intrinsic need satisfaction. That is, people who show greater anterior insular cortical activities during the initiation of intrinsic motivation would also be expected to show greater spontaneous satisfactions during intrinsically motivated behavior in the form of greater self-reported perceived autonomy and perceived competence, which are the subjective experiences that underlie intrinsic motivation. Participants with higher intrinsic psychological needs (for autonomy, competence) in general would tend to show higher intrinsic need satisfaction when encountering intrinsically motivating situations and, as a result, to show greater neural activities of the anterior insular cortex during the initiation of intrinsic motivation. Hypothesis 2 was that anterior insular cortical activities would be associated not only with intrinsic motivation but also with the subjective (i.e., self-report) experience of intrinsic psychological need satisfaction.

38 27 Research Question 3 Are there any additional neural substrates of intrinsic motivation? If so, what are those additional neural substrates? Hypothesis 3 Previous neuroscientific studies on intrinsic motivation have shown that intrinsic motivation during the task performance recruits the neural activities of the ventral striatum, just as extrinsic types of motivation do. This raises the possibility that intrinsic motivation and extrinsic motivation might not only have distinct neural activities (as per Hypothesis 1), but also that they might also have shared neural activities. Because ventral striatum activities have been found in studies of both extrinsic motivation and intrinsic motivation, it makes sense to offer an exploratory hypothesis that the two types of motivation might also share neural substrates. The ventral striatum was hypothesized to be one such shared neural substrate. Hypothesis 3 was that not only would anterior insular cortical activities be associated with intrinsic motivation but also that ventral striatal activities would be associated with intrinsic motivation.

39 28 CHAPTER III STUDY 1: NEURAL DIFFERENCES BETWEEN INTRINSIC REASONS FOR DOING VERSUS EXTRINSIC REASONS FOR DOING: AN FMRI STUDY Introduction In deciding whether or not to engage themselves in an activity, people consider their reasons for doing so. People can decide to engage in a task because they expect doing so will bring them an attractive consequence (e.g., anticipating money for completing a project), but people can also decide to engage in a task because they expect doing so will bring spontaneous self-satisfactions (e.g., experiencing enjoyment while completing a project). Social and educational psychologists argue that action energized and directed by such intrinsic reasons is qualitatively different from action energized and directed by extrinsic reasons (Ryan & Deci, 2000). This means that intrinsic motivation, which is an inherent and task-endogenous type of motivation, orients people toward an activity because of the anticipation of spontaneous self-satisfactions whereas extrinsic motivation (e.g., incentive-based motivation), which is an acquired and task-exogenous type of motivation, orients people toward an activity because of its acquired extrinsic values (Deci & Ryan, 1985). Whereas intrinsic motivation is commonly accepted as an important type of motivation by social and educational psychologists, it has not been accepted as a core motivational construct in studies rooted in the behavioral traditions, such as organizational psychology, behavioral sciences, and neuroscience. These traditions focus

40 29 rather exclusively on incentive motivation (Berridge, 2004; Locke & Henne, 1986). Findings from numerous neuroscience studies suggest that incentive-based decision making processes and goal-directed behaviors are associated with neuronal responses to stimuli that have become associated through experience with rewarding consequences (Cardinal et al., 2002; McClure et al., 2004; Schultz, 2000). Results of animal studies have consistently shown that brain regions, such as the limbic system (e.g., amygdala), the ventral striatum (e.g., nucleus accumbens), the anterior and posterior cingulate cortex, and the frontal cortex (e.g., dorsolateral prefrontal cortex, ventromedial prefrontal cortex including the orbitofrontal cortex, anterior cingulate cortex), are related to rewardexpectation states and reward-related responses or behaviors (Cardinal et al., 2002; Hayden, Nair, McCoy, & Platt, 2008; Mogenson et al., 1980; Robbins & Everitt, 1996; Schultz, 2000). Researchers have also suggested that these brain regions are similarly activated during human reward processing (Cardinal et al., 2002; Hampton & O Doherty, 2007; McClure et al., 2004; O Doherty, 2004; Plassmann, O Doherty, & Rangel, 2007). While impressive in many respects, an exclusive focus on incentive motivation falls short of addressing contemporary motivational concepts, such as intrinsic motivation, autonomy, flow, challenge-seeking, curiosity, and so on (Reeve, 1996). Recognizing that contemporary neuroimaging investigations typically exclude inherent and taskendogenous types of motivational concepts, the viability of expanding the neural understanding of motivation was pursued by initiating a pioneering study of intrinsic motivation by scanning participants neural activities when they decided to act for intrinsic reasons versus when they decided to act for extrinsic reasons using event-related functional magnetic resonance imaging (fmri).

41 30 The prediction for the neural bases of extrinsic motivation (i.e., incentive motivation) is not novel and instead reflects the well-established findings that the valuation system, such as ventromedial prefrontal cortical activities (e.g., the orbitofrontal cortex) and anterior and posterior cingulate activities, would be more recruited by decision making based on weighing attractive extrinsic reasons for doing (Bray, Shimojo, & O Doherty, 2010; Britton et al., 2006; Hayden et al., 2008; Maddock, Garrett, & Buonocore, 2003; Plassmann et al., 2007). The prediction for the neural bases of intrinsic motivation, however, is novel and represents a key open question in the study of affective neuroscience. When people are intrinsically motivated, they act out of personal interest and because they find the task at hand to be inherently enjoyable and capable of producing spontaneous self-satisfactions such as It s interesting and I enjoyed that. As people become aware of how tasks affect their subjective feelings when they formulate a conscious experience of my feelings about that thing they show greater insular cortex activities (Craig, 2009, p. 65). Hence, it was predicted that the insular cortex would be more recruited by decision making based on awareness of positive inherent feeling states (intrinsic reasons for doing) because insular activities is related practically to all inherent feelings (Craig, 2009) but particularly to inherent need satisfactions (Cardinal et al., 2002; Craig, 2002; Naqvi et al., 2007; Singer, Critchley, & Preuschoff, 2009).

42 31 Method Participants Ten undergraduates (6 females and 4 males; mean age: 19.7 ± 0.87), who were recruited from introductory educational psychology classes at the University of Iowa, participated 1. They were neurologically healthy, right-handed, native English speakers who had normal or corrected-to-normal vision. All participants provided informed consent in accordance with the regulations of the Institutional Review Board of the University of Iowa. Stimuli In this study, phrases were used to describe situations from the following three conditions: intrinsic motivation, extrinsic motivation, and amotivation (a control condition). The phrases were developed based on self-determination theory s conceptual and operational definitions of these three types of motivation (Ryan & Deci, 2000; see Table 1). Sixty familiar situations (e.g., reading a book, working on a computer, planning a project) were developed and three different reasons for doing each task were inserted to characterize the activity as motivated by the type of motivation unique to the experimental condition. 1 In the neuroimaging studies, the sample size concerns not only the number of participants but also the number of observations (i.e., task trials in this study) of each participant (Huettel, Song, & McCarthy, 2004). In this study, to compensate the small number of participants, a large number of observations (i.e., 60 trials for each condition) were used.

43 32 Table 1 Examples of phrases from each of the three experimental conditions used in the experimental task. Intrinsic motivation phrases Extrinsic motivation phrases Amotivation phrases Writing an enjoyable paper Writing an extra-credit paper Writing an assigned paper Working with freedom Working for incentives Working with pressure Participating in a fun project Participating in a money-making project Participating in a required project Having options and choices Having prizes and awards Having pressures and obligations Working because its fun Working because I want money Working because I have to Working on the computer out of curiosity Working on the computer to meet a deadline Working on the computer for bonus points Feeling interested Anticipating a prize Feeling frustrated In the intrinsic motivation condition, the phrases described situations that motivate people due to internal causalities, such as interest or enjoyment (e.g., writing an enjoyable paper, working on the computer out of curiosity, pursuing my personal interests in class). In the extrinsic motivation condition, the phrases described situations that motivate people due to attractive extrinsic incentives (e.g., writing an extra-credit paper, working on the computer for bonus points, pursuing an attractive reward in class). For the amotivation (literally without motivation ) condition, the phrases described situations that lead to merely compliant behavior (e.g., writing an assigned paper, working on the computer to meet a deadline, pursuing a routine task in class). The amotivation phrases were used as filler items to avoid participants skewed yes responses. The 60 sets of phrases were selected from a larger pool of 90 sets of phrases

44 33 based on a pilot test in which a separate group of participants rated the phrases using a computer presentation. The phrases across the three conditions were matched in terms of sentence structure and word length. Task and Procedure An event-related fmri experiment, which consisted of three runs, was performed. Each run lasted 10 minutes and consisted of 60 trials, which were randomly taken from each of the three conditions (20 trials per condition) and were presented in random order. In each trial (see Figure 1), a phrase was presented for three seconds to describe a situation related to one of the three conditions. During those three seconds, participants were asked to read the phrase and make a decision, Do you want to do this? yes or no?, by pressing the left button with the forefinger finger (for yes) or the right button with the middle finger (for no). Following this response, there was a jitter of 2-12 seconds (mean = 7 seconds) between each trial 2. Then, the next trial began, which presented a phrase describing another one of the three conditions. During the experimental session, participants first received the task instruction and practiced the experimental task before performing the real task during the brain image scans. Participants anatomic images were first acquired and then functional 2 If the longer interval (e.g., more than 10 seconds) between successive trials was used, the estimation efficiency (i.e., ability to differentiate individual hemodynamic response functions of stimuli) is improved but the detection power (i.e., ability to detect neural activations) is decreased. In contrast, the shorter interval (e.g., two seconds) makes the detection power increased but the estimation efficiency is worsened. To satisfy both issues, the method randomizing the medium-length intervals is generally used (Huettel, Song, & McCarthy, 2004).

45 34 images were scanned while participants performed the experimental task. After the brain image scans, participants were debriefed about the experiment and received compensations for their participation. Note. IM: intrinsic motivation; EM: extrinsic motivation; Amot: amotivation. Figure 1 Task and experimental design. fmri Data Acquisition Imaging was performed with a 3T Trio MRI scanner (Siemens, Erlangen, Germany). First, T1-weighted anatomic images (TR = 1590 ms, TE = 3.58 ms, flip angle = 10, FOV = 256 X 256, and slice thickness = 2 mm) were acquired for anatomical localization using a MP-RAGE sequence in order to facilitate the precise determination of the structures corresponding to the functional activation foci. After obtaining anatomic images, 16-slice functional images were acquired using a T2*-weighted gradient-echo echo planar imaging (EPI) sequence sensitive to blood oxygenation level-

46 35 dependent (BOLD) contrast (TR = 2000 ms, TE = 30 ms, flip angle = 90, FOV = 220 X 220, 64 X 64 matrix, and slice thickness = 5 mm with 1mm gap). fmri Data Analysis Imaging preprocessing, individual analyses and statistical analyses were performed using AFNI (Cox, 1996; The first eight images of each run were discarded to allow hemodynamics and MRI signals to reach a steady state. In preprocessing, the functional images were checked to determine whether there were signal artifacts which could be made by participants head movement, scanner irregularities, and so on by using an AFNI program called 3dToutcount. Then, the functional images were interpolated to the same time point at the beginning of the TR for slice timing correction by using an AFNI program called 3dTshift. After the functional images of each participant were aligned to the structural images of each participant by using an AFNI program called 3dAllineate, the aligned functional images were registered to the base volume for head motion correction by using an AFNI program called 3dvolreg. These motion-corrected brain images were spatially smoothed with a 4 mm full-width at half-maximum (FWHM) Gaussian kernel by using an AFNI program called 3dmerge3. By this smoothing process, the functional images had smaller anatomical variability across participants, and this smoothing process helped reduce the errors caused by the 3 Even though the brain regions are anatomically the same, they are not exactly in the same locations across different people. Therefore, participants neural activations are smoothed to some degree (e.g., 4~8 mm) and then compared. The smoothing value was smaller than that used in Study 2 and Study 3 due to the smaller number of participants. However, 4 mm is in the acceptable range for the future statistical analyses (Huettel, Song, & McCarthy, 2004).

47 36 multiple comparisons. After the values of background voxels (i.e., voxels outside the brain) were excluded, the functional data were normalized as a percent of the mean by using an AFNI program called 3dcalc for running future statistical analyses. The functional images of each run were separately preprocessed, and then the two runs of each participant were concatenated before individual analyses by using an AFNI program called 3dTcat. In individual analyses, the preprocessed time-series data were analyzed by a general linear model by using an AFNI program called 3dDeconvolve. In this general linear model, hemodynamic response functions (HRF) 4 of 9 regressors were computed. Three regressors were for experimental conditions, intrinsic motivation (X 1 ), extrinsic motivation (X 2 ), and amotivation (X 3 ). To control for the effects of motion artifacts, six regressors for motion parameters were included as covariates in the model. The regression equation was: Y = β 1 X 1 + β 2 X 2 + β 3 X 3 + β 4 X 4 + β 5 X 5 + β 6 X 6 + β 7 X 7 + β 8 X 8 + β 9 X 9 + error variance. For group analyses, each individual s statistical data were transformed to Talairach space (Talairach & Tournoux, 1988). By using an AFNI program the high-resolution structural images of each participant were transformed to the standardized structural images. Then, by using an AFNI program called adwarp, the functional images of each participant were transformed to the standardized high-resolution structural images of each participant. At that time, the functional images were resampled to 2 X 2 X 2 mm 3 voxels. 4 Hemodynamic response function (HRF) means signal changes in blood flow, which can be observed due to blood oxygenation changes.

48 37 In the group analyses, subtraction analyses were performed to examine the neural differences between the two different types of motivation (intrinsic motivation vs. extrinsic motivation) by using an AFNI program called 3dANOVA2. For correcting multiple comparison inferences in whole-brain analyses, the cluster-wise threshold was employed based on Monte-Carlo simulations (Forman et al., 1995), which set at a value of.043 determined by a conjoined voxel-wise threshold (p <.005), a connectivity radius of 2.0 mm, and a minimum volume of 272 mm 3 (34 contiguous voxels). The significant activations for the subtraction analyses and the regression analysis were reported as Talairach coordinates after the MNI coordinates converted to the Talairach space by using a mni2tal algorithm (Lacadie, Fulbright, Rajeevan, Constable, & Papademetris, 2008). After activated brain regions in the above analyses were set as regions of interests (ROIs), time-series BOLD signal changes of each condition in these ROIs were examined in order to confirm the neural difference results from the subtraction analyses. To statistically compare the BOLD signal changes across conditions, repeated measures ANOVAs and Student-Newman-Keuls post hoc tests were conducted. SPSS 17.0 was used for these analyses. Results Behavioral Results Participants yes/no finger-press responses to the action question (Do you want to do this?) served as a behavioral indicator of approach-based motivated action. The mean percentages and the standard errors of participants motivated (yes) responses for phrases of the intrinsic motivation and extrinsic motivation conditions were 92.4 ± 2.09 % and

49 ± 5.52 % respectively, which were not significantly different from each other (d = 0.68), but only 23.8 ± 3.24 % for phrases in the amotivation condition, which were significantly lower than both experimental conditions, F (2,8) = , p <.05, d = 6.94 for intrinsic motivation versus amotivation, d = 2.81 for extrinsic motivation versus amotivation, with amotivation < intrinsic motivation = extrinsic motivation, using Student-Newman-Keuls post hoc tests (see Figure 2. A). Participants responses were therefore consistent with the experimental manipulation approach-oriented motivated responses for the intrinsic motivation and extrinsic motivation phrases and unmotivated responses for the amotivation phrases thereby confirming that the experimental manipulation was successful. Means and standard errors for the reaction time responses for phrases of the intrinsic motivation, extrinsic motivation, and amotivation conditions were ± 62.5 ms, ± 72.0 ms, and ± 63.1 ms respectively. Results showed that participants showed significantly shorter reaction times in the intrinsic motivation condition than in the extrinsic motivation condition (d = 2.83) and in the amotivation condition (d = 2.01) while participants reaction times in the extrinsic motivation condition and those in the amotivation condition were not significantly different (d = 0.55), F (2,8) = 48.10, p <.05, with intrinsic motivation < extrinsic motivation = amotivation, using Student-Newman-Keuls post hoc tests (see Figure 2. B).

50 39 Note. IM: intrinsic motivation; EM: extrinsic motivation; Amot: amotivation. * p <.05. Figure 2 Participants mean yes percentage and mean reaction time. fmri Results Results of the subtraction analyses (see Table 2) between the intrinsic motivation and extrinsic motivation conditions showed that the right insular cortex was more activated in the intrinsic motivation condition than in the extrinsic motivation condition (corrected p <.043; see Figure 3. A). Time-series BOLD signal change patterns of this right insular cortex activities between the intrinsic motivation and extrinsic motivation conditions were also extracted, and these data also showed increase activations in the intrinsic motivation condition that were consistent with the results of the subtraction analysis (see Figure 3. B).

51 40 Table 2 Results of the subtraction analysis between the IM and EM conditions. Region BA Volume Side Talairach Coordinates x y z Maximum t value IM EM Insular cortex R EM IM Posterior cingulate cortex R Note. The cluster-wise threshold (correct p <.043) is determined by voxel-wise threshold (p <.005), the connectivity radius (2.0 mm), the minimum volume (34 contiguous voxels, 272 mm 3 ), and the FWHM (4 mm). IM: intrinsic motivation; EM: extrinsic motivation. In contrast, the extrinsic motivation condition showed greater neural activities of the right posterior cingulate cortex, which is a brain region of the valuation system, than did the intrinsic motivation condition (corrected p <.043; see Figure 4. A). Time-series BOLD signal change patterns of this right posterior cingulate cortex activities between the intrinsic motivation and extrinsic motivation conditions were again extracted, and these data also showed increase activations in the extrinsic motivation condition that were consistent with the results of the subtraction analysis (see Figure 4. B).

52 41 Note. IM: intrinsic motivation; EM: extrinsic motivation. Figure 3 A. The insular cortex was more activated in the intrinsic motivation condition than in the extrinsic motivation condition. B. The time-series BOLD signal changes of the insular cortex are presented.

53 42 Note. IM: intrinsic motivation; EM: extrinsic motivation. Figure 4 A. The posterior cingulate cortex was more activated in the extrinsic motivation condition than in the intrinsic motivation condition. B. The timeseries BOLD signal changes of the posterior cingulate cortex are presented.

54 43 Discussion The purpose of this study was to address the new question of Are the neural bases of intrinsic motivation different from those of extrinsic motivation? To address this question, the neural differences were identified as people made decisions whether or not to engage in familiar activities but for the very different reasons that related either to intrinsic motivation or to extrinsic motivation. Doing so was expected to provide the evidence necessary to extend the neuroscientific conception of motivation beyond an exclusive focus on extrinsic motivation to include intrinsic motivation as well. This study provides evidence that participants did recruit different patterns of neural activities during the decision making process depending on intrinsic versus extrinsic reasons for doing. The insular cortex functioned more during the decision making process based on intrinsic reasons for doing, while the posterior cingulate cortex functioned more during the decision making process based on extrinsic reasons for doing. The general function of the insular cortex is the processing of subjective feelings generated by bodily information (Craig, 2009; Damasio et al., 2000; Pessoa, 2008; Phan, Wager, Taylor, & Liberzon, 2002). In particular, the insular cortex is known to be associated with self-satisfactions, intuitive feelings, and basic urges (Brody et al., 2002; Naqvi et al., 2007). In contrast, the general function of the posterior cingulate cortex is evaluative processing of external stimuli (Hayden et al., 2008; Maddock et al., 2003). Among the valuation system, the posterior cingulate cortex is particularly known to be related to subjective value formation using social knowledge (Johnson et al., 2006; Schiller, Freeman, Mitchell, Uleman, & Phelps, 2009).

55 44 Based on these results, it can be assumed that, in this study, participants in the intrinsic motivation condition decided that they wanted to engage in the activities based on the presence of spontaneous self-satisfactions (e.g., enjoyment, interest, freedom), while participants in the extrinsic motivation condition decided that they wanted to engage in the activities based on socially-acquired values (e.g., incentive, extra-credit, prize). These assumptions are supported by the reaction time results showing that participants engaged in faster responses to the intrinsic motivation phrases than to the extrinsic motivation phrases. This supports the interpretation that participants made relatively quick gut felt decisions about intrinsic reasons for acting while they made calculated cost-benefit decisions (e.g., is this consequence attractive enough to be worth the effort?) about extrinsic reasons for acting (Bechara & Damasio, 2005). Intrinsic motivation theorists propose that human motivation is not singular (Ryan & Deci, 2000). They argue that qualitatively different types of motivation exist and, further, that each type of motivation is generated by different energy sources. In particular, they distinguish intrinsic motivation, which is generated by inherent processes, from extrinsic motivation, which is generated through environmental contingencies (Deci & Ryan, 1985). Neural evidence from the present study supports these assumptions. When participants in the present study imagined the intrinsic motivation situations, they were assumed to decide to engage in the situations based on their inherent-feeling need satisfaction processing. This is an important point, because intrinsic motivation theorists define intrinsic motivation as that which arises from the satisfaction of inherent psychological needs (for autonomy, competence, and relatedness; Ryan & Deci, 2000). If the situations were perceived as inherently need satisfying, positive feelings led

56 45 participants to freely want to approach the described situation. In contrast, when participants imagined the extrinsic motivation situations, they made their decision to engage in the situations based on the rational consideration of whether the described situation offered an attractive enough benefit to warrant action. This means that intrinsic motivation is produced more by the presence of endogenous positive feelings, which emanate out of the intuitive processing of spontaneously experienced self-satisfactions, while extrinsic motivation is produced more by the environmentally-associated benefits that task engagement is expected to generate, which emanate out of the processing of stored values and environmental contingencies. Based on self-determination theory and the neural understanding of the insular cortex, intrinsic need satisfaction is assumed to be the psychological property that explains the link between intrinsic motivation and insular cortex activities. However, there exists little empirical evidence supporting this hypothesis. In this regard, I conducted the second experiment additionally identifying the relation between intrinsic need satisfaction and the neural activities of the insular cortex during the experience of intrinsic motivation.

57 46 CHAPTER IV STUDY 2: EXPERIENCING SELF-DETERMINED VERSUS NON-SELF- DETERMINED REASONS FOR ACTING: THE NEURAL CORRELATES OF DIFFERENT MANIFESTATIONS OF PERSONAL AGENCY Introduction People are sometimes motivated into action because they expect what they do will produce an attractive external contingency (e.g., a reward); other times, people are motivated into action because they expect what they do will produce feelings of spontaneous satisfaction (e.g., interest, enjoyment). That people can be energized and directed into action (i.e., motivated) extrinsically or intrinsically is a basic, though controversial, issue in contemporary motivation study (Deci et al., 1999a; Lepper, Corpus, & Iyengar, 2005; Ryan & Deci, 2000; Sansone & Harackiewicz, 2000). Both extrinsic (i.e., incentive) and intrinsic types of motivation produce volitional, agentic action. When either extrinsically or intrinsically motivated, the person has the subjective experience of personal causation and of being the agentic cause of the outcome-seeking motoric action. What distinguishes between the two types of motivation is the sought-after outcome, or reason why the person engages in volitional, agentic action. If the reason for action is incentive-based (e.g., I read the book to gain parental approval), then the motivated action is extrinsically motivated in that it is both environmentally-generated and environmentally-regulated. That is, as environmental incentives arise, change, and disappear, the person s volitional and agentic action changes

58 47 in kind. If the reason for action is self-based (e.g., I read the book because it interests me), then the motivated action is intrinsically motivated in that it is both self-generated and self-regulated. As self-satisfaction experiences (e.g., interest, enjoyment, autonomy) arise, change, and disappear, the person s volitional and agentic action changes in kind. According to self-determination theory (Deci & Ryan, 1985; Ryan & Deci, 2000), a key distinction can be made within the concept of volitional, agentic action in that some volitional actions are initiated and regulated by an expression of the self, whereas other volitional actions are initiated and regulated by environmental forces. The former is referred to as self-determined behavior, and its prototype is intrinsically-motivated behavior; the latter is referred to as environmentally-determined (or non-selfdetermined ) behavior, and its prototype is extrinsically-motivated behavior (Ryan & Deci, 2000). This distinction between self-determined and non-self-determined volitional, agentic action is important theoretically (as above) but also practically as people (e.g., students, workers, athletes) who are intrinsically motivated function more positively in a wide variety of important ways than do people who are extrinsically motivated (with positive functioning being indicated by the extent of learning, engagement, performance, achievement, and well-being outcomes; Black & Deci, 2000; Deci, Schwartz, Sheinman, & Ryan, 1981; Jang, Reeve, Ryan, & Kim, 2009; Vansteenkiste, Simons, Lens, Sheldon, & Deci, 2004). In neuroscience, there is little evidence to clarify whether self-determined intrinsic motivation constitutes a qualitatively different type of motivation from non-selfdetermined extrinsic motivation. The absence of such evidence suggests a unidimensional conceptualization of the motivation underlying volitional and agentic action,

59 48 one that questions the validity of the self-determination theory view conceptualizing that, first, different types of volitional and agentic motivation exist and that, second, one type of motivation (i.e., self-determined intrinsic motivation) leads to significantly more positive functioning than does the other (i.e., non-self-determined extrinsic motivation). Neuroscience-based agency research focuses rather exclusively on the notion of who initiates and regulates actions, not on the notion of why the person does. That is, numerous studies have tried to determine the neural substrates of agency by examining the neural differences between self-generated versus other-generated behavior. In these studies, the contrast is between personal action (e.g., move a joystick) that is caused by self versus personal action that is caused by someone else such as the experimenter (e.g., I held the joystick and I moved it vs. I held the joystick but the experimenter moved it ). Findings from these studies suggest that the neural activities of the prefrontal cortex, insular cortex, cerebellum, and motor-related regions (e.g., supplementary motor area, pre-supplementary motor area, precentral gyrus, and postcentral gyrus) are related to the execution, observation, or imagination of self-generated behavior (Gallagher, 2000; Haggard, 2008; Nachev, Kennard, & Husain, 2008). Specifically, some studies demonstrated that the activities of the anterior insular cortex were more associated with self-generated (i.e., more agentic in this literature) behavior while the activities of the angular gyrus were more associated with other-generated (i.e., less agentic in this literature) behavior (Farrer et al., 2003; Farrer & Frith, 2002). The present study was designed to extend the previous distinction between volitional, agentic, and self-generated action versus non-volitional, non-agentic, and other-generated action to test the idea that two important types of motivation exist within

60 49 the former concept namely, within the concept of volitional, agentic, and self-generated behavior. Specifically, the present study was designed to identify the neural differences between self-determined versus non-self-determined volitional, agentic behavior (e.g., I did it volitionally to satisfy my interest vs. I did it volitionally to obtain the reward) and to check whether these neural results are consistent with the results of Study 1. The research strategy was to have people imagine the enactment of volitional, agentic behavior on the same task but for two different reasons, one of which was a selfdetermined (i.e., intrinsic) reason and the other of which was a non-self-determined (i.e., extrinsic) reason. For the self-determined reasons, reasons for action emphasized by selfdetermination theorists, including acting out of interest, enjoyment, autonomy, and competence, were selected (see Ryan & Deci, 2000). What these reasons have in common is an inner endorsement of one s actions, which is the sense that one s volitional, agentic behavior is actually self-authored, emanates from the self, and is one s own (Ryan & Deci, 2000). For the non-self-determined reasons, reasons for action undertaken in order to attain an attractive environmental reward or incentive, including acting in order to obtain money, a high grade, public recognition, and extra credit points, were selected. What these reasons have in common is an other endorsement of the action and the sense that one s behavior is other-authored, emanates from the presence of environmental contingencies, and is someone else s idea. Imagination of behaviors was used as the experimental paradigm because numerous studies have shown that imagining behavior has worked as well as the actual execution of behavior for generating a sense of volition and agency (Decety et al., 1994; Gallese & Goldman, 1998; Ruby & Decety, 2001). Importantly, this strategy had the

61 50 advantage of sampling neural activities in response to a broad and representative range of self-determined versus non-self-determined reasons for acting. To formulate the experimental hypotheses, it is argued that extrinsicallymotivated behavior and non-self-generated behavior are functional equivalents in everyday practice, as both are generated and regulated by the offering of attractive incentives and consequences (e.g., my boss offered a bonus if I worked, my teacher promised extra credit points if I studied), even if the extrinsically-motivated behavior is volitional and agentic in the sense that the person initiates the action. Recognizing this, it was hypothesized that non-self-determined extrinsically-motivated reasons for volitional and agentic acting would be related to angular gyrus activities (as per Farrer et al., 2003; Farrer & Frith, 2002). The angular gyrus is a part of the parietal brain region that was shown to be related to extrinsic motivation in Study 1 (see Figure 4). Likewise, it was argued that intrinsically-motivated behavior and self-generated behavior are functional equivalents in everyday practice, as both are generated and regulated by the experiencing of self-satisfactions (e.g., I worked because it was fun, I studied because I was curious to learn something new). Recognizing this, it was hypothesized that self-determined intrinsically-motivated reasons for volitional and agentic acting would be associated with anterior insular cortex activities (as per Farrer et al., 2003; Farrer & Frith, 2002). Stated differently, the hypothesis was that only self-determined, intrinsically-motivated reasons for volitional action would be associated with anterior insular cortex activities, while non-self-determined, extrinsically-motivated reasons for volitional action would not. In fact, it was hypothesized that non-self-determined, extrinsically-motivated reasons for

62 51 volitional action would actually be associated with angular gyrus activities, which characterized non-agentic action in previous studies. To confirm that the hypothesized anterior insular cortex activities were particularly linked to the expectation of intrinsic satisfactions, it was identified whether participants subjective experiences of perceived autonomy and perceived competence correlated with the extent of anterior insular cortex activities (as well as other neural activities) during self-determined behavior. These two particular types of intrinsic satisfaction were assessed because they constitute the definitional bases of intrinsic motivation: Intrinsic motivation is based on the innate, organismic needs for competence and self-determination. It energizes a wide variety of behaviors and psychological processes for which the primary rewards are the experiences of effectance and autonomy. (Deci & Ryan, 1985, p. 32). Method Participants 16 Korean undergraduates (8 females and 8 males; aged 19-24), who were recruited from a large university in Korea, participated in the fmri study. They were right-handed and had no history of neurological illness. All participants provided informed consent in accordance with the regulations of the Institutional Review Board of Korea University and received compensation for their participation.

63 52 Stimuli Korean phrases were used to describe situations from the following three conditions: self-determined intrinsically-motivated reasons for action (IM), non-selfdetermined extrinsically-motivated reasons for action (EM), and neutral reasons for action (Neutral). The phrases were developed based on theoretical postulates and operational definitions from self-determination theory (Deci & Ryan, 1985; Ryan & Deci, 2000; see Table 3). While participants read the phrases written in Korean, the example of phrases shown in Table 3 are provided in English for the purposes of this paper. Table 3 Examples of phrases from each of the three experimental conditions used in the experimental task. Intrinsic motivation phrases Extrinsic motivation phrases Neutral phrases Writing an enjoyable paper Writing an extra-credit paper Writing a class paper Working with freedom Working for incentives Working with time to spare Participating in a fun project Participating in a money-making project Participating in a routine project Having options and choices Having prizes and awards Having things to do Studying for fun Studying for a grade Studying because it is time Feeling curious Feeling rewarded Feeling neutral Feeling interested Anticipating a prize Feeling normal

64 53 Each phrase consisted of two parts, one part described a familiar situation to imagine acting on (e.g., reading a book, planning a project) and the other part provided one of the three different reasons for acting. In the IM condition, the phrases described situations with reasons for acting that featured internal causalities, such as interest, enjoyment, or perceived autonomy (e.g., writing an enjoyable paper). In the EM condition, the phrases described situations with reasons for acting that featured attractive extrinsic incentives (e.g., writing an extra-credit paper). In the Neutral condition, the phrases described situations with only a neutral reason for acting (e.g., writing a class paper). The phrases across the three experimental conditions were matched in terms of sentence structure and word length. To verify the suitability of the phrases, a pilot test was conducted in advance. 23 Korean undergraduates (10 females, 13 males; aged 20-28), recruited from the same large university in Korea, participated. In this computerized pilot test, participants rated 180 phrases that consisted of 60 phrases from each of the three experimental conditions. During the pilot test, each phrase, randomly selected from the array of 180 phrases, was presented sequentially, and participants were asked to rate each described situation on a series of four 1-7 uni-polar scales in terms of how behaviorally energizing, intrinsically motivating, extrinsically motivating, and attractive it was perceived to be (see Figure 5). Prior to viewing the phrases, participants received instructions on the conceptual definitions of what constituted behavioral energization, intrinsic motivation, extrinsic motivation, and attractiveness. Data of one male participant were excluded from the analyses because his responses were not recorded (equipment failure). Results from the pilot test appear in the Results section. From the larger pool of 180 phrases tested in the

65 54 pilot test, 150 phrases (50 triads) rated as expected on these four dependent measures and were selected for inclusion in the actual study. Figure 5 Procedure of the pilot test. Task and Procedure During the fmri scan, there were two separate runs, and each run consisted of 75 trials of 25 different sets. In each trial (see Figure 6), one phrase, randomly selected from the whole array of the phrases, was presented for four seconds, and a fixation cross was presented at the inter-trial interval (ITI) for an average of three seconds (1000ms- 7000ms). Participants rated how much they wanted to engage in each of the described

66 55 situations on a 1-4 scale ( Do you want to do this? : 1 = not at all; 2 = a little; 3 = some; 4= a great deal). The experimental manipulations (e.g., 4-second phrase presentation, asking the participants to make this engagement rating) encouraged the participants to mentalize the described situation. To indicate their judgments, participants were asked to press one of the four buttons by using their right hand. Before entering the fmri setting, participants completed the basic psychological needs scale (described below) and received the task instruction. Once the participant was situated in the fmri setting, he or she completed four practice trials. During the fmri scan, functional images were acquired while participants performed the experimental task, and then anatomic images were acquired. At the end of the experiment, participants were debriefed about the experiment and received their compensation for participation. Note. IM: intrinsic motivation; EM: extrinsic motivation; Neu: neutral. Figure 6 Task and experimental design.

67 56 Measure Basic Psychological Needs Scale (see Appendix). The basic psychological needs scale (Gagné, 2003) featured 13 items to assess the extent to which the psychological experiences of autonomy and competence occur in the person s life. Sample items are I feel like I can decide for myself how to live my life (autonomy) and I often do not feel very capable (competence, reverse scored). The scale has been widely used and has been shown to produce acceptable psychometric properties. Each set of items for autonomy (average α =.71; α s ranged from.68 to.73) and that for competence (average α =.73; α s ranged from.71 to.75) showed acceptable internal consistency (Gagné, 2003; Niemiec, Ryan, & Deci, 2009; Wei, Shaffer, Young, & Zakalik, 2005). In addition, this scale has been shown to correlate highly with another measure of basic psychological needs (r =.68 for autonomy, r =.68 for competence) and to predict productive and satisfying learning outcomes, such as achievement, engagement, and intrinsic motivation (Gagné, 2003; Jang et al., 2009). In the present study, the Korean translated version of the basic psychological needs scale (Jang et al., 2009) was used. To complete the scale, participants rated on a 1-7 Likert scale (Strongly disagree-strongly agree) to indicate how true each statement was for them in general. Scores in the present study showed high internal consistency (α =.75,.75,.83 for autonomy, competence, and the total score). fmri Data Acquisition Imaging was performed with a 3T Trio MRI scanner (Siemens, Erlangen, Germany). First, 32-slice functional images were acquired using a T2*-weighted gradient-echo echo planar imaging (EPI) sequence sensitive to blood oxygenation level-

68 57 dependent (BOLD) contrast. The following imaging parameters were used: TR = 2000 ms, TE = 30 ms, flip angle = 90, FOV = 224 X 224, in-plane resolution = 3.5 X 3.5 mm, and slice thickness = 4 mm (no gap). High-resolution T1-weighted structural images were acquired by using a MP-RAGE sequence. These images were used for anatomical localization to facilitate the precise determination of the structures corresponding to the functional activation foci. The following imaging parameters were used: TR = 1900 ms, TE = 2.52 ms, flip angle = 9, FOV = 256 X 256, and slice thickness = 1 mm (no gap). fmri Data Analysis The brain images were analyzed using AFNI (Cox, 1996; The first three images of each run were discarded to allow hemodynamics and MRI signals to reach a steady state. The procedure of the preprocessing was consistent with that of Study 1. First, the functional images were checked to determine whether there were signal artifacts which could be made by participants head movement, scanner irregularities, and so on. Then, the functional images were spatially and temporally realigned (i.e., slice timing correction and head motion correction) for data correction. These realigned brain images were spatially smoothed with a Gaussian kernel of 5 mm full-width at half-maximum (FWHM) Gaussian kernel. After the values of background voxels (i.e., voxels outside the brain) were excluded, the functional data were normalized as a percent of the mean for running future statistical analyses. The functional images of each run were separately preprocessed, and then the two runs of each participant were concatenated before individual analyses.

69 58 In individual analyses, the preprocessed time-series data were analyzed by a general linear model (GLM) using the three regressors of experimental conditions for IM, EM, and Neutral, and the six regressors for motion parameters as covariates which were convoluted with hemodynamic response functions (HRF). For group analyses, each individual s statistical data were transformed to Talairach space (Talairach & Tournoux, 1988). First, the high-resolution structural images of each participant were transformed to the standardized structural images. Then, the functional images of each participant were transformed to the standardized high-resolution structural images of each participant. At that time, the functional images were resampled to 2 X 2 X 2 mm 3 voxels. In the group analyses, subtraction analyses were conducted to compare the neural differences between the IM and EM conditions. A regression analysis was also conducted to examine the correlations between participants scores of the basic psychological needs scale and the neural activities in the IM condition. For correcting multiple comparison inferences in these whole-brain analyses, the cluster-wise threshold (corrected p <.05) was employed based on Monte-Carlo simulations (Forman et al., 1995), which was determined by both voxel-wise threshold (p <.005) and cluster size (n = 53, a minimum volume of 424 mm 3 ). The significant activations for the subtraction analyses and the regression analysis were reported as Talairach coordinates after the MNI coordinates converted to the Talairach space by using a mni2tal algorithm (Lacadie et al., 2008). After activated brain regions in the above analyses were set as regions of interests (ROIs), the BOLD signal changes of each condition in these ROIs were examined in order to compare the neural activation magnitudes of the ROIs across conditions. To

70 59 statistically compare the BOLD signal changes across conditions, repeat measures ANOVAs and Student-Newman-Keuls post hoc tests were conducted. SPSS 17.0 was used for these analyses. Results Pilot Test Results Results confirmed that the phrases in the IM and EM conditions worked as intended. Specifically, participants rated the phrases in the IM condition as more intrinsically motivating than they rated the phrases in the EM condition, t (20) = 5.77, p <.05, d = 1.40 (Ms, 5.8 vs. 4.5; see Figure 7. A), rated the phrases in the EM condition as more extrinsically motivating than they rated the phrases in the IM condition, t (20) = 6.58, p <.05, d = 1.15 (Ms, 6.0 vs. 4.7; see Figure 7. A), and rated the phrases in both the IM and EM conditions as more behaviorally energizing, F (2,20) = 49.64, p <.05, d = 2.23 for IM versus Neutral, d = 1.43 for EM versus Neutral, d = 0.37 for IM versus EM (IM, EM, and Neutral Ms, 5.9 = 5.7 > 4.9; see Figure 7. B) and as more positively valenced, F (2,20) = 55.33, p <.05, d = 2.25 for IM versus Neutral, d = 1.37 for EM versus Neutral, d = 0.43 for IM versus EM (IM, EM, and Neutral Ms, 5.7 = 5.5 > 4.7; see Figure 7. C), than they rated the phrases in the Neutral condition.

71 60 Note. IM: intrinsic motivation; EM: extrinsic motivation; Neu: neutral. * p <.05. Figure 7 Results of the pilot test. Participants ratings of intrinsic motivation and extrinsic motivation (A), behavioral energization (B), and positive valence of attractiveness (C).

72 61 Behavioral Results The main effect of experimental condition on participants ratings of how much they wanted to engage in the described situations across the three conditions was significant, as the situations described in the IM and EM phrases were perceived as more motivating than were the situations described in the Neutral phrases, F (2,14) = 71.48, p <.05, d = 3.11 for IM versus Neutral, d = 2.33 for EM versus Neutral, d = 0.42 for IM versus EM (IM, EM, Neutral, Ms 3.3 = 3.2 > 2.4; see Figure 8. A). The main effect of experimental condition on participants reaction times (RT) to indicate extent of judgmental simplicity in deciding how much they wanted to engage in the described situations across the three conditions was significant, as mean RTs (in ms) in the IM and EM conditions (which did not differ significantly from one another) were significantly faster than were the RTs in the Neutral condition, F (2,14) = 10.82, p <.05, d = 1.24 for IM versus Neutral, d = 0.93 for EM versus Neutral, d = 0.37 for IM versus EM (IM, EM, Neutral Ms, = < ; see Figure 8. B).

73 62 Note. IM: intrinsic motivation; EM: extrinsic motivation; Neu: neutral. * p <.05. Figure 8 Participants mean behavioral energization rating (A) and mean reaction time (B) in seconds. fmri Results Results of the subtraction analyses (see Table 4) showed that the left anterior insular cortex cortex (see Figure 9. A) as well as the left superior temporal gyrus, the bilateral cerebellum, the left posterior insular cortex, the bilateral supplementary motor area, the right dorsolateral prefrontal cortex, the right occipital lobe, the right precentral gyrus, the right postcentral gyrus, and the right middle frontal gyrus were more activated in the IM condition than in the EM condition (corrected p <.05). BOLD signal changes of the hypothesized brain region (i.e., anterior insular cortex) were compared across conditions. Results confirmed that anterior insular cortex activities were activated in the IM condition while they were deactivated in the EM condition, F (2,14) = 13.84, p <.05, d

74 63 = 0.69 for IM versus Neutral, d = 0.62 for EM versus Neutral, d = 1.41 for IM versus EM (IM > Neutral > EM; see Figure 9. B). Table 4 Results of the subtraction analysis between the IM and EM conditions. Region BA Volume Side Talairach Coordinates x y z Maximum t value IM EM Superior temporal gyrus L Cerebellum 2864 R L R L Insular cortex L L Supplementary motor area R L Dorsolateral prefrontal cortex R Occipital lobe R Precentral gyrus R Postcentral gyrus R Middle frontal gyrus R EM IM Angular gyrus L Note. The cluster-wise threshold (correct p <.05) is determined by voxel-wise threshold (p <.005), the connectivity radius (2.0 mm), the minimum volume (53 contiguous voxels, 424 mm 3 ), and the FWHM (5 mm). IM: intrinsic motivation; EM: extrinsic motivation.

75 64 Note. AIC: anterior insular cortex; IM: intrinsic motivation; EM: extrinsic motivation; Neu: neutral. Figure 9 A. There were significantly greater brain activations of the anterior insular cortex in the IM condition than in the EM condition. B. The BOLD signal changes of the anterior insular cortex across conditions are presented.

76 65 The subtraction analysis further showed that the left angular gyrus (see Figure 10. A) was more activated in the EM condition than in the IM condition (corrected p <.05). BOLD signal change results of this brain region across conditions confirmed that the activities of the angular gyrus were activated in the EM condition while they were deactivated in the IM condition, F (2,14) = 12.15, p <.05, d = 0.21 for IM versus Neutral, d = 0.54 for EM versus Neutral, d = 1.28 for IM versus EM (EM > IM; see Figure 10. B). Only the extent to which participants showed anterior insular cortex activations in the IM condition correlated significantly with their self-reported intrinsic satisfactions in life (i.e., scores of autonomy and competence on the BPSN; see Table 5; Figure 11. A). Participants self-reported intrinsic satisfactions correlated significantly and positively with their bilateral anterior insular cortex activations (r =.81, p <.05 for the left anterior insular cortex, see Figure 11. B; r =.88, p <.05 for the right anterior insular cortex, see Figure 11. C). Table 5 Results of the regression analysis between participants self-reported intrinsic satisfactions and the neural activities in the IM condition. Region BA Volume Side Talairach Coordinates x y z Maximum t value Insular cortex L R Note. The cluster-wise threshold (correct p <.05) is determined by voxel-wise threshold (p <.005), the connectivity radius (2.0 mm), the minimum volume (53 contiguous voxels, 424 mm 3 ), and the FWHM (5 mm).

77 66 Note. IM: intrinsic motivation; EM: extrinsic motivation; Neu: neutral. Figure 10 A. There were significantly greater brain activations of the angular gyrus in the EM condition than in the IM condition. B. The BOLD signal changes of the angular gyrus across conditions are presented.

78 67 Note. AIC: anterior insular cortex; IM: intrinsic motivation. Figure 11 A. Anterior insular cortex activities in the IM condition were the only neural activities which correlated positively with participants perceived intrinsic satisfactions. The correlations between the BOLD signal changes of the left anterior insular cortex (B) and the right anterior insular cortex (C) in the IM condition and participants scores of the basic psychological needs scale are presented. Discussion Previous research on agentic action has shown that people display distinct neural activities when their behavior is self-generated rather than other-generated. Selfgenerated agentic action, however, arises for two reasons intrinsically from what self-

79 68 determination theory labels as truly self-determined motivation as people engage in action from internal causalities and for internal satisfactions such as interest and enjoyment versus extrinsically from what self-determination theory labels as non-selfdetermined motivation as people engage in action from environmental causalities and for extrinsic incentives and rewards. This distinction has been shown to be important in behavioral research as self-determined motivation leads to more positive functioning and to more positive outcomes than does non-self-determined motivation. Recognizing the theoretical and practical significance of this distinction in the quality of personal agency, the present study sought to identify the neural substrates of the experience of selfdetermined personal agency and to distinguish them from the neural substrates of the experience of non-self-determined personal agency. The neural results of this study showed that the anterior insular cortex, the motorrelated regions, and the cerebellum were more activated when participants imagined selfdetermined action than when they imagined non-self-determined action, even while imagining acting in the same situation. Considering these neural activities are known to be associated with the sense of volition and agency (Farrer et al., 2003; Farrer & Frith, 2002; Gallagher, 2000; Haggard, 2008; Ruby & Decety, 2001), this finding suggests that self-determined intrinsically-motivated behavior is more volitional and agentic than is non-self-determined extrinsically-motivated behavior. The neural activities of the anterior insular cortex in Study 2 have particular importance. First, these results replicate the finding in Study 1. Considering that the participants of Study 1 were English-speaking USA undergraduates while those of Study 2 were Korean-speaking Korean undergraduates, these results suggest generalizability

80 69 across people with different cultural backgrounds. In fact, the neural activities of the anterior insular cortex were slightly different between the two studies that the anterior insular cortex in the right hemisphere was observed in the subtraction between the intrinsic and extrinsic motivation conditions of Study 1 while the left anterior insular cortex was observed in the same subtraction of Study 2. However, there is no need to focus on this hemispheric difference because the correlational result between the neural activities and participants self-report confirmed the link between the anterior insular cortex in both hemispheres and participants perceived intrinsic need satisfaction. This correlational result also provides evidence explaining the relation between intrinsic motivation and the neural activities of the anterior insular cortex. That is, among the neural activities during the experience of intrinsic motivation, the neural activities of the anterior insular cortex are assumed to be particularly linked to intrinsic need satisfaction, a basis of intrinsic motivation. The insular cortex is known to be linked to representation of bodily states (Craig, 2009; Damasio, 1999). This means that information of bodily states is stored in and retrieved from the somatosensory map, and the insular cortex works for the function of this somatosensory map. Therefore, the activities of the insular cortex, including the anterior insular cortex, have been repeatedly observed in numerous studies, including Study 1, on psychological processes which are influenced by bodily information, such as feeling processes (Craig, 2009; Damasio et al., 2000), feeling-based decision-making processes (Bechara & Damasio, 2005), and so on. Insular cortex activities, particularly anterior insular cortex activities, in the studies on agency can be understood in this same vein (Farrer & Frith, 2002; Gallagher, 2000; Ruby & Decety, 2001). That is, the sense of agency is also influenced by insular cortex activities which

81 70 are linked to the processes of the self (i.e., monitoring bodily information). Furthermore, anterior insular cortex activities were observed to be more activated as the degree of the sense of agency was increased (Farrer et al., 2003). Based on the findings from the studies on addiction and craving, the insular cortex is suggested to be associated specifically with bodily needs among bodily information. That is, insular cortex activities are linked to the urge for satisfying these bodily needs. This suggestion has been supported by the evidence showing that the insular cortex, including the anterior insular cortex, reacts to the target cues of cravings (Brody et al., 2002; Critchley, Wiens, Rotshtein, Öhman, & Dolan, 2004; Pelchat et al., 2004), dysfunctions of the insular cortex are related to abnormal cravings, such as addicted behaviors (Goldstein et al., 2009; Naqvi & Bechara, 2009), and people having addicted behaviors quit their addicted behaviors after incurring damage to the insular cortex (Naqvi et al., 2007). In the present study, the anterior insular cortex activities uniquely revealed significant and positive correlations with participants subjective experiences of perceived intrinsic satisfaction. That is, participants who showed higher intrinsic satisfaction in general revealed higher extent of anterior insular cortex activities when they imagined self-determined behavior and expected sought-after intrinsic satisfaction. The results of the present study therefore show the close relationships between anterior insular cortex activities and participants striving for intrinsic satisfaction. The motor-related regions, such as the supplementary motor area, the precentral gyrus, and the postcentral gyrus, and the cerebellum were expected to be activated by the experimental paradigm. In the studies on agency, especially the studies in which participants were required to imagine the enactment of behavior, these two brain regions

82 71 have been repeatedly observed (Farrer & Frith, 2002; Gallagher, 2000; Ruby & Decety, 2001). These observations were consistent with the general understanding that the motor-related regions are linked to motor imagery and the cerebellum is linked to the prediction of consequences of this motor imagery (Blakemore & Decety, 2001; Blakemore, Frith, & Wolpert, 2001; Lotze et al., 1999). The neural results of the present study also revealed that the left angular gyrus was the only brain region showing more neural activations during the sense of non-selfdetermined behavior than during the sense of self-determined behavior. Neural evidence suggests that these neural activities are associated with the sense of a loss of agency. Even though the right hemisphere of these posterior parietal regions were emphasized as dominant regions activated by the sense of loss of agency, the left hemisphere has been also observed to be activated together with the right hemisphere in many studies on agency (Farrer et al., 2003; Farrer & Frith, 2002). These results suggest that non-selfdetermined behavior is less agentic than is self-determined behavior, which is consistent with the suggestion based on the anterior insular cortex activities of this study. As explained earlier, these posterior parietal activities during non-self-determined behavior were assumed to be particularly linked to extrinsic reasons for acting. This means that participants appeared to sense low degrees of self-authored agency when they expected to gain extrinsic satisfaction from behavior. In numerous studies on agency, the posterior parietal regions tend to be more activated during observation or imagination of other-generated behavior than during those of self-generated behavior (Farrer & Frith, 2002; Ruby & Decety, 2001). Therefore, these brain regions have been generally assumed to be activated as other-generated

83 72 behavior reduces the sense of agency. Additionally, the posterior parietal regions are more activated when participants observe greater discrepancies between expected vs. actual outcomes of self-generated behavior (Farrer et al., 2003). This finding suggests that these brain regions are more activated as behavior, even self-generated behavior, is perceived to be less accurately self-controlled (i.e., less agentic). In the current study, the posterior parietal regions were more activated when participants imagined behavior that was initiated and regulated by environmental forces (i.e., extrinsic reasons for acting) than when participants imagined behavior that was initiated and regulated by the self (i.e., intrinsic reasons for acting). This means that the posterior parietal regions are more activated when extrinsic reasons for acting cause self-generated behavior than when intrinsic reasons for acting cause self-generated behavior. Considering these posterior parietal regions are known to be related to the understanding of social knowledge (Chiao et al., 2009; Culham & Valyear, 2006; Fogassi et al., 2005), perceiving environmental influences on the self-related processes is assumed to reduce the sense of agency. The meaning of agency basically revolves around the phenomenological sense of my own voluntary behavior (Bandura, 2001; Deci & Ryan, 1991; Gallagher, 2000). However, there have been different ideas on the meaning of my own. One approach has emphasized that my own is determined by the notion of who generates behavior (self vs. others). In this approach, the sense of agency is the feeling that I generate behavior (Gallagher, 2000). Another approach has emphasized that my own is determined by the notion of why I generate behavior (Deci & Ryan, 1991). This approach suggests that non-self-determined behavior is less agentic when it is enacted for the pursuit of attractive environmental contingencies (that are often determined by others)

84 73 and is more agentic when it is enacted for the pursuit of intrinsic satisfactions (that are inherently determined by self-satisfactions). This finding supports the latter approach to conceptualize the meaning of agency as being relatively less about who generates behavior (i.e., self vs. others) and relatively more about the reasons why the self generates voluntary action. Two aspects of the methodology limit the conclusions that can be drawn from the present study. The first limitation is that participants imagined enacting self-generated behavior, rather than actually acting it out. While past research has shown that imagining such behavior works as well as actually executing it in terms of generating a sense of agency, it still seems necessary to test our hypothesis using actual action. The second limitation is that participants also imagined their reasons for acting, rather than experienced the self-satisfaction and reward receipt directly. This is a limitation because additional neural activity may emerge with the actual receipt of intrinsic satisfaction and extrinsic reward that were not observed in the present study (e.g., reward-related striatal activations). For these reasons, I conducted the third experiment that can identify the neural substrates of intrinsic motivation during the engagement of actual interesting tasks.

85 74 CHAPTER V STUDY 3: NEURAL SUBSTRATES OF INTRINSIC MOTIVATION DURING THE TASK PERFORMANCE: AN FMRI STUDY Introduction The neural correlates of intrinsic motivation were examined in Study 1 and Study 2. Results showed that the neural activities of intrinsic motivation could be distinguished from extrinsic motivation (Studies 1 and 2), from amotivation (Study 1), and from neutral motivation (Study 2). Specifically, activations of the anterior insular cortex, which are known to be related to the processing of intrinsic needs, were observed when participants decided to act due to intrinsic reasons. However, there is a limit to what the findings of Study 1 and Study 2 can explain. This is because Study 1 and Study 2 examined the neural correlates of intrinsic motivation only during the imagination of action, as participants were asked to make a decision (Do you want to do this?) related to a task known to be associated with intrinsic motivation (e.g., writing an enjoyable paper, feeling interested). The imagination of action may or may not be the same as the actual in-vivo experience of intrinsically motivated behavior. The purpose of Study 3 was to examine the neural correlates when participants performed interesting tasks that generate intrinsic motivation. The research strategy was to offer participants two versions of the same task one that was capable of generating intrinsic motivation and the other that was not and compare the neural activities during these two versions of a task to see if the participants neural activities during an interesting version of a task (i.e,. during intrinsic motivation) was significantly different

86 75 from participants neural activities during a non-interesting version of the same task (i.e., during neutral motivation). To create interesting versus non-interesting versions of the same task, the conceptual definition of intrinsic motivation was considered, as was a review of the type of experimental tasks that are used heavily in the social psychological and educational psychological literatures. From these considerations, one task was designed to contrast an opportunity to pique and satisfy curiosity versus not (following Berlyne, 1960; Kang et al., 2009; Loewenstein, 1994; Ryan & Deci, 2000) while the other task was designed to contrast an opportunity to seek out and master an optimal challenge or not (following Reeve, 1989; Ryan & Deci, 2000). These tasks are described in the Method section, but the important point is that both were designed to offer participants an opportunity for intrinsic need satisfaction (experience autonomy and competence) from the environmental conditions of curiosity-inducing questions and optimal challenge. There are a few neuroscientific investigations that are relevant to the study of the in-vivo experience of intrinsic motivation. One study examined intrinsic achievement motivation during the working-memory task performance and found that the experience of intrinsic achievement motivation on a problem-solving task gave rise to putamen activities (Mizuno et al., 2008). Another study examined curiosity, and this study also found that the experience of intrinsically-motivated curiosity gave rise to caudate activities (Kang et al., 2009). In addition, Murayama and his colleagues (2010) found neural evidence supporting the undermining effects of extrinsic rewards on intrinsic motivation during the performance of an interesting stopwatch task by observing the neural decreases of caudate activities. That is, the presence of intrinsic motivation gave

87 76 rise to an increase in caudate activities, while the undermining of intrinsic motivation gave rise to a decrease in caudate activities. Based on the findings from the previous neuroscientific studies on intrinsic motivation, it can hypothesized that what will be unique about the actual experience of intrinsic motivation (versus its imagined experience) will be increased activations in the ventral striatum. Therefore, combining the findings from Studies 1 and 2 with the pioneering neuroscientific studies of intrinsic motivation (i.e., the three experiments outlined above) led to the following hypothesis: When people engage in interesting versions of a task, they will show increased neural activities in both their anterior insular cortex as well as in the ventral striatum, compared to when people engage themselves in non-interesting versions of the same tasks. Method Participants 16 Korean undergraduates (8 females and 8 males; mean age: 23.4 ± 3.01), who were recruited from a large university in Korea, participated in the fmri study. They were right-handed and had no history of neurological illness. All participants provided informed consent in accordance with the regulations of the Institutional Review Board of Korea University and received compensation for their participation. Tasks Two tasks were employed, one that was designed to generate intrinsic motivation via curiosity-inducing questions (Kang et al., 2009) and a second task that was designed to generate intrinsic motivation via optimal challenge (Reeve, 1989).

88 77 Curiosity-inducing task. Curiosity is a cognitively-based emotion that occurs whenever a person experiences a gap in his or her knowledge (Loewenstein, 1994), and it is fundamental to intrinsic motivation as it is the basis for exploratory, manipulatory, and experimental behaviors (Deci & Ryan, 1985). When environmental events unfold in expected ways, people do not experience gaps in their knowledge and do not experience curiosity or the intrinsic motivation to explore. When environmental events unfold in unexpected ways, however, people do experience gaps in their knowledge and do experience curiosity and the intrinsic motivation to explore in such a way that exploration provides the information people need to resolve the knowledge gap, which yields learning. From this conceptualization of what makes a task interesting or not, two versions of a question-and-answer task were developed, one that induced curiosity and one that did not (see Table 6). The task goal was to find the correct answer for the target question. The curious questions were about the unknown and new knowledge in various domains (e.g., What animal can shed up to 30,000 teeth in its lifetime?). These curious questions were adopted from Kang and her colleagues study (2009) and translated into Korean, and from Korean trivia quiz books and websites. In contrast, the non-curious questions were about common knowledge (e.g., What country is the world s most populous?). These non-curious questions were adopted from general-knowledge quiz books and elementary school textbooks. The sentence structure and word length were matched between the curious and non-curious questions. In addition, to control for novel word effects, the correct answer words for both curious and non-curious questions were well-known words (e.g., shark for the above curious question; China for the above non-curious

89 78 question). Twenty-one pairs of curious versus non-curious questions were selected for inclusion in the experiment, and these 21 pairs of questions were selected based on a computerized pilot test of 25 possible pairs of curious versus non-curious questions. Table 6 Examples of curious and non-curious questions. Curious questions Answers Non-curious questions Answers What animal sheds up 30,000 teeth in its lifetime? Shark What country is the world s most populous? China What book is the most shoplifted book? The Bible What is the official currency of the United States? US dollar What instrument was invented to sound like a human singing? Violin What does the red traffic light mean? Stop What was the first animated film to win an Academy Award? Beauty and the Beast Which company manufactured iphone and ipad? Apple Which continent is divided into the most countries? Africa What is the capital city of France? Paris What country has the longest coastline Canada How many days are in the month of March? 31 days Which company first manufactured CDs? Philips What is the addictive substance in cigarettes? Nicotine Challenge-inducing task. A challenge is something that requires stretching one s abilities and involves trying something new. The intrinsic needs for autonomy and competence underlie intrinsically-motivated behavior to seek out and attempt to master optimal challenges, to use one s creativity and resourcefulness, and to engage oneself in challenges that are neither too easy nor too difficult. Like curiosity, optimal challenge is

90 79 fundamental to intrinsic motivation as it is the basis for exploratory, manipulatory, and experimental behaviors (Deci & Ryan, 1985). As people explore their surroundings and as people seek out opportunities to stretch and extend their capacities, they encounter challenges within environmental activities that vary in their difficulty. When tests of skill are too easy, people do not experience psychological challenge and they do not experience intrinsic motivation to master that challenge. When tests of skill are appropriately difficult, people do experience psychological challenge and they do experience intrinsic motivation to undertake and seek to master that challenge. The environmental task selected to contrast challenge versus non-challenge in the present study was an anagram task. The challenging and non-challenging (i.e., easy) English anagrams (see Table 7) were developed based on the experimental task of a previous study (Reeve, 1989). The task goal was to unscramble a 5-letter jumbled meaningless word into the correct word. To differentiate the challenging levels, 5-letter jumbled words were scrambled according to the complex and difficult-to-predict rules (e.g., OHTMN transposed from MONTH ) while 5-letter jumbled words for the non-challenging anagrams were scrambled according to the simple and repeated rules (e.g., CLCOK transposed from CLOCK ). To control out novel word effects, the correct answers for both challenging and non-challenging anagrams were commonly used words. To make the challenging anagrams challenging but solvable (i.e., not extremely difficult) within the time limit of the current fmri experiment (7 seconds), a hint for each of the challenging and nonchallenging anagrams was presented in the middle of the problem presentation. Each of the hints for anagrams informed the participant as to what the first two letters of the

91 80 correct answer word were (e.g., MO for the above challenging anagram; CL for the above non-challenging anagram). Twenty-one pairs of challenging versus non-challenging anagrams were selected for inclusion in the experiment, and these 21 pairs of anagrams were selected based on a computerized pilot test of 25 possible pairs of challenging versus non-challenging anagrams. Table 7 Examples of challenging and non-challenging anagrams. Challenging anagrams Answers Non-challenging anagrams Answers OHTMN MONTH CLCOK CLOCK CRPEI PRICE WHTIE WHITE NRITA TRAIN HAYPP HAPPY HUOCG COUGH IFRST FIRST SPEUA PAUSE THIKN THINK CIOTN TONIC SEMLL SMELL CISNE SINCE OEACN OCEAN Procedure Using the event-related fmri design, three separate 8-minute runs were employed. In each run, 28 trials, 7 trials per each of four experimental conditions (i.e., curious/noncurious question conditions and challenging/non-challenging anagram conditions), were randomly presented. In each trial (see Figure 12), a question or an anagram was

92 81 presented for seven seconds. In each of the challenging/non-challenging anagram trials, after three seconds passed, a hint informing the first two letters of the correct answer was given together with the problem for four seconds while no hint was given on any of the curious/non-curious question trials. During this 7-second question/anagram presentation, participants were asked to think about the answer. Then, the correct answer was presented for three seconds. After a one-second fixation cross was presented, participants were asked to rate how much this question or anagram was interesting on a 1-4 scale (1 = not at all; 2 = a little; 3 = some; 4= a great deal) for two seconds. To indicate their judgments, participants were asked to press one of the four buttons by using their right hand. Following this rating, the next trial began after another fixation cross was presented at the inter-trial interval (ITI) for an average of four seconds (2000ms-6000ms). Before the fmri scanning, participants received the task instruction. Once the participant was situated in the fmri setting, he or she completed four practice trials. During the fmri scan, functional images were acquired while participants performed the experimental task, and then anatomic images were acquired. At the end of the experiment, participants were debriefed about the experiment and received their compensation for participation.

93 82 Note. CU: curious question condition; NCU: non-curious question condition; CH: challenging anagram condition; NCH: non-challenging anagram condition. Figure 12 Task and experimental design. fmri Data Acquisition The method of the fmri data acquisition was the same to that of study 2. fmri Data Analysis The brain images were analyzed using AFNI (Cox, 1996; The first three images of each run were discarded to allow hemodynamics and MRI signals to reach a steady state. The procedure of the preprocessing was the same to that of study 2. In individual analyses, the preprocessed time-series data were analyzed by a general linear model (GLM) using the four regressors of conditions for curious questions,

94 83 non-curious questions (as a corresponding control for curious questions), challenging anagrams, and non-challenging anagrams (as a corresponding control for challenging anagrams), and the six regressors for motion parameters as covariates which were convoluted with hemodynamic response functions (HRF). For group analyses, each individual s statistical data were transformed to Talairach space (Talairach & Tournoux, 1988). First, the high-resolution structural images of each participant were transformed to the standardized structural images. Then, the functional images of each participant were transformed to the standardized highresolution structural images of each participant. At that time, the functional images were resampled to 2 X 2 X 2 mm 3 voxels. In the group analyses, subtraction analyses were first conducted to compare the neural differences between the curious versus non-curious questions and between the challenging versus non-challenging anagrams. For correcting multiple comparison inferences in these whole-brain analyses, the cluster-wise threshold (corrected p <.049) was employed based on Monte-Carlo simulations (Forman et al., 1995), which was determined by both voxel-wise threshold (p <.001) and cluster size (n = 26, a minimum volume of 208 mm 3 ). A conjunction analysis was then performed to examine the common neural activities of the curious questions and the challenging anagrams compared to those of the non-curious questions and the non-challenging anagrams. For the conjunction analysis, statistical maps for the two contrasts (i.e., curious questions versus non-curious questions, challenging anagrams versus non-challenging anagrams) were created. These maps contained voxels which were significantly activated above the threshold (t > 4.05, which

95 84 is p < 0.001, for intensity threshold; n = 26, a minimum volume of 208 mm 3 for cluster size). With these two maps, a conjunction map was constructed. The significant activations for the subtraction analyses and the conjunction analysis were reported as Talairach coordinates after the MNI coordinates converted to the Talairach space by using a mni2tal algorithm (Lacadie et al., 2008). After activated brain regions in the above analyses were set as regions of interests (ROIs), the BOLD signal changes of each condition in these ROIs were examined in order to compare the neural activation magnitudes of the ROIs across conditions. For these analyses, SPSS 17.0 was used. Results Behavioral Results Participants mean rating of how interesting the questions or anagams were was compared between the two conditions of each task. Results showed that both the curious questions (F (1,15) = 69.64, p <.05, d = 2.15 ; curious questions, non-curious questions, Ms 3.15 > 1.78; see Figure 13. A) and the challenging anagrams (F (1,15) = 20.42, p <.05, d = 1.22 ; challenging anagrams, non-challenging anagrams, Ms 2.39 > 1.99; see Figure 13. B) were perceived as more interesting than were their corresponding comparison conditions. These findings are important because they show that participants experienced the curiosity-inducing questions and the challenging anagrams as significantly more interesting (i.e., as significantly more intrinsically motivating) than the non-curiosityinducing questions and the non-challenging anagrams.

96 85 Note. CU: curious question condition; NCU: non-curious question condition; CH: challenging anagram condition; NCH: non-challenging anagram condition. * p <.05. Figure 13 Participants mean interest rating. fmri Results Results of the conjunction analysis showed that the following brain regions were more activated in both the curious question and the challenging anagram conditions than in the non-curious question and the non-challenging anagram conditions: the left inferior frontal gyrus, the left anterior cingulate cortex, the bilateral occipital lobe, the right cerebellum, and the left anterior insular cortex (corrected p < 0.049; Table 8; Fig. 14. A).

97 86 Table 8 Results of the conjunction analysis of the contrast between curious versus non-curious questions and the contrast between challenging versus nonchallenging anagrams. Region BA Volume Side Talairach Coordinates x y z Inferior frontal gyrus L Anterior cingulate cortex L Occipital lobe L R Cerebellum 528 R Anterior insular cortex L Note. The cluster-wise threshold (correct p <.049) is determined by voxel-wise threshold (p <.001), the connectivity radius (2.0 mm), the minimum volume (26 contiguous voxels, 208 mm 3 ), and the FWHM (5 mm). BOLD signal changes of the brain regions activated in the above conjunction analysis (except the occipital lobe because it was not the hypothesized brain region) were compared across conditions. Results confirmed that the neural patterns of the conjunction analysis results (i.e., interesting condition > non-interesting condition) were correct in all hypothesized brain regions, the anterior insular cortex (t (15) = 5.58, p <.05, d = 1.43 for curious versus non-curious questions; t (15) = 6.70, p <.05, d = 1.77 for challenging versus non-challenging anagrams), the cerebellum (t (15) = 7.58, p <.05, d = 1.89 for curious versus non-curious questions; t (15) = 5.95, p <.05, d = 2.04 for challenging versus non-challenging anagrams), the inferior frontal gyrus (t (15) = 8.31, p <.05, d = 2.12 for curious versus non-curious questions; t (15) = 6.94, p <.05, d = 1.78 for challenging versus non-challenging anagrams), and the anterior cingulate cortex (t (15) = 5.71, p <.05, d = 1.48 for curious versus non-curious questions; t (15) = 6.64, p <.05, d = 1.78 for challenging versus non-challenging anagrams) (see Figure 14. B).

98 87 Note. CU: curious question condition; NCU: non-curious question condition; CH: challenging anagram condition; NCH: non-challenging anagram condition. Figure 14 A. There were commonly greater brain activations of the anterior cingulate cortex (a), the inferior frontal gyrus (b), the anterior insular cortex (c), and the cerebellum (d) in the CU and CH conditions than in the NCU and NCH conditions. B. The BOLD signal changes of these brain regions across conditions are presented.

99 88 In addition to looking at the neural differences between the interesting version of the tasks (curious, challenge) versus the non-interesting version of the task (non-curious, non-challenging), the neural differences were examined that were specific to both tasks. Results of the individual subtraction analysis that contrasted only the curious questions versus the non-curious questions appear in Table 9, while the results of the subtraction analysis that contrasted only the challenging anagrams versus the non-challenging anagrams appear in Table 10. As shown in Table 9, the curious question condition showed more neural activities of the frontal regions and the ventral striatum as well as the brain regions activated in the above conjunction analysis while the non-curious question condition showed more neural activities of the parietal and temporal regions. As shown in Table 10, the challenging anagram condition showed more neural activities of the frontal regions as well as the brain regions activated in the above conjunction analysis while there were more neural activities of the temporal regions in the non-challenging anagram problem condition.

100 89 Table 9 Results of the contrast between curious versus non-curious questions. Region BA Volume Side Talairach Coordinates x y z Maximum t value Curious Non-curious Frontal Inferior frontal gyrus L L Anterior insular cortex L Superior frontal gyrus / Anterior cingulate cortex 6, L Middle frontal gyrus L L Temporal Middle temporal gyrus L Occipital Occipital lobe L R Limbic Ventral striatum 408 L R Cerebellum 1744 R R Non-curious Curious Parietal Posterior cingulate cortex R Precentral gyrus L Postcentral gyrus R Precuneus R R Precuneus R Posterior insular cortex L Inferior parietal lobe L R Temporal Middle temporal gyrus R Inferior temporal gyrus L Occipital Occipital lobe R Note. The cluster-wise threshold (correct p <.049) is determined by voxel-wise threshold (p <.001), the connectivity radius (2.0 mm), the minimum volume (26 contiguous voxels, 208 mm 3 ), and the FWHM (5 mm).

101 90 Table 10 Results of the contrast between challenging versus non-challenging anagrams. Region BA Volume Side Talairach Coordinates x y z Maximum t value Challenging Non-challenging Frontal Inferior frontal gyrus / anterior insular cortex 45, L Anterior cingulate cortex R Inferior frontal gyrus R Anterior insular cortex R Parietal Inferior parietal lobe L Supramarginal gyrus R Precentral gyrus L R Occipital Occipital lobe L R R L Limbic Thalamus 320 L Cerebellum 984 R R L Non-challenging - Challenging Frontal Medial frontal gyrus L Parietal Posterior cingulate cortex L Precuneus R Temporal Middle temporal gyrus L Superior temporal gyrus R R R Note. The cluster-wise threshold (correct p <.049) is determined by voxel-wise threshold (p <.001), the connectivity radius (2.0 mm), the minimum volume (26 contiguous voxels, 208 mm 3 ), and the FWHM (5 mm).

102 91 In particular, ventral striatum activities, which are well-known for reward processing, were observed only in the contrast between the curious versus non-curious question conditions, but not in the contrast between the challenging versus nonchallenging anagram conditions (see Figure 15. A). To confirm the ventral striatum activation differences between the two comparisons, BOLD signal changes of the ventral striatum were compared across conditions and between the two conditions of each task (see Figure 15. B). Results showed that the following neural pattern of the ventral striatum activities was observed: curious question condition > challenging anagram condition > non-curious question condition = non-challenging anagram condition (F (3,13) = 12.42, p <.05). However, the results also showed that the ventral striatum was relatively more activated in the challenging anagrams condition than in the nonchallenging anagrams condition (t (15) = 2.54, p <.05, d = 0.65) just as it was in the curious question condition than in the non-curious question condition (t (15) = 6.45, p <.05, d = 1.66).

103 92 Note. CU: curious question condition; NCU: non-curious question condition; CH: challenging anagram condition; NCH: non-challenging anagram condition. Figure 15 A. There were significantly greater brain activations of the ventral striatum in the CU condition than in the NCU condition. B. The BOLD signal changes of the ventral striatum across conditions are presented.