THE ROLE OF SELF-CONSCIOUS EMOTIONS ON INFERENCE GENERATION A Thesis by Jill D. Heisler Bachelor of Arts, Wichita State University, 2010 Submitted to the Department of Counseling, Educational Leadership, Educational and School Psychology and the faculty of the Graduate School of Wichita State University in partial fulfillment of the requirements for the degree of Master of Education May 2014 i © Copyright 2014 by Jill D. Heisler All Rights Reserved ii THE ROLE OF SELF-CONSCIOUS EMOTIONS ON INFERENCE GENERATION The following faculty members have examined the final copy of this thesis for form and content, and recommend that it be accepted in partial fulfillment of the requirement for the degree of Master of Education with a major in Educational Psychology. _______________________________________ Catherine M. Bohn-Gettler, Committee Chair _______________________________________ Kim McDowell, Committee Member _______________________________________ Susan Unruh, Committee Member iii DEDICATION To my parents, who inspired my love of reading. iv ACKNOWLEDGMENTS I would like to thank my advisor, Dr. Catherine Bohn-Gettler, for her guidance and advice throughout my education. I would also like to thank Doris Burgert for her support and encouragement. I appreciate the following research assistants for their help in conducting this study: Brady Marzolf and Jeanette De La Torre. Finally, I would like to express gratitude toward my committee members, Dr. Kim McDowell and Dr. Susan Unruh, for providing me with feedback and taking the time to be members of my committee. v ABSTRACT Emotion may play a critical role in a reader’s ability to comprehend text. Past research has shown that happy and sad emotions influence processing, which can impact comprehension. More specific emotions, such as self-conscious emotions, require a reader to allocate more cognitive resources to process those emotions. This study examines the effects of pride, shame, guilt, and neutral emotions on the reader’s ability to generate planning knowledge and physical knowledge inferences. After an autobiographical memory task, participants read texts that require a planning knowledge or physical knowledge inference to be generated. Response times and accuracy rates for the inference generation task were examined. Though there was not a statistically significant effect of emotion found when examining accuracy rates for knowledge validating questions, the p-value was less than .10, and demonstrated that pride may facilitate general knowledge activation during reading. vi TABLE OF CONTENTS Chapter Page 1. INTRODUCTION 1 2. LITERATURE REVIEW 3 2.1 2.2 2.3 2.5 3 4 8 18 3. METHOD 21 3.1 3.2 21 21 21 22 22 24 24 3.3 4. 5. Text Comprehension Inference Generation Mood and Emotion Current Study and Hypothesis Participants Materials 3.2.1 State Shame and Guilt Scale (SSGS) 3.2.2 Emotion Induction 3.2.3 Inference Task 3.2.4 Demographic Information Procedure RESULTS 27 4.1 4.2 27 30 Verifying the Validity of the Emotion Induction Procedure Inference Task s DISCUSSION 5.1 Processes Involved in Inference Generation 5.2 Potential Mediating Variables 5.3 Future Directions 5.4 Implications 40 42 43 44 46 REFERENCES 47 APPENDIXES 53 A. B. C. D. E. SSGS Emotion Induction Scripts Inference Generation Task – Planning Knowledge Stimuli Inference Generation Task – Physical Knowledge Stimuli Demographic Survey vii 54 56 58 62 66 CHAPTER 1 INTRODUCTION Reading comprehension is vital for the success of children and adults, whether they are students in academic settings or simply enjoying a novel. During the education process, students depend on their ability to read and comprehend text for learning and completing assignments in every academic subject, such as reading texts to write a research paper, or following instructions for a science experiment. Adults likewise rely on the ability to comprehend text in their daily lives, such as reading a memo at work, following directions to bake a loaf of bread, or filling out a tax return. Comprehension involves the simultaneous use of multiple processes to create a coherent mental representation of the text. Although a variety of theories have been proposed to define and explain comprehension, a dominant theory, the tri-partite model, specifies comprehension as a process that involves integrating the text content with one’s knowledge (Kintsch & van Dijk, 1978). This integration can be referred to as inference generation, and it is a vital part of comprehension because readers must construct inferences to encode the meaning of text (Graesser, Singer, & Trabasso, 1994). The current study examines the inferential processes that can lead to a better understanding of how individuals comprehend text. Various factors can influence the way a person comprehends text. One such factor includes reader variables, such as general attention, attention-allocation skills, inferential skills, background knowledge, and basic reading skills (Thurlow & van den Broek, 1997; van den Broek & Kremer, 1999). One variable particularly important to the current study and that can influence comprehension is the reader’s moods and emotions. Previously, researchers have examined mood and its influence on text comprehension (Bohn-Gettler & Rapp, 2011; Egidi & Gerrig, 2009). For example, Bohn-Gettler and Rapp (2011) found that readers in a positive mood 1 generated more text-based inferences and fewer non-coherence processes compared to readers in a neutral mood. Therefore, a positive mood could be beneficial for certain processes involved in text comprehension. Some researchers examine more specific emotions that can be evoked in academic settings (Pekrun, 2006). These academic emotions relate immediately to academic achievement and the outcomes of academic tasks. Some of these academic emotions are also classified as self-conscious emotions, which are emotions related to a person’s self-concept or mental representation one has regarding oneself (Tracy & Robins, 2007). These emotions include pride, shame, and guilt. Self-conscious emotions require self-evaluation, and therefore require an individual to utilize his or her cognitive resources to generate and process the emotion. The processes involved in comprehending text, generating inferences, and managing moods and emotions can all place additional loads on an individual’s limited cognitive resources (Ellis, Thomas, & Rodriguez, 1984). Thus, too much strain on those resources in one direction can cause another factor to receive less attention. According to Forgas’ (1995) affect infusion model, affect, mood, and emotions can play a part in some levels of cognition when high cognitive effort and constructive thinking are required. More complex inferences require high cognitive effort and constructive thinking. Thus, moods and emotions should influence inference generation during reading comprehension. Understanding how emotion can affect inferential processing is important to understanding reading comprehension and its constraints. This study combined emotion and inference generation to better understand reading comprehension. In the current study, participants were induced to experience the self-conscious, academic emotions of pride, shame, or guilt. After inducing these emotions, participants read texts that require inferences to be generated, and researchers examined the relationship between emotion and inference generation. 2 CHAPTER 2 LITERATURE REVIEW Text Comprehension Tri-partite model of reading comprehension. Kintsch and van Dijk (1978) proposed a tri-partite model to define and explain the processes underlying reading comprehension. In this model, there are multiple levels in which information is encoded from the text into a mental representation. A mental representation is a cognitive account of the ideas and information presented in the text. A reader must have a highly coherent mental representation of the text in order to have a high level of reading comprehension. Coherence represents the degree to which the text is correctly encoded into a complete mental representation within the mind of a reader (van den Broek, Lorch, Linderholm, & Gustafson, 2001). The first level of the tri-partite model is the surface structure, the next is the propositional level, and the last level is the situation model. The surface structure represents the most superficial level of understanding. At this stage, the reader encodes words without establishing meaning (Kintsch & van Dijk, 1978). Consider this example: “The child broke the vase. He hid the pieces from his mother.” The reader would encode only the words and the sentence structure, but not necessarily the meaning of the words. The propositional level represents a deeper understanding of the textual information encoded into the mental representation. At this level, the reader encodes the general meaning of the sentence. The reader paraphrases the gist meanings of text but does not make connections to background knowledge or within the text itself (Kintsch & van Dijk, 1978). Applied to the example of the child, the reader may encode that the child broke the object and concealed the fragments. However, the reader would not move beyond this degree of understanding. 3 The situation model level of comprehension represents the deepest level of processing. At this level, the reader uses background knowledge and context to more fully understand the meaning of the text. The reader will also generate inferences. Inferences fill in missing information in the text, such as adding to a text to connect new content to already processed information (Kintsch & van Dijk, 1978). For instance, referring back to the example sentence about the child breaking the vase, the reader might use context clues (e.g., the pieces were hid from the mother) and background knowledge (e.g., that mothers often get upset about broken objects) to infer that the child hid the broken pieces in an attempt to avoid his mother’s disapproval. Inference Generation Inference generation is an essential part of the text comprehension process. Graesser et al.’s (1994) constructionist theory posits that individuals construct inferences by activating background knowledge structures held in the long-term memory, and that information is used when encoding the meaning of the text. The background knowledge used during inference generation can range from specific to generic knowledge that would apply to the text. For example, if an individual was reading a narrative passage set in a park, the reader might incorporate generic knowledge about what parks contain. Based on this knowledge, he might infer that trees, grassy areas, and play structures form the scene in the passage. However, sometimes more specific knowledge is needed. For instance, imagine a science student is reading the process for a laboratory experiment. When reading the instructions for an experiment, he might rely on his more specific background knowledge of sodium bicarbonate to infer the reaction of the chemical when mixed with vinegar. 4 In inference generation research, there exists a debate about whether inferences are generated automatically or strategically. Automatic processes refer to cognitive procedures that are not in the reader’s conscious control, do not require many attentional resources, are effortless, interfere minimally with other cognitive tasks, and are rapid. In contrast, strategic processes refer to intentional processes used by a reader to better understand the text (Magliano, Trabasso, & Graesser, 1999). Automatic processes can reduce the burden on the individual’s working memory, which can only hold a limited amount of information. Therefore, cognitive resources are limited and automated processes allow for those cognitive resources to be used for other tasks (Thurlow & van den Broek, 1997). As an example, novice readers expend most of their limited cognitive resources on identifying words in a text because this is not yet an automatic process, which leaves fewer remaining resources for inference generation. In contrast, skilled readers are more automatic with word identification skills and creating meaning in combinations of word units. Therefore, a skilled reader’s cognitive resources can be allocated toward the more complex processes required to comprehend text. One type of inference that can be generated automatically or strategically is a causal inference. Causal inferences link a cause and effect when it is not explicitly stated in the text, and are often required for maintaining coherence. Trabasso, van den Broek, and Suh (1989) suggest that comprehension is a search for meaning, and that search uses general knowledge to explain causal relationships. These causal inferences are important to coherence because they help establish causal connections to link events in texts. During comprehension, a reader uses causal connections to create a causal chain, which serves as a network that connects important events throughout a narrative. The reader utilizes the causal chain to understand relationships between parts of the text as well as to generate causal inferences (Tapiero, van den Broek, & Quintana, 5 2002; Thurlow & van den Broek, 1997; Trabasso & van den Broek, 1985). For instance, when reading, “the door slammed, and the man woke up,” the reader would infer that the sound of the door slamming caused the man to awaken. Although research has shown that this type of inference is often generated quickly and with little effort, there exists variation among readers. Less skilled readers do not show strong patterns of consistently making this type of inference. In contrast, highly skilled readers are more likely to consistently generate causal inferences, suggesting that causal inferences may become more automatic as reading skills increase (Long, Oppy, & Seely, 1994; Thurlow & van den Broek, 1997). Causal inferences can include backward, forward, and elaborative inferences. To generate backward inferences, readers will attempt to better comprehend a sentence by referring to a previous statement in the text or to their pre-existing background knowledge stored in long-term memory (Graesser et al., 1994). For example, when reading, “Sally received an invitation to a party. Sally went to the mall to buy a new dress,” the reader could use the information that Sally was invited to a party to explain why Sally went to the mall to purchase a dress. Ultimately, the reader could infer that Sally bought the dress to wear to the party. Forward inferences utilize knowledge structures already in the long-term memory (i.e., prior knowledge and experiences) or information provided in the text to predict what will happen next (Graesser et al., 1994). When considering the example of Sally and the party, a reader could infer a number of different possibilities. The reader may infer that Sally bought a new dress, that Sally could not find a suitable dress, or that Sally did in fact attend the party after finding the dress. Although none of these options were explicitly stated in the text, the reader could use his or her own experiences of preparing for parties or information provided about Sally to predict what will happen next in the passage. 6 Elaborative inferences are inferences that add additional information to the text from a reader’s background knowledge and can be either backward or forward inferences (McKoon & Ratcliff, 1989). One type of elaborative inference is a consistent inference, which occurs when a reader elaborates on the text to make it logically consistent (Thurlow & van den Broek, 1997). If a person reads, “the child ran by the lamp and it broke,” the reader may infer that the child bumped into the lamp, thus causing its ruin. Or, the reader could infer that the child deliberately hit the lamp because he wanted to watch it break. Elaborative inferences are not always required for creating a coherent representation of the text (Thurlow & van den Broek, 1997). In this instance, determining the child’s intent is not necessary to understanding the sentence. Elaborative inferences include several knowledge areas, including planning and physical knowledge (Shears & Chiarello, 2004; Shears, Miller, Ball, Hawkins, Griggs, & Varner, 2007). Physical inferences are generated based on physical actions that are occurring within the text. These inferences are usually simple to generate and can be either predictive or backward in nature. For example, in the text, “The earthquake shook the house. The house collapsed," the reader would infer that the shaking from the earthquake caused the house to fall. Planning, or goal-directed, inferences are defined as a series of purposeful actions that support the achievement of a goal. Planning inferences can include knowledge regarding efficiency, temporal ordering, and rational sequencing. For example, the text, “The lazy student did not study for the test. He copied the answers from his friend,” requires the reader to generate more inferences compared to the physical inference. The reader must infer that the student had a goal of passing the exam, which caused him to copy answers to achieve that goal. In general, knowledge-based inferences are drawn from a person’s general knowledge (Graesser et al., 1994). However, there could be a difference in the way that knowledge is used 7 during the inference generation process or the way in which various areas of knowledge can form connections in the text. Based on the link between comprehension and the pursuit of causal relations suggested by Trabasso et al. (1989), there is a hierarchy of knowledge based on logical necessity, meaning that knowledge dealing with physical situations could be more readily available than knowledge dealing with planning situations. This may be due to the physical situation being more urgent and possibly linked to danger (depending on the context). Therefore, planning inferences could be categorized differently than physical inferences. Research suggests that planning inferences require more cognitive resources compared to physical inferences, because planning inferences require the reader to make the link between the text, goals, and goal achievement rather than simply linking physical cause and effect (Shears & Chiarello, 2004; Shears et al., 2007). Although multiple inferences can be generated during the comprehension process, various factors can influence the type and frequency of inference generation. van den Broek and Kremer (1999) propose that three factors influence comprehension: text variables, the instructional context, and reader variables. Each factor can individually influence comprehension, and the variables may interact with one another. The current study considers two of the three facets. First, it considers the text variable of what type of inference is suggested in the text. Second, it considers the instructional context, namely the emotions reader experience as a function of an induction procedure. Mood and Emotion When examining individuals, affect can play a key role in cognitive processing in a variety of ways. For example, in educational settings, researchers have examined anxiety and its influences on learning and test performance (Pekrun, 2006). However, moods and emotions 8 should first be considered in the context of their broader theoretical frameworks. These theoretical models focus on affect (a general term used for both mood and emotion), mood (an enduring affective state with low intensity), and emotion (a higher intensity affective state with a short duration,) and their roles in cognitive processes (Forgas, 1995). Emotions can be categorized in terms of valence (the pleasant or unpleasant nature of the emotion) and activation (the level of arousal the emotion causes). For example, a pleasant activating emotion could be enjoyment, and a pleasant deactivating emotion could be relief. Furthermore, an unpleasant activating emotion could be anxiety, and an unpleasant deactivating emotion could be hopelessness (Pekrun, Goetz, Frenzel, Barchfeld, & Perry, 2011). In the context of these broader theoretical frameworks, specific emotions can be considered in terms of how they influence cognitive processing. Resource allocation models. One conceptualization of how affect can influence cognitive processing comes from resource allocation models. Such models suggest that any task requires a certain degree of cognitive effort, and the amount of cognitive effort allocated to that task affects memory. Individuals have limited cognitive resources, and emotion can represent a “drain” on these cognitive resources because attention must be allocated toward processing the emotion. Therefore, emotion places an additional load on the already limited cognitive resources that can be available for completing a task (Ellis, Thomas, & Rodriquez, 1984). This framework is the baseline for the resource allocation model, which describes the relationship between emotion and cognitive resources (Ellis & Ashbrook, 1988). Ellis, Moore, Varner, Ottaway, and Becker (1997a) asked participants to write down thoughts as they learned material. Those participants in a negative mood wrote more irrelevant thoughts and recalled less material than those in the control group. This study demonstrated that individuals in negative 9 moods expended additional attentional resources when dealing with their mood, leaving fewer of those resources to help complete the task at hand (Ellis & Ashbrook, 1988; Ellis et al., 1997a). Kihlstrom (1989) added to this by suggesting that individuals in a depressed mood think about their mood more than those not in a depressed mood. Therefore, thinking about one’s mood further consumes those attentional resources needed to complete other cognitive tasks. Ellis, Ottaway, Varner, Becker, and Moore (1997b) applied the resource allocation model to readers’ abilities to identify contradictions in texts. When instructed to find contradictions, participants with a sad mood identified only one third as many contradictions as those in the control group and made more false identifications. In addition, readers in sad moods reported experiencing more irrelevant thoughts that were disruptive to the comprehension process. Negative moods also caused readers to be less accurate when making evaluations regarding how well they comprehended the text (Ellis et al., 1997b; Stringer & Heath, 2006). Such work demonstrates that affect can influence certain comprehension processes. The resource allocation model offers an explanation of how emotion can impact cognitive processes in general. However, this model does not address the circumstances under which affect influences cognitive thinking processes, such as varying difficulty levels of tasks. Furthermore, it does not differentiate between the roles that positive and negative affect can have on processing, but rather assumes that emotion in general consumes cognitive processing. The affect infusion model fills this gap by examining under what conditions mood can have an impact on cognitive processes, as well as the differential effects that could result from positive versus negative moods. Affect infusion model. Forgas (1995) proposed a four-leveled approach to processing in the affect infusion model (AIM). Affect infusion occurs when moods and emotions influence 10 cognitive thinking processes. Thus, the four levels refer to levels of processing that either allow or do not allow for affect to influence processing (affect infusion). The first two levels, direct access and motivated processing, have low affect infusion, meaning that affect is unlikely to play a role in processing. When utilizing direct access, individuals will activate already-stored information about familiar objects (Forgas, 1995). For instance, if a person has learned that dogs have fur, that person will continue to access this information such that he will assume that any dog he sees will have fur. He will access this information without a great degree of cognitive effort (if any), and mood will not affect the access of this simple information because no constructive elaboration is involved. The second level, the motivated processing level, also involves directly accessing information from long-term memory. However, accessing this information is more goal-directed in that the information needed is more specific. For example, an individual will use this method when he has a clear information goal, meaning that a person will search for very specific information in long-term memory. This processing method requires a high degree of cognitive effort and predetermined thinking (Forgas, 1995). For example, a contestant on a game show would use this processing level when searching for the answer to a difficult trivia question. Although motivated processing requires a greater degree of cognitive effort than direct access, it does not allow for affect infusion because, similar to direct access, constructive elaboration is not necessary. The next two levels, heuristic and substantive processing, allow for more affect infusion because they require constructive elaboration. Constructive elaboration is a cognitive process that involves open information search strategies within an individual’s mind and elaboration on existing cognitive representations. In heuristic processing, a person will process information 11 based on general rules already established through experience (Forgas, 1995). For example, if a woman knows a few people from neighborhood X and those people engage in gossiping, she may use heuristic processing to infer that all people from neighborhood X are gossips. If this woman does not like gossips, she might associate her negative feelings with all people from neighborhood X. Another example of heuristic processing is treating affect as information, which is when a person will treat his or her emotions as actual objective information about a particular object, person, or event. Affect-as-information is used as a heuristic short cut to making judgments, rather than objectively going through problem-solving strategies to generate judgments (Forgas 1995). Heuristic processing requires lower degrees of cognitive effort than some other forms of processing, but it is a form of constructive processing. The final level of the affect infusion model, substantive processing, uses both high cognitive effort and a high degree of constructive thinking. Thus, substantive processing allows for affect infusion. Substantive processing occurs when a person engages in comprehensive and analytic processing of a particular subject (Forgas, 1995). For instance, an employer would use this type of processing when evaluating a candidate for an open job position. He may use affect toward the candidate when they meet in an interview, as well as information of the applicant’s job history and skills, to engage in problem solving that would lead to an informed decision about whether or not to hire the candidate. In this scenario, the employer would exert a high degree of cognitive effort and would also be motivated to make accurate interpretations in order to correctly evaluate the applicant. This stage is the most demanding and is often used when the subject is either complex or unfamiliar. One would use this type of processing if he lacked a specific answer, had enough cognitive capacity to engage in such processing, and required 12 accuracy in his judgment. Furthermore, critical thinking and creativity often require substantive processing. Therefore, the affect infusion model hypothesizes that there are certain “types” of cognitive processing that allow for moods and emotions to influence processing. The affect infusion model also examines how the valence of an individual’s mood can impact an individual’s focus of attention and method of processing (Forgas, 1995). Individuals in a negative mood will use a more methodical processing style, such that they attend to each piece of information contained within a task, regardless of its relevance to the task. When in a negative mood, participants are more detail-oriented and more likely to use conservation of information, meaning they encode the information but may not process it more deeply (Bless & Fiedler, 1995). In contrast, individuals in a positive mood will be more focused on the relevant aspects of a task. Thus, they can broaden their attention to make connections between important elements of a task, but can also activate relevant information from long-term memory that is related to the task but not contained within the task itself. This widened focus of attention enables individuals to utilize more creative processing due to an activation of general knowledge structures. Thus, such processing should assist in inference generation. The previous research focuses on cognition in general, but those principles can also apply to reading situations, in particular, inference generation. Bohn-Gettler and Rapp (2011) suggest that mood influences comprehension during the process of reading, and it can also influence post-reading memory. When induced to feel a positive mood, readers paraphrased and generated more text-based inferences and fewer non-coherence processes (such as associative inferences) compared to readers in a neutral mood. This aligns with the findings of Bless and Fiedler (1995). Because the readers are in a positive mood, they can broaden their focus and make connections 13 between important elements of the text and information from long-term memory. Sad-induced mood participants generated fewer text-based inferences than those in the happy-induced mood group. As the affect infusion model suggests, readers in a negative mood will use a more methodical processing style and make fewer connections between ideas. Those readers were more detail oriented, regardless of the relevance of those details to the text as a whole. These studies examine the valence of mood in broad terms of positive or negative. However, a wide range of more specific emotions can occur in realistic learning settings. It is important to consider more specific emotions because each emotion can provide new challenges to the reader. For example, excitement and relief are considered positive emotions. However, these emotions have very different influences on learners. Excitement tends to activate an individual and cause him to seek more learning tasks. Relief can cause a student to stop seeking further learning tasks (Pekrun, 2006). Studying more specific emotions can provide information needed to understand learning. Academic emotions. In the control-value theory of achievement emotions, Pekrun (2006) defines achievement emotions as those related immediately to academic achievement and the outcomes of an academic task. Those emotions may include: enjoyment, boredom, pride, shame, and frustration, among others. A student’s perception of the control he has over an academic activity, along with the perceived importance of the activity, combine to elicit specific achievement emotions. When a student feels control over a learning task, values the task and performs well, the resulting emotions are considered positively valenced activating emotions that can include enjoyment, hope, and pride. These emotions benefit academic performance and increase motivation. When a student feels control over and values a task but performs poorly, negatively valenced activating emotions, such as shame, anxiety, and anger can be elicited. 14 These emotions can either decrease motivation or increase motivation depending on the amount of control the student feels over the task and the level of importance the student places on the task (Pekrun, 2006). For example, shame can either increase or decrease resiliency based on the student behavior. When a student perceives the task to be highly important, the student will likely be motivated to improve task performance (Turner & Schallert, 2001). For example, if a student values a college course he is taking and performs poorly on the mid-term exam, he will be motivated to improve his performance to raise his grade on the final exam. Some academic emotions decrease motivation. When a student feels a lack of control and does not perceive the task to be important but performs well, the emotion will be a positively valenced deactivating emotion, such as relief. For example, a high school student is required to take a certain course but does not value the information provided in the course. He needs to pass an important exam in order to receive an acceptable letter grade in the course. He will feel relief after he completes the task with a passing grade, but will not be motivated to increase his performance since he already avoided failure. If a student lacks control, does not value the task and performs poorly, negatively valenced deactivating emotions will be elicited, such as hopelessness and boredom. This can result in decreased performance and motivation (Pekrun, 2006). For example, when a struggling student attempts to comprehend a too-difficult text but receives no additional support, he will likely not believe that he can improve his reading skills enough to correctly complete assignments. He will begin to feel hopeless and think he will not ever be able to perform well. He will not be motivated to increase his performance because he believes that it is impossible. Self-Conscious Emotions. Academic emotions are examined in students and are specific to the nature of the task. Although these emotions can fit into a variety of classifications, another 15 way to categorize some of these emotions is into self-conscious versus basic emotions. The current study will focus on self-conscious emotions because they require more extensive cognitive processing than basic emotions. This processing may have more of an impact on academic tasks. Basic emotions are those emotions that are based in biology, experienced across all cultures, associated with specific and universal facial expressions, and developed in the first nine months of infancy. These emotions are an automatic and immediate response to a stimulus (Tracy & Robins, 2007). Basic emotions include anger, fear, disgust, sadness, happiness, and surprise (Shaver, Schwartz, Kirson, & O’Conner, 1987). Basic emotions are often related to survival or reproduction goals. For example, if a man was standing in a street and a car was coming toward him, he would experience fear. This fear would most likely cause him to jump out of the way, thus enabling his continued survival. The academic emotions of pride, shame, and guilt are classified as self-conscious emotions. Self-conscious emotions relate to a person's self-concept (the mental representation one has regarding oneself) and his interpretation of his relationships to other individuals, groups of individuals, or society in general. Tracy and Robins (2007) suggest that self-conscious emotions should be distinguished from other emotions. Unlike basic emotions, self-conscious emotions are not present during infancy and develop between 18 months and 3 years of age, because they require a developing theory of mind, or how an individual understands himself (Lagattuta & Thompson, 2007). Because these emotions are more complex, they lack distinctive and universal facial expressions. Self-conscious emotions are also cognitively complex because they require self-evaluation. For example, if a person experiences shame, he must evaluate how 16 the event conflicted with his current self-representation. This is not an automatic response to the event. Tracy and Robins (2007) propose a theoretical process model of self-conscious emotions. After an event occurs, the individual enters the first appraisal stage, called survival-goal relevance. In this stage, one evaluates whether the event is vital to survival, reproduction, or identity. If the event is related to survival or reproduction, the person will elicit a basic emotion, such as fear or happiness. If the event is related to an identity goal, the individual will enter a stage in which his or her attentional focus is on the self. In the next stage, one focuses on self-representations, which can include current, ideal, or ought representations of the self (the last being representations based on expectations others have for the individual). At least one of these self-representations must be activated in order to elicit self-conscious emotions (Tracy & Robins, 2007). In the third stage, goal relevance, the individual evaluates whether the event is relevant to his identity goals. If the event supports or contradicts the individual’s self-representations, the cognitive process will move to the next stage in self-conscious emotion elicitation. In the fourth stage, individuals will compare the event with established selfrepresentations and decide whether the event pertains to their identity goals from the second stage (Tracy & Robins, 2007). If the event is goal-congruent, the emotion elicited will be positively valenced. If the event is not goal-congruent, a negatively valenced self-conscious emotion will be elicited. Once an individual determines whether the event is goal-congruent or goal-incongruent, the individual enters the fifth stage and will evaluate the cause of the event (Tracy & Robins, 2007), which can be perceived as either internal or external in nature. In conjunction with the 17 theory of the locus of control, an individual must attribute causation to an event. Locus of control refers to what an individual perceives is the cause for an event or outcome (Dweck, 1986). If the individual attributes causation to internal forces (i.e., that something about him or connected to him caused the event) and the outcome is negative, he will feel shame or guilt. If the outcome is positive, pride will be elicited. When the individual attributes causation to external causes, basic emotions are elicited. Shame, Guilt, and Pride. The current study will examine the specific self-conscious emotions of shame, pride, and guilt. These are important to study because these emotions are achievement emotions as well as self-conscious emotions. Shame and guilt are elicited when a person’s stable and global self-representations are called into question. If the individual believes the self-attribute is uncontrollable, shame will emerge. When the self-attribute is controllable, guilt transpires (Tracy and Robins, 2007). Much like shame and guilt, hubris and authentic pride are a result of stable selfrepresentations. Hubristic pride is an overall sense of pride that is not tied to a specific event but is rather a global self-representation (Tracy and Robins, 2007; Williams & DeSteno, 2008). Hubristic pride can lead to negative social consequences. Authentic pride is due to specific achievements. Authentic pride can play a beneficial adaptive role in that it motivates an individual to develop skills, and it guides decisions that relate to social goals. The current study will attempt to induce authentic pride because participants will be asked to recall specific events. Current Study and Hypothesis The current study examined the role of pride, shame, and guilt in relation to inference generation in readers. Previously, researchers examined general positive and negative affect and its influence on reading and inference generation (Bless & Fiedler, 1995; Bohn-Gettler & Rapp, 18 2011). Additionally, Pekrun (2006) has done a great deal of research on academic emotions and their influences on learning. This study examined emotions that are more complex than general positive and negative affect in connection with inference generation, which is important for a greater understanding of how emotions naturally elicited in academic settings can play a role in comprehension processes. The current study extends previous research by combining many of these elements to further understand emotions and inference generation. In the present study, participants reflected upon and wrote about an experience in which they felt pride, shame, or guilt, in an effort to induce that particular emotion. Then, these participants engaged in a task that measures the generation of planning versus physical inferences. This study design, along with the theoretical framework provided, lends itself to several hypotheses. Prior research demonstrates that autobiographical memory tasks are effective in manipulating emotion (Kraus, Adler, & Chen, 2012; Laird & Strout, 2007; Schwarz & Clore, 1983; Westermann, Spies, Stahl, & Hesse, 1996). Therefore, the first hypothesis was that the emotion manipulation will be effective, which will result in increases in pride for those in the pride condition, increases in shame for those in the shame condition, increases in guilt for those in the guilt condition, and no change for those in the control condition. Because certain inferences require more cognitive resources than others, and because affect infusion is more likely to occur for tasks that require substantive processing (Bohn-Gettler & Rapp, 2011; Ellis et al., 1997b; Forgas, 1995; Kihlstrom, 1989; Stringer & Heath, 2006), this study examined inferences that require differential levels of cognitive effort (physical and planning knowledge inferences). Planning knowledge inferences require more cognitive resources because the reader must make connections between the text, goals, and goal 19 achievement, whereas physical knowledge inferences simply link cause and effect (Shears & Chiarello, 2004; Shears et al., 2007). Therefore, the second hypothesis was that readers would be more likely to generate physical compared to planning knowledge inferences. This will result in faster and more accurate responses to physical knowledge inference texts than planning knowledge inference texts. Because self-conscious emotions require a good deal of cognitive processing, individuals induced to elicit these emotions may require more time to generate inferences (Pekrun, 2006). Because planning knowledge inference generation occurs at the situation model level and uses substantive processing, this requires participants to use higher levels of cognitive effort and constructive elaboration (Forgas, 1995). Affect infusion is more likely to occur for tasks that require substantive processing. Thus, the third hypothesis was that planning knowledge inferences will be more likely to allow for affect infusion than physical knowledge inferences. Therefore, there should be an interaction between emotion and inference type, such that planning knowledge inferences should be more affected by the induced emotion than physical inferences. Thus, differences in speed and accuracy as a function of emotion will be more pronounced for the planning knowledge inference texts. Although self-conscious emotions require individuals to use their cognitive resources to process the emotion, a difference may occur between the types of emotions being elicited. Pride is a positively valenced activating emotion and can benefit academic performance and increase motivation (Pekrun, 2006). This should result in causing individuals in the pride condition to generate inferences quickly and more accurately. Thus, the fourth hypothesis was that pride should facilitate inference generation, whereas guilt and shame should hamper it. This will result in faster and more accurate responses for pride. 20 CHAPTER 3 METHOD Participants A total of 91 participants in undergraduate and graduate degree programs at a Midwestern university were recruited for this study. Participants were recruited through psychology and education courses. Of the participants who responded, 20 were male and 71 were female. Caucasians comprised 92.3% of the participants, 4.4% were Hispanic, 1.1% were Asian, 1.1% were mixed race, and 1.1% did not respond. When asked about their year in school, 82.4% responded that they were undergraduate students and 17.6% responded that they were graduate students. The median age of participants was 23 years (M = 25.43). Students received course credit for their participation. The participants were all native English speakers so that language would not be a confounding factor. Materials State Shame and Guilt Scale (SSGS; Marschall, Saftner, & Tangney, 1994). The SSGS (see Appendix A) is a scale that measures state emotions of pride, shame, and guilt. The scale consists of fifteen items. Five of these items measure pride (e.g. “I feel pleased about something I have done”), five items measure shame (e.g. “I feel humiliated, disgraced”), and five items measure guilt (e.g. “I feel bad about something I have done”). Participants respond to each item by rating their emotions on a 5-point Likert scale, with one being “not feeling this way at all,” and five being “feeling this way strongly.” The instructions emphasize that participants should respond to items based on how they currently feel. The scale was developed to measure the manipulation of shame in participants. The SGSS is shown to have good reliability with 21 Cronbach’s coefficient alphas of .89 for shame, .82 for guilt, and .87 for pride (Marschall et al., 1994). Emotion induction. Participants were randomly assigned to a pride, shame, guilt, or neutral emotion condition. The neutral emotion served as the control group. Emotions were induced using an autobiographical memory task (Kraus et al., 2012; Laird & Strout, 2007; Schwarz & Clore, 1983; Westermann et al., 1996; see Appendix B). A trained researcher asked participants to remember a specific event in their past that made them feel proud, ashamed, or guilty. Those in the neutral condition were asked to remember waking up this morning. The researcher then asked participants to close their eyes and visualize the event as if they were experiencing it once more. Participants were asked to recall their feelings, specific details of what happened at the time, the people with them, the images they saw, the sounds they heard, and the thoughts they had. Finally, the researcher prompted the participants to write about the event in as much detail as possible (Kraus et al., 2012). Prior research has demonstrated that this technique is effective in manipulating emotion. This task took about 15 minutes to complete. Inference task (See Appendices C and D). Participants read 20 four-sentence texts designed to elicit planning or physical knowledge inferences (developed by Shears & Chiarello, 2004, & Shears et al., 2007; see Table 1 for sample texts). Physical knowledge inferences required the reader to link cause and effect based on physical actions in the text. However, planning inferences required the reader to generate an inference about a character’s goal and the behaviors necessary to achieve that goal. Each text had two versions: inference and control. In the inference version, the first and last sentences worked together to facilitate the generation of an inference. Sentences two and three simply contained additional information about the context. In the control version, the first 22 and last sentences did not work together to facilitate an inference, because the first and last sentences were unrelated. Table 1. Examples of Planning Knowledge and Physical Knowledge Inference Texts Type Text Physical Knowledge Inference Malcolm realized Valentine's Day was tomorrow. He was not prepared. He had to figure something out quickly. He went to the candy shop. Malcolm realized the final exam was tomorrow. He was not prepared. He had to figure something out quickly. He went to the candy shop. Jessica was in a car accident. She was really upset at first. She finally calmed down. Jessica was injured. Physical Knowledge Control Jessica was in a better mood. She was really upset at first. She finally calmed down. Jessica was injured. Planning Knowledge Inference Planning Knowledge Control Text Probe Word Inference Probe Word Knowledge Verification Question Tomorrow Girlfriend Do people give candy on special days? Crash Are people hurt in car accidents? Injured In association with each of the four-sentence texts, a text based probe word, an inference probe word, and a knowledge validating question were be presented. A text based probe word was a word that was explicitly stated in the text. In the examples regarding Malcolm and the candy shop (see Table 1), the text-based probe word was “tomorrow,” which appears in the first sentence. A correct recognition of this probe word would indicate the participant read the text. An inference based probe word was a word that was not explicitly stated in the four-sentence text, but the word could be generated from an inference made by the reader based on the inference text (but not the control text). In the text about Malcolm (see Table 1), the inference 23 based probe word was “girlfriend.” Incorrectly indicating that this word did, in fact, appear in the text would indicate that the inference was generated. More false recognitions should occur when reading the inference version compared to the control version. A knowledge validating question asked the participant a question based on the text. Responses to this question confirmed that the participant read the text and that the participant had the background knowledge necessary to make the inference. The answer to this question should be made quickly and accurately after an inference has been generated. In the Malcolm and the candy shop text, the knowledge validating question was “Do people give candy on special days?” A faster answer time, along with increased accuracy rates, would demonstrate that knowledge had been activated by the text. The inference version of the text should lead to faster answer times and more correct answers. The stimulus set included 80 texts comprised of 20 planning inference texts, 20 physical inference texts and 20 planning control texts, and 20 physical control texts. Each participant read 5 planning inference texts, 5 physical inference texts, and 10 control texts in a random order. Demographic Information (See Appendix D). A demographic questionnaire was given to participants at the end of the experiment. The questionnaire asked the participants to provide their gender, age, year in school, ethnicity, race, learning disability, and native language. Procedures After a consent process, participants completed the SSGS (Marschall et al., 1994) to establish a baseline for their emotions of pride, shame, and guilt. After completing the baseline SSGS, participants were randomly assigned to a pride, shame, guilt, or neutral emotion condition. Participants completed the autobiographical memory task (Kraus et al., 2012; Westermann et al., 1996; Laird & Strout 2007) to induce emotion. The autobiographical memory task was completed in small groups with either all members in each group assigned to the same 24 emotion condition or instructions were presented at the top of a paper. Participants were asked to recall a time that they felt proud (pride condition), ashamed (shame condition), guilty (guilt condition), or waking up that morning (neutral condition). Researchers asked the participants a series of questions to help them recall the situation. Then the participants were asked to write about the memory. After the participants wrote about the event, the participant had the option to place the journal entry into a sealed envelope or take the journal entry home. The journal entries were not read by the researchers and this information was disclosed to the participants. Following the mood induction, the participants were asked to re-take the SSGS to determine changes in their emotion as a result of the induction technique. After the second administration of the SSGS, the participants engaged in the inference generation task. Each participant completed the task on a computer using E-Prime software. The participants were given verbal and written instructions that they should read each text for comprehension. After each text, two probe words were presented, and participants decided, as quickly and accurately as possible, whether each word appeared in the text. After the probe words, participants responded to the yes/no knowledge validation question. To indicate whether the probe words appeared in the text, and to answer the yes/no knowledge validating questions, the participant were instructed to press keys labeled “yes” (the “0” key) or “no” (the “.” key) with their right index and middle fingers, respectively. The participants were asked to keep their hands placed on the “yes” and “no” keys so that measurements of speed and accuracy would not be compromised. After the instructions, the participants pressed the space bar. A message appeared on the screen that read, “Get ready to read sentences.” All stimuli on the screen were centrally presented. The participants pressed the “yes” key to begin reading the first text. Participants 25 began with five practice trials to become familiarized with the program. The sentences appeared on the screen one sentence at a time, and the participants pressed the “yes” key to indicate they were ready for the next sentence. After each text, a message that read, “Get ready to respond quickly and accurately to words” appeared. The participants were presented with the text probe word and the inference probe word in randomized order, and participants indicated whether the words appeared in the text. Following the second probe word, the participants responded to the knowledge validating question. The computer software recorded the number of correct responses, the error rates, and the response times for each probe word and question response. The screen cleared following the participant’s response to the question. The participant continued the same process for responding throughout the task. In total, the participant completed 5 practice trials and 20 trials (5 planning inference texts, 5 physical inference texts, and 10 control texts) in approximately one 20-minute session. Upon completion of the inference generation task, participants in the shame and guilt condition watched video clips that were intended to induce a happy mood for five to six minutes (Mirous, 2012). Finally, participants were asked to complete the demographics survey and were debriefed. 26 CHAPTER 4 RESULTS Verifying the Validity of the Emotion Induction Procedure To ensure the autobiographical emotion induction procedure was effective for each emotion induction condition, repeated measures ANOVAs were used. The dependent variables included the SSGS scores, and separate ANOVAs were run for each emotion measured on the scale (pride, shame, guilt). The independent variables included the emotion induction condition (pride, shame, guilt, and neutral, between subjects), and whether the SSGS was being measured at pre- or post-induction (within subjects). No outliers were removed, because the data was generally normally distributed. See Table 2 for descriptive statistics by condition. Table 2. Summary of Summed Scores Before and After Emotion Induction Emotion Condition Pride (n = 22) Shame (n = 22) Guilt (n = 24) Control (n = 23) Pride Sum Scores PrePostInduction Induction M SD M SD Shame Sum Scores PrePostInduction Induction M SD M SD Guilt Sum Scores PrePostInduction Induction M SD M SD 19.46 3.39 19.18 3.87 7.14 2.40 6.05 1.56 8.5 2.54 7.41 2.63 19.68 3.4 12.91 5.1 8.23 2.93 8.64 4.44 10.23 4.72 10.32 5.68 18.92 4.19 17.38 4.42 7.00 2.59 8.5 3.32 8.08 2.62 11.42 4.73 19.65 3.82 18.91 4.54 7.13 2.42 6.23 1.65 7.44 2.73 6.70 2.62 Pride Emotion Scores. When pride was the dependent variable, the main effect of pre- vs. postinduction was significant, F(1, 87) = 10.09, p = .002, Æž2 = .10, such that pride scores were higher at pre- compared to post-test. The interaction between pre-post and emotion condition was not significant, F(3, 87) = 1.03, p = .38, Æž2 = .03. The main effect of emotion condition was not significant, F(3, 87) = .48, p = .69, Æž2 = .02. 27 Shame Emotion Scores. When shame was the dependent variable, the effect of pre- vs. post-induction was not significant, F(1, 87) = .01, p = .93, Æž2 < .001. The main effect of emotion condition was significant, F(3, 87) = 2.87, p = .04, Æž2 = .09. Post-hoc Tukey tests showed two non-statistically significant effects, but effects with p < .10 : Those in shame condition had higher shame scores than those in the pride and neutral conditions (p = .07 and p < .09, respectively). However, the interaction between pre-post and emotion condition was also significant, F(3, 87) = 5.07, p = .003, Æž2 = .15. Post hoc Tukey tests revealed that at pre-test, there were no differences between the emotion induction groups (p-values > .05). At post-test, the pride and neutral conditions had lower shame scores in comparison to the guilt and shame conditions (p-values < .05). There were no differences between the guilt and shame conditions (p > .05). Guilt Emotion Scores. When guilt was the dependent variable, the main effect of pre-vs. post-induction was not significant, F(1, 87) = 1.08, p = .30, Æž2 = .01. The main effect of emotion condition was significant, F(3, 87) = 4.89, p = .003, Æž2 = .14. Post-hoc Tukey tests indicated that those in the pride condition had lower guilt scores than those in the shame condition (the effect that was not significant, but p = .09). Those in the neutral condition had lower guilt scores than those in the shame (p = .007) and guilt (p < .03) conditions. However, the interaction between pre- and post-emotion induction and emotion condition was significant, F(3, 87) = 7.17, p < .001, Æž2 = .20. Post-hoc Tukey tests indicated that at pre-test, those in the shame condition had higher guilt scores than those in the guilt and neutral conditions (ps < .05). At post-test, those in the pride and neutral conditions had lower guilt scores than those in the shame and guilt conditions (ps < .05). 28 Conclusion. According to the results of the repeated measures ANOVAs, the emotion induction was effective for the emotion conditions of guilt and shame. However, the emotion induction procedure may not have been effective for the pride condition. This may be due to the high pre-induction pride scores: The pre-induction pride scores were approaching ceiling, and thus the pride scores had little room to increase. 25 20 15 Pre-Induction 10 Post-Induction 5 0 Pride Sum Score Shame Sum Score Guilt Sum Score Figure 1. Mean scores of participants in pride emotion condition. 25 20 15 Pre-Induction 10 Post-Induction 5 0 Pride Sum Score Shame Sum Score Guilt Sum Score Figure 2. Mean scores of participants in shame emotion condition. 29 20 15 10 Pre-Induction Post-Induction 5 0 Pride Sum Score Shame Sum Score Guilt Sum Score Figure 3. Mean scores of participants in guilt emotion condition 25 20 15 Pre-Induction 10 Post-Induction 5 0 Pride Sum Score Shame Sum Score Guilt Sum Score Figure 4. Mean scores of participants in neutral emotion condition Inference Tasks To test the effects of emotion and knowledge area on inference generation, mixed ANOVAs were utilized. The independent variables included knowledge area (planning vs. physical; within subjects), sentence type (inference versus control; within subjects), and emotion condition (pride, shame, guilt, neutral; between subjects). The dependent variables included 30 accuracy rates and response times for each type of probe word, including: text probe words, inference probe words, and knowledge validating questions. Please see Table 3 for descriptive statistics of the accuracy rates and response times by condition and inference type. Text Probe Accuracy Rates. First, the accuracy rates for the text probe words were examined. The main effect of knowledge area was significant, such that accuracy rates were higher for physical compared to planning texts, F(1, 90) = 7.58, p =.008, Æž2 = .08. The main effect for sentence type was also significant, such that accuracy rates were higher for control compared to inference texts, F(1, 90) = 12.64, p =.001, Æž2 = .12. None of the other main effects or interactions were significant (F-values < 3.14). Main Effect of Knowledge Area .940 .930 .920 .910 .900 .890 Planning Knowledge Physical Knowledge Figure 5. Mean percentage accuracy rate for text probe words by knowledge area. 31 Main Effect of Sentence Type .960 .940 .920 .900 .880 Control Texts Inference Texts Figure 6. Mean percentage accuracy rate for text probe words by sentence type. Text Probe Response Times. Next, the response times for the text probe words were examined. The main effect of knowledge area was significant, such that response times were slower for planning compared to physical texts, F(1, 90) = 8.63, p=.004, Æž2 = .09. The main effect of sentence type was also significant, such that response times were slower for control compared to inference texts, F(1, 90) = 8.00, p = .006, Æž2 = .08. The interaction between knowledge area and sentence type was significant, F(1, 90) = 8.53, p = .004, Æž2 = .09. Post-hoc Tukey tests indicated that for the planning knowledge texts, response times were slower for the control compared to the inference texts (p < .05). This difference did not exist for physical knowledge texts. For the control texts, the response times were slower for the planning compared to physical knowledge texts (p < .05). This difference did not exist for the inference texts. None of the other main effects or interactions were significant (F-values < .95). 32 Main Effect of Knowledge Area 1140.000 1120.000 1100.000 1080.000 1060.000 1040.000 1020.000 1000.000 980.000 Planning Knowledge Physical Knowledge Figure 7. Mean response times (in milliseconds) for text probe words by knowledge area. Main Effect of Sentence Type 1140.000 1120.000 1100.000 1080.000 1060.000 1040.000 1020.000 1000.000 Control Texts Inference Texts Figure 8. Mean response times (in milliseconds) for text probe words by sentence type. 33 Interaction Between Knowledge Area and Sentence Type 1250.000 1200.000 1150.000 1100.000 Control Text 1050.000 Inference Text 1000.000 950.000 Planning Knowledge Area Physical Knowledge Area Figure 9. Interaction between knowledge area and sentence type with mean response times (in milliseconds) to text probe words. Inference Probe Accuracy Rates. The accuracy rates for inference probe words were examined. The interaction between knowledge area and sentence type was significant, F(1, 90) = 25.91, p < .001, Æž2 = .224. Post-hoc Tukey tests indicated that for the planning knowledge texts, accuracy rates were greater for control texts compared to inference texts (p < .05). For physical knowledge texts, accuracy rates were greater for inference compared to control texts (p < .05). For control sentence types, accuracy rates for planning knowledge areas were greater compared to physical knowledge areas (p < .05). For inference sentence types, accuracy rates were greater for physical knowledge areas compared to planning knowledge areas (p < .05). None of the other main effects or interactions were significant (F-values < 2.61). 34 Interaction Between Knowledge Area and Sentence Type .980 .960 .940 .920 .900 .880 .860 .840 Control Texts Inference Texts Planning Knowledge Physical Knowledge Area Area Figure 10. Interaction between knowledge area and sentence type with mean percentage accuracy rates to inference probe words. Inference Probe Response Times. The response times for inference probe words were examined. The interaction between knowledge area and sentence type was significant, F(1, 90) = 6.54, p =.01, Æž2 = .07. Post-hoc Tukey tests indicated that for physical knowledge texts, response times were slower for control sentence types compared to inference sentence types (p < .05). The difference did not exist for planning knowledge areas. For control sentence types, response times for slower for physical knowledge areas compared to planning knowledge areas. The difference did not exist for inference sentence types. None of the other main effects or interactions were significant (F-values < 3.31). 35 Interaction Between Knowledge Area and Sentence Times 1250.000 1200.000 1150.000 1100.000 1050.000 1000.000 950.000 Control Texts Inference Texts Planning Knowledge Area Physical Knowledge Area Figure 11. Interaction between knowledge area and sentence type with mean response times (in milliseconds) to inference probe words. Knowledge Validating Question Accuracy Rates. The accuracy rates for knowledge validating questions were examined. The main effect of knowledge area was significant, such that accuracy rates were higher for physical knowledge areas compared to planning knowledge areas, F(1, 90) = 14.57, p < .001, Æž2 = .14. The main effect of sentence type was significant, such that accuracy rates were higher for control texts compared to inference texts, F(1, 90) = 5.19, p =.023 , Æž2 = .06. The interaction between knowledge area and emotion condition was not significant, F(3, 87) = 2.44, p = .07, Æž2 = .08. Since the p-value is close to a significant effect, information was inferred using this finding. For the neutral emotion condition, accuracy rates were higher for physical knowledge areas compared to planning knowledge areas (p < .05). This difference did not exist for the shame, guilt, and pride emotion conditions. For planning knowledge areas, accuracy rates were higher for those in the pride emotion condition compared to the neutral emotion condition (p < .05). This difference did not exist for physical knowledge areas. 36 The interaction between sentence type and emotion condition was not significant, F(3, 87) = 2.14, p =.10, Æž2 = .07. As with the interaction between knowledge area and emotion condition, researchers reviewed this finding in terms of implications for the study. For the neutral emotion condition, accuracy rates were higher for inference texts compared to control texts (p < .05). This difference did not exist for the shame, guilt, and pride emotion conditions. For inference sentence types, accuracy rates were higher for the pride emotion condition compared to the neutral emotion condition (p < .05). In addition, accuracy rates were higher for pride compared to shame and guilt emotion conditions regarding inference sentence type, but it was not significant, but p < .10. None of the other main effects or interactions were significant (Fvalues < 1.49). Interaction Between Knowledge Area and Emotion Condition Accuracy Rates .980 .960 .940 Pride Condition .920 Shame Condition .900 Guilt Condition .880 Neutral Condition .860 .840 Physical Knowledge Planning Knowledge Area Area Figure 12. Interaction between knowledge area and emotion condition with mean percentage accuracy rates for knowledge validating questions. 37 Interaction Between Sentence Type and Emotion Condition Accuracy Rates .960 .940 Pride Condition .920 Shame Condition .900 Guilt Condition Neutral Condition .880 .860 Inference Text Control Text Figure 13. Interaction between sentence type and emotion condition with mean percentage accuracy rates for knowledge validating questions. Knowledge Validating Question Response Times. None of the main effects or interactions were significant (F-values < 1.65). 38 Table 3 Accuracy Rates and Response Times for the Inference Generation Task Inference Sentence Type Emotion Knowledge Probe Word Accuracy Rates Response Times Condition Area Mean SE Mean SE Physical .91 .02 994.46 59.89 Pride Planning .93 .02 1072.95 76.57 Physical .89 .02 1060.08 59.89 Shame Text Based Planning .96 .02 1028.20 76.57 Probe Words Physical .94 .02 1106.29 57.34 Guilt Planning .99 .02 1013.51 73.31 Physical .92 .02 1043.25 58.57 Neutral Planning .96 .02 1071.86 74.89 Physical .88 .03 1365.54 99.37 Pride Planning .95 .02 1176.71 59.60 Physical .87 .03 1269.57 99.37 Inference Shame Planning .93 .02 1108.33 59.60 Based Probe Words Physical .90 .02 1234.19 95.14 Guilt Planning .95 .02 1170.22 57.06 Physical .90 .03 1148.83 97.19 Neutral Planning .94 .02 1117.39 58.29 Physical .95 .02 2392.80 135.16 Pride Planning .92 .03 2367.02 167.59 Physical .96 .02 2304.98 135.16 Knowledge Shame Planning .92 .03 2285.46 167.59 Validating Questions Physical .95 .02 2252.46 129.40 Guilt Planning .92 .03 2558.98 160.45 Physical .98 .02 2379.36 132.19 Neutral Planning .93 .03 2475.39 163.90 39 Control Sentence Type Accuracy Rates Response Times Mean SE Mean SE .92 .03 1141.64 78.49 .92 .02 1049.66 60.82 .87 .03 1227.79 78.49 .89 .02 1012.21 60.82 .91 .03 1242.38 75.14 .90 .02 1068.52 58.23 .87 .03 1189.18 76.76 .92 .02 1005.76 59.48 .95 .02 1158.79 65.57 .90 .02 1198.06 74.67 .95 .02 1142.55 65.57 .88 .02 1152.91 74.67 .97 .02 1112.47 62.78 .92 .02 1175.63 71.49 .96 .02 1071.41 64.13 .92 .02 1142.54 73.03 .96 .02 2275.65 168.61 .94 .02 2296.86 156.32 .95 .02 2347.62 168.61 .89 .02 2278.00 156.32 .90 .02 2559.49 161.43 .92 .02 2467.54 149.67 .95 .02 2625.62 164.91 .85 .02 2733.17 152.89 CHAPTER 5 DISCUSSION The goal of the present study was to determine the effects of particular self-conscious emotions (pride, shame, and guilt) on inference generation and knowledge activation. Participants demonstrated statistically significant differences when responding to probe words (text and inference based) as a function of the knowledge area required to generate the inference (planning and physical knowledge areas). However, when accounting for emotion condition, there was not a statically significant effect for probe word recognition with regard to accuracy rates and reading times. There were interactions that were not significant, but the p-value was less than .10 for the accuracy of participants’ responses to knowledge validating questions. The results suggest that emotion may affect knowledge activation, but not the other processes involve in inference generation. The first hypothesis, which stated that the emotion manipulation would be effective, was generally supported. Although the interaction between pre-post induction and emotion condition was not significant for the pride scores, there was a significant interaction for shame and guilt scores. Participants in the pride and neutral emotion conditions demonstrated lower shame scores compared to those in the guilt and shame condition after the emotion induction. Compared to those in the shame and guilt condition, participants in the pride and neutral conditions had lower guilt scores after that emotion induction. These results add to the literature that autobiographical memory tasks can be effective for emotion manipulation (Kraus et al., 2012; Laird & Strout, 2007; Schwarz & Clore, 1983; Westermann et al., 1996). However, pride inductions may be less effective for normal student populations who may already have high baseline pride scores. 40 The second hypothesis, which stated that readers would be more likely to generate physical compared to planning knowledge inferences, was supported. Participants demonstrated faster and more accurate responses to physical knowledge inference texts than planning knowledge inference texts. Accuracy rates were higher and response times were faster when viewing text probe words in physical knowledge texts compared to planning knowledge texts. Accuracy rates were also higher for knowledge validating questions when reading physical knowledge compared to planning knowledge texts. This aligns with findings by Shears et al. (2007) that knowledge area influenced inference generation. In particular, planning knowledge inferences were generated more slowly and less accurately than physical knowledge inferences, indicating that the inference was less likely to be generated. The third hypothesis, which stated that pride should facilitate inference generation, whereas guilt and shame should hamper it, was partially supported when considering knowledge validating questions. The main effect for emotion condition was not statistically significant. However, when examining the interaction between sentence type and emotion condition for accuracy rates regarding the knowledge validating questions, the p-value was not significant, but p < .10. Those in the neutral emotion condition had higher accuracy rates on inference texts compared to control texts. For texts that facilitated inference generation, accuracy rates were higher for those participants in the pride emotion condition compared to participants in the shame, guilt, and neutral emotion conditions. Unexpectedly, a significant interaction between knowledge area, emotion, and inference generation was not found when examining probe word recognition (text probe word and inference probe word). This may be due to the complexity of specific academic emotions, such that mediating variables (such as working memory) may be implicated. Potential mediating variables will be described in more detail later. 41 The fourth hypothesis, which stated that there would be an interaction between emotion and knowledge area, was partially supported. A significant interaction between knowledge area and emotion when examining probe word recognition (text probe word and inference probe word) was not found. However, when examining the accuracy rates of the knowledge validating questions, the p-value was not significant, but p = .07. Those in the neutral emotion condition had higher accuracy rates for the physical knowledge areas compared to planning knowledge areas. Accuracy rates were higher for those in the pride emotion condition compared to the neutral emotion condition for responses to knowledge validating questions in planning knowledge areas. This indicated that emotion might play a greater role when examining knowledge activation in comparison to other aspects of inference generation. Processes Involved in Inference Generation When reading, a person must begin by identifying letters and words and determine what the words mean as a unit. Skilled readers usually do this automatically. The reader then must integrate each new sentence into the context of the rest of the text. Often this requires a reader to make multiple inferences to establish a coherent mental representation of the text. Texts are usually incomplete in nature, meaning that they do not tell the reader everything that needs to be known to understand the text. Thus readers must fill in, or infer, missing pieces of the text with background knowledge of the topic. This requires activating knowledge (either relevant or irrelevant to the topic) and then making appropriate links between the pieces of activated knowledge (Thurlow & van den Broek, 1997). The results suggest that perhaps emotions can affect inference generation, but perhaps only during the knowledge activation phase. This aligns with the affect infusion model in that positive emotions are associated with wider knowledge activation nets. 42 The results of the study contribute to the existing research on inference generation in connection with the affect infusion model. The finding that pride did not have a significant effect, but that p < .10 for knowledge area questions but not other aspects of inference generation might indicate that pride facilitates the activation of knowledge, which aligns with the affect infusion model’s notion that positive affect facilitates making connections between concepts (Forgas, 1995). Hence, knowledge activation could play a mediating role between the effects of emotion and text processing. Expanding on the affect infusion model, Bohn-Gettler and Rapp (2011) found that participants with happy-induced moods engaged in more text-based inferences compared to participants with sad-induced moods. Pride could be compared to happy-induced mood because both are classified as positive emotions. Previous research found that knowledge-based inferences depend on general knowledge and general knowledge activation (Graesser et al., 1994). This could explain why participants in the pride emotion condition had better accuracy rates on knowledge activation questions. Making text-based inferences requires some activation of prior knowledge. If pride facilitated that activation, comprehension could improve among participants. Potential Mediating Variables According to the resource-allocation model, emotional duress increases the likelihood of irrelevant thoughts that interfere with cognitive activities (Ellis & Ashbrook, 1988). Emotions such as depression and sadness can occupy already limited attentional resources, thus leaving fewer cognitive resources available for other tasks. The emotions of shame and guilt could be compared to depression and sadness in that all are negative emotions. Participants in the shame and guilt emotion conditions did not perform as well on knowledge activation questions 43 compared to the participants in the pride condition. This effect could be a function of readers’ attentional resources being allocated toward processing the emotion, which would therefore leave fewer resources available for knowledge activation. Working memory is another potential mediating variable that could play a role in the current study. As described earlier, working memory refers to the capacity a person has to mentally manage information within a specific task. The amount of activation available in working memory varies among individuals, and processing and storage are mediated by activation in working memory (Just & Carpenter, 1992). Shears et al. (2007) found that working memory played a role when readers generated knowledge-based inferences based on specific knowledge areas (physical or planning). Furthermore, Bohn-Gettler and Rapp (2011) found that the effects of emotion on reading depended upon working memory such that readers with higher working memory scores were better able to regulate emotions. Hence, their reading processes were not as likely to be modified as a function of emotion. Therefore, if the current study had included a measure of working memory capacity, it might allow for finding significant differences among participants regarding emotion management and inference generation. It may be that the effects of shame and guilt would be more pronounced among individuals with low working memory. For example, those participants with larger working memory capacities may be able to process an emotion while still completing an inference generation task. However, participants with smaller working memory capacities may have difficulty allocating resources to both emotion management and inference generation. Future Directions A number of design considerations improved the strength of the current study. First, all of the measures have been used in previous research, hence documentation exists that the measures 44 are reliable and valid. Second, this study was performed in a laboratory setting with many controlled variables. Researchers were able to control variables such as testing room atmosphere (room temperature, sound, furniture, etc.), time allowed to complete the activity, and random assignment. Such experimental control contributes to increased reliability and internal validity. However, real academic settings do not usually lend themselves to such a controlled environment. Hence, future research should examine whether these findings would extend into more naturalistic learning settings. However, one challenge to consider is the effects of emotion among readers with disabilities. Originally, the current study intended to include participants with reading disabilities. After solicitation measures and sufficient response time, only two people participated that had a documented reading disability. No participants responded to advertisements sent via the university’s office of disability services. Therefore, information could not be collected or measured with this population. Reading disabilities can influence the ways in which a reader encode and retrieve text, the number of inferences a reader generates, and the working memory span of the reader (Andersson & Ostergren, 2012; Borella, Carretti, & Pelegrina, 2010; Compton, Fuchs, Fuchs, Lambert, & Hamlett, 2012; Laing & Kamhi, 2002). Individuals with reading disabilities are also more prone to experiencing negative emotions in academic settings, such as anxiety and shame (Brown, Higgins, Pierce, Hong, & Thoma, 2003; Denhart, 2008; Maag & Reid, 2006; Shessel & Reiff, 1999; Wilson, Armstrong, Furrie, & Walcot, 2009). Thus, it was hypothesized that there may have been a stronger link between inference generation, knowledge areas, and emotion in participants with reading disabilities. Future research may consider recruiting participants from non-university populations or specific courses for individuals with reading disabilities, obtaining a more racially and gender diverse 45 sample, and incorporating measures such as working memory capacity. Although controlling for working memory may increase internal validity, it potentially reduces external validity in that you cannot control for working memory in academic settings. Additionally, future research might further explore the idea of resource allocation by examining neuroimaging, which might shed light on whether emotion leads to the engagement of various brain areas. Furthermore, Shears et al. (2007) found that there was a difference between knowledge areas when generating inferences. Often standardized tests do not distinguish between inference types and knowledge areas assessed during inference tasks. Future research could further study the need for these distinctions. Implications This research may lead to important educational applications and implications. Overall, these findings support a growing body of work documenting the effects of emotion on comprehension. Pekrun (2002) found that academic emotions relate to a student's motivation, learning strategies, cognitive resources, self-regulation, and academic achievement. The current study supports the notion that academic emotions may also play a role in knowledge activation. Furthermore, the present study demonstrates that emotion may affect the earlier steps of inference generation in its effect on knowledge activation. If emotion can facilitate knowledge activation, this may have important implications for educators seeking to improve the comprehension skills of their students. In classrooms, emotions can affect the learning environment. Teachers should encourage environments that promote positive emotions, such as pride, to help facilitate knowledge activation among students. 46 REFERENCES 47 References Andersson, U. & Ostergren, R. (2012). Number magnitude processing and basic cognitive functions in children with mathematical learning disabilities. Learning and Individual Differences, 22(6), 701 – 714. Bless, H. & Fielder, K (1995). Affective states and the influence of activated general knowledge. Personality and Social Psychology Bulletin, 21,766-778.doi: 10.1177/0146167295217010 Bohn-Gettler, C. M. & Rapp, D. N. (2011). Depending on my mood: Mood-driven influences on text comprehension. Journal of Educational Psychology, 103(3), 562-577.doi: 10.1037/a0023458 Borella, E., Carretti, B., & Pelegrina, S. (2010). The specific role of inhibition in reading comprehension in good and poor comprehenders. Journal of Learning Disabilities, 43(6), 541-552. doi: 10.1177/0022219410371676 Brown, M. R., Higgins, K., Pierce, T., Hong, E, & Thoma, C. (2003). Secondary students' perceptions of school life with regard to alienation: The effects of disability, gender and race. Learning Disability Quarterly, 26(4), 227 - 238. Compton, D. L., Fuchs, L. S., Fuchs, D., Lambert, W., & Hamlett, C. (2012). The cognitive and academic profiles of reading and mathematics learning disabilities. Journal of Learning Disabilities, 45(1), 79 - 95.doi: 10.1177/0022219410393012 Denhart, H. (2008). Deconstructing barriers: Perceptions of students labeled with learning disabilities in higher education. Journal of Learning Disabilities 41(6), 483-497. doi: 10.1177/0022219408321151 Dweck, C. (1986). Motivational processes affecting learning. American Psychologist, 41(10), 1040-1048. Egidi, G., & Gerrig, R. J. (2009). How valence affects language processing: Negativity bias and mood congruence in narrative comprehension. Memory & Cognition, 37(5), 547-55.doi: 10.3758/MC.37.5.547 Ellis, H. C., & Ashbrook, P. W. (1988). Resource allocation model of the effects of depressed mood states on memory. In K. Fiedler & J. Forgas (Eds.), Affect, cognition and social behavior (pp.25-43). Toronto, Ontario, Canada: Hogrefe. 48 Ellis, H.C., Moore, B. A., Varner, L. J., Ottaway, S. A., & Becker, A.S. (1997a). Depressed mood, task organization, cognitive interference, and memory. Journal of Social Behavior and Personality, 12(2), 453 - 470. Ellis, H. C., Ottaway, S. A., Varner, L. J., Becker, A. S., & Moore, B. A. (1997b). Emotion, motivation, and text comprehension: The detection of contradictions in passages. Journal of Experimental Psychology, 126(2), 131-146.doi: 10.1037/0096-3445.126.2.131 Ellis, H.C., Thomas, R.L., & Rodriguez, I.A. (1984). Emotional mood states and memory: Elaborative encoding, semantic processing, and cognitive effort. Journal of Experimental Psychology: Learning, Memory, and Cognition, 10, 470-482. Forgas, J.P. (1995). Mood and judgment: The Affect Infusion Model (AIM). Psychological Bulletin 117(1), 39-66. Graesser, A.C., Singer, M., & Trabasso, T. (1994). Constructing inferences during narrative textcomprehension. Psychological Review, 101(3), 371-395. Just, M. A. & Carpenter, P. A. (1992). A capacity theory of comprehension: individual differences in working memory. Psychological Review, 99(1), 122-149. Kihlstrom, J. F. (1989). On what does mood-dependent memory depend? Journal of Social Behavior and Personality, 4, 23-32. Kintsch, W., & van Dijk, T. A. (1978).Toward a model of text comprehension and production. Psychological Review, 85(5), 363-394.doi: 10.1037/0033-295X.85.5.363 Kraus, M. W., Adler, N., & Chen, T. W. D. (2012). Is the association of subjective SES and selfrated health confounded by negative mood? An experimental approach. Health Psychology. doi: 10.1037/a0027343 Lagattuta, K. H. & Thompson, R. A. (2007). The development of self-conscious emotions: Cognitive processes and social influence. In J.L. Tracy, Robins, and Tangney, The SelfConscious Emotions: Theory and Research (pp. 91 - 113). New York, NY: The Guilford Press. Laing, S. P. & Kamhi, A. G. (2002). The use of think-aloud protocols to compare inferencing abilities in average and below-average readers. Journal of Learning Disabilities 35(5), 436 - 447. 49 Laird, J. D. & Strout, S. (2007). Emotional Behaviors as Emotional Stimuli. In J. A. Coan & J.J.B. Allen (Eds.), Handbook of Emotion Elicitation and Assessment (54 – 64). New York: Oxford University Press. Long, D. L., Oppy, B. J., & Seely, M. R. (1994). Individual differences in the time course of inferential processing. Journal of Experimental Psychology: Learning, Memory, and Cognition, 20, 1456-1470. Maag, J. W., & Reid, R. (2006). Depression among students with learning disabilities: assessing the risk. Journal of Learning Disabilities 39(1), 3 - 10. Magliano, J. P., Trabasso, T., & Graesser, A. C. (1999). Strategic processing during comprehension. Journal of Educational Psychology, 91(4), 615 - 629. Marschall, D. E., Saftner, J. L., & Tangney, J. P. (1994). The State Shame and Guilt Scale. George Mason University, Fairfax, VA. McKoon, G. & Ratcliff, R. (1989). Semantic associations and elaborative inference. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15(2), 326-338. Mirous, H. J. (2010). Mood makes or breaks meaning: Mood effects on inference processing during narrative comprehension in adults & children. (Doctoral Dissertation). Retrieved from ProQuest Dissertations and Theses. (Accession Order No. AAT 3508337). Pekrun, R. (2006). The control-value theory of achievement emotions: Assumptions, corollaries, and implications for educational research and practice. Educational Psychology Review, 18, 315-341.doi:10.1007/s10648-006-9029-9 Pekrun, R., Goetz, T., Frenzel, A. C., Barchfeld, P., & Perry, R. P. (2011). Measuring emotions in students' learning and performance: The achievement emotions questionnaire. Contemporary Educational Psychology, 36, 36 - 48. Schwarz, N., & Clore, G. L. (1983). Mood, misattribution, and judgment of well-being: Informative and directive functions of affective states. Journal of Personality and Social Psychology 45(3), 513 - 523. Shaver, E, Schwartz, J., Kirson, D., & O'Connor, C. (1987). Emotion knowledge: Further exploration of a prototype approach. Journal of Personality and Social Psychology, 52, 1061-1086. 50 Shears, C. & Chiarello, C. (2004). Knowledge-based inferences are not general. Discourse Processes, 38(1), 31-55.doi:10.1207/s15326950dp3801 Shears, C., Miller, V., Ball, M., Hawkins, A., Griggs, J., & Varner, A. (2007). Cognitive demand differences in causal inferences: Characters' plans are more difficult to comprehend than physical causation. Discourse Processes, 43(3), 255-278. doi: 10.1080/01638530701226238 Shessel, I., & Reiff, H. B. (1999). Experiences of adults with learning disabilities: Positive and negative impacts and outcomes. Learning Disability Quarterly 22(4), 305 - 3176. Stringer, R., & Heath, N. (2006). Possible relationships between depressive symptoms and reading. Canadian Journal of School Psychology. 21, 93-105. Tapiero, I., van den Broek, P., & Quintana, M. P. (2002). The mental representation of narrative texts as networks: the role of necessity and sufficiency in the detection of different types of causal relations. Discourse Processes 34(3), 237-258. Thurlow, R., & van den Broek, P. (1997). Automaticity and inference generation during reading. Reading & Writing Quarterly, 13, 165 - 181. Trabasso, T., & van den Broek, P. W. (1985).Causal thinking and representations of narrative events. Journal of Memory and Language, 24, 612–630. Trabasso, T., van den Broek, P., & Suh, S. Y. (1989). Logical necessity and transitivity of causal relations in stories. Discourse Processes, 12, 1–25. Tracy, J. L. & Robins, R. W. (2007). The self in self-conscious emotions: A cognitive appraisal approach. In J.L. Tracy, Robins, and Tangney, The Self-Conscious Emotions: Theory and Research (pp. 3-20). New York, NY: The Guilford Press. Turner, J. E., & Schallert, D. L. (2001). Expectancy–value relationships of shame reactions and shame resiliency. Journal of Educational Psychology, 93(2), 320-329.doi: 10.1037/00220663.93.2.320 van den Broek, P., & Kremer, K.E. (1999). The mind in action: What it means to comprehend during reading. In B. Taylor, M. Graves & P. van den Broek (Eds.), Reading for meaning (pp. 1-31). New York, NY: Teachers College Press. 51 van den Broek, P., Lorch, R. F., Linderholm, T., & Gustafson, M. (2001). The effects of readers’ goals on inference generation and memory for texts. Memory & Cognition, 29(8), 10811087. Westermann, R., Spies, K., Stahl, G., & Hesse, F. W. (1996). Relative effectiveness and validity of mood induction procedures: A meta- analysis. European Journal of Social Psychology, 26(4), 557-580. Williams, L. A. & DeSteno, D. (2008). Pride and perseverance: The motivational role of pride. Journal of Personality and Social Psychology, 94(6), 1007 - 1017. doi: 10.1037/00223514.94.6.1007 Wilson, A. M., Armstrong, C. D., Furrie, A., & Walcot, E. (2009). The mental health of Canadians with self-reported learning disabilities. Journal of Learning Disabilities 42(24), 24 - 40. doi: 10.1177/0022219408326216 52 APPENDIXES 53 APPENDIX A SSGS The following are some statements which may or may not describe how you are feeling right now. Please rate each statement using the 5-point scale below. Remember to rate each statement based on how you are feeling right at this moment. Not feeling this way at all 1. I feel good about myself. Feeling this way somewhat Feeling this way very strongly 1 ------- 2 ------- 3 ------- 4 ------- 5 2. I want to sink into the floor and disappear. 1 ------- 2 ------- 3 ------- 4 ------- 5 3. I feel remorse, regret. 1 ------- 2 ------- 3 ------- 4 ------- 5 4. I feel worthwhile, valuable. 1 ------- 2 ------- 3 ------- 4 ------- 5 5. I feel small. 1 ------- 2 ------- 3 ------- 4 ------- 5 6. I feel tension about something I have done. 1 ------- 2 ------- 3 ------- 4 ------- 5 7. I feel capable, useful. 1 ------- 2 ------- 3 ------- 4 ------- 5 8. I feel like I am a bad person. 1 ------- 2 ------- 3 ------- 4 ------- 5 54 9. I cannot stop thinking about something bad I have done. 1 ------- 2 ------- 3 ------- 4 ------- 5 10. I feel proud. 1 ------- 2 ------- 3 ------- 4 ------- 5 11. I feel humiliated, disgraced. 1 ------- 2 ------- 3 ------- 4 ------- 5 12. I feel like apologizing, confessing. 1 ------- 2 ------- 3 ------- 4 ------- 5 13. I feel pleased about something I have done. 1 ------- 2 ------- 3 ------- 4 ------- 5 14. I feel worthless, powerless. 15. I feel bad about something I have done. 1 ------- 2 ------- 3 ------- 4 ------- 5 1 ------- 2 ------- 3 ------- 4 ------- 5 55 APPENDIX B EMOTION INDUCTION SCRIPTS Script for Pride Condition (Adapted from the script used in Kraus et al., 2012) Researcher: Please close your eyes and remember a specific time when you felt proud or satisfied with something you did. Try to remember the event as if you were actually experiencing it for a second time. (pause) What lead up to that event? (pause) How did you feel in the moment? (pause) Who was with you during that time? (pause)During the moment, what did you see and hear? (pause) What thoughts were running through your mind? (pause) What actions were you doing during the moment? On the paper provided, write about this event in as much detail as possible. Script for Shame Condition (Adapted from the script used in Kraus et al., 2012) Researcher: Please close your eyes and remember a specific time when you felt shamed, disrespected, or devalued. Try to remember the event as if you were actually experiencing it for a second time. (pause) What lead up to that event? (pause) How did you feel in the moment? (pause) Who was with you during that time? (pause)During the moment, what did you see and hear? (pause) What thoughts were running through your mind? (pause) What actions were you doing during the moment? On the paper provided, write about this event in as much detail as possible. Script for Guilt Condition (Adapted from the script used in Kraus et al., 2012) Researcher: Please close your eyes and remember a specific time when you felt guilt or remorse about something you did. Try to remember the event as if you were actually experiencing it for a second time. (pause) What lead up to that event? (pause) How did you feel in the moment? 56 (pause) Who was with you during that time? (pause)During the moment, what did you see and hear? (pause) What thoughts were running through your mind? (pause) What actions were you doing during the moment? On the paper provided, write about this event in as much detail as possible. Script for Neutral Condition (Adapted from the script used in Kraus et al., 2012) Researcher: Please close your eyes and remember waking up this morning. Try to remember the event as if you were actually experiencing it for a second time. (pause) What lead up to that event? (pause) How did you feel in the moment? (pause) Who was with you during that time? (pause)During the moment, what did you see and hear? (pause) What thoughts were running through your mind? (pause) What actions were you doing during the moment? On the paper provided, write about this event in as much detail as possible. 57 APPENDIX C INFERENCE GENERATION TASK – PLANNING KNOWLEDGE STIMULI Type Inference Control Inference Control Inference Control Inference Control Inference Control Inference Control Text Nina's birthday is coming up. She likes balloons. Everyone will be excited. Her dad bought her a cake. Nina's throat is very sore. She likes balloons. Everyone will be excited. Her dad bought her a cake. The doctor is late for surgery. He likes to be on time. His patient is in the operating room. The doctor is speeding on the freeway The doctor is ready for surgery. He likes to be on time. His patient is in the operating room. The doctor is speeding on the freeway. The 15 year old boy was caught drinking beer. His parents were angry. The boy felt guilty. His parents grounded him. The 15 year old boy was babysitting. His parents were angry. The boy felt guilty. His parents grounded him. Sally went to the bank. She drove quickly. She had to get there before closing time. Her rent was due in the morning. Sally went to the parade. She drove quickly. She had to get there before closing time. Her rent was due in the morning. Cindy was low in energy. She had just finished working. Cindy was driving home. She drank a raspberry smoothie. Cindy was listening to the radio. She had just finished working. Cindy was driving home. She drank a raspberry smoothie. Malcolm realized Valentine's Day was tomorrow. He was not prepared. He had to figure something out quickly. He went to the candy shop. Malcolm realized the final exam was tomorrow. He was not prepared. He had to figure something out quickly. He went to the candy shop. 58 Knowledge Verification Question Text Probe Word Inference Probe Word Cake Party Is cake good for sore throats? Freeway Rush Can doctors be late? Boy Punish Rent Cash Do people pay rent? Drank Boost Can a smoothie increase your energy? Tomorrow Girlfriend Do people give candy on special days? Do parents punish their children? Type Inference Control Inference Control Inference Control Inference Control Inference Control Inference Control Inference Control Text Ellen's friend's baby shower is coming up. Ellen made invitations. Her friend is having a boy. Ellen went shopping. Ellen's friend is on her way over. Ellen made invitations. Her friend is having a boy. Ellen went shopping. Erin got straight A's on her report card. It was a huge accomplishment for her. Her parents were so proud. Her parents took her to dinner. Erin received a beautiful postcard. It was a huge accomplishment for her. Her parents were so proud. Her parents took her to dinner. Melvin loves piano music. He tries to support the arts. Music could entertain him for hours. He is taking piano lessons. Melvin loves watching movies. He tries to support the arts. Music could entertain him for hours. He is taking piano lessons. Sherry was constantly eating. It was her favorite thing to do. She loved to try new cuisines. She bought a membership to 24 Hour Fitness. Sherry was constantly traveling. It was her favorite thing to do. She loved to try new cuisines. She bought a membership to 24 Hour Fitness. The little girl was starting school. She was five years old. She wanted to make new friends. She woke up before dawn. The little girl was in school. She was five years old. She wanted to make new friends. She woke up before dawn. Connie's sunglasses were broken. They were her favorite pair. She wanted to replace them. She went to the mall. Connie's sunglasses were the latest style. They were her favorite pair. She wanted to replace them. She went to the mall. Jodie visited the slaughterhouse. She had never been inside the building. The experience had a big impact on her. Jodie became a vegetarian the next day. Jodie visited the historical building. She had never been inside the building. The experience had a big impact on her. Jodie became a vegetarian the next day. 59 Text Probe Word Inference Probe Word Knowledge Verification Question Present Friend Do people buy gifts for friends? Card Reward Do postcards get grades? Lessons Play Do people practice the piano? Constantly Exercise Do people travel while exercising? School Excited Does school start before dawn? Pair Buy Visited Disgusted Are sunglasses needed inside? Do vegetarians eat meat? Type Inference Control Inference Control Inference Control Inference Control Inference Control Inference Control Inference Control Text Elise's clothes were dirty. It had been a month since she'd washed them. She couldn't put it off any longer. She did her laundry. Elise's front steps were dirty. It had been a month since she'd washed them. She couldn't put it off any longer. She did her laundry. Judy wanted to go to the concert. She had been wanting to for years. Judy finally made a commitment to do it. She stopped at the box office. Judy wanted to learn French. She had been wanting to for years. Judy finally made a commitment to do it. She stopped at the box office. The day was getting hot. The sun was shining brightly. Mark considered his plans. Mark decided to go to the beach. The day was going to be busy. The sun was shining brightly. Mark considered his plans. Mark decided to go to the beach. The lazy student did not study for the test. He went to bed late last night. He was tired in class. He copied the answers from his friend. The lazy student rode the bus. He went to bed late last night. He was tired in class. He copied the answers from his friend. Jake lost the bike race. He loves to ride his bike. He has been riding for eight years. He decided to train even harder. Jake lost the bike lock. He loves to ride his bike. He has been riding for eight years. He decided to train even harder. Alex's paint supplies were getting low. He always forgets to refill before it's too late. He has a lot of errands to run. He went to the art supply store. Alex's car was low on fuel. He always forgets to refill before it's too late. He has a lot of errands to run. He went to the art supply store. Martha's daughter is graduating soon. She had just turned eighteen. Martha is so proud of her. Martha went to the flower shop. Martha's daughter is in her room. She had just turned eighteen. Martha is so proud of her. Martha went to the flower shop. 60 Text Probe Word Inference Probe Word Knowledge Verification Question Laundry Machine Stopped Buy Are concert tickets sold in grocery stores? Day Cool Can it be less hot at the beach? Answers Cheat Do students have to copy? Decided Next Can training improve skills? Paint Brushes Do people paint? Daughter Surprise Can flowers be gifts? Can laundry be washed? Type Inference Control Inference Control Inference Control Inference Control Text The wedding was next week. Everyone but Vincent was ready. He was so nervous. Vincent had to rent a tuxedo. The camping trip was next week. Everyone but Vincent was ready. He was so nervous. Vincent had to rent a tuxedo. Martin has been stressed lately by his work. He has so many deadlines. He has more responsibility than ever. He is doing meditation. Martin enjoys his work. He has so many deadlines. He has more responsibility than ever. He is doing meditation. The evening news revealed the dangers of sun exposure. The news is always filled with advice. People tend to trust the news. People go out to buy sunscreen. The evening news revealed the dangers of travel. The news is always filled with advice. People tend to trust the news. People go out to buy sunscreen. Sarah was feeling very sick. Her stomach was in knots. She knew something was about to happen. She went to the doctor. Sarah was feeling very lucky. Her stomach was in knots. She knew something was about to happen. She went to the doctor. 61 Knowledge Verification Question Text Probe Word Inference Probe Word Tuxedo Attend Do people wear dress clothes camping? Work Relax Is meditation relaxing? People Burn Can travel be fun? Doctor Cure Can doctors make you feel better? APPENDIX D INFERENCE GENERATION TASK – PHYSICAL KNOWLEDGE STIMULI Type Inference Control Inference Control Inference Control Inference Control Inference Control Inference Control Text Jessica was in a car accident. She was really upset at first. She finally calmed down. Jessica was injured. Jessica was in a better mood. She was really upset at first. She finally calmed down. Jessica was injured. Mark's car tire went flat. He didn't have any money left. He had just replaced them last week. He couldn't drive anywhere. Mark's car tires were new. He didn't have any money left. He had just replaced them last week. He couldn't drive anywhere. The wine glass fell from the table. It was still full of red wine. The glass had been left there by accident. Pieces of glass went everywhere. The wine glass sat on the table. It was still full of red wine. The glass had been left there by accident. Pieces of glass went everywhere. Sarah took the aspirins. She had been busy all afternoon. She had such a headache. Her pain went away. Sarah looked for her purse. She had been busy all afternoon. She had such a headache. Her pain went away. The bright sun was shining on the field. It was still winter time. Al had just built a snowman. Al's snowman began to melt. The clouds shaded the field. It was still winter time. Al had just built a snowman. Al's snowman began to melt. Dorothy poured the bucket of water on the fire. It was a small camp fire. She was ready to leave. The fire went out. Dorothy placed the bucket of water by the fire. It was a small camp fire. She was ready to leave. The fire went out. 62 Text Probe Word Inference Probe Word Knowledge Verification Question Injured Crash Are people hurt in car accidents? Drive Stuck Do people stop driving because of new tires? Table Shattered Does glass break? Pain Relief Do aspirins relieve pain? Field Heat Will sun melt snow? Bucket Smoke Is fire extinguished by water? Type Inference Control Inference Control Inference Control Inference Control Inference Control Inference Control Inference Control Text Allen forgot to feed his goldfish all week. He needed countless reminders. He could never remember things. Allen's goldfish died. Allen forgot to read his book all week. He needed countless reminders. He could never remember things. Allen's goldfish died. Ronny had too much caffeine. He didn't know what to do with himself. He tried to keep himself entertained. He couldn't go to sleep. Ronny had too much to eat. He didn't know what to do with himself. He tried to keep himself entertained. He couldn't go to sleep. Marie flipped the switch. She'd had such a long day at work. She just wanted to relax. The light went out. Marie flipped the pages of her magazine. She'd had such a long day at work. She just wanted to relax. The light went out. Joe struck the birthday balloon with a pin. He was just playing around. He was always finding new ways to have fun. The balloon burst. Joe struck the birthday card with a pin. He was just playing around. He was always finding new ways to have fun. The balloon burst. Greg bicycled across the broken glass. He's usually very cautious. He was in a hurry to get to work on time. Greg's tire went flat. Greg carefully bicycled across the busy street. He's usually very cautious. He was in a hurry to get to work on time. Greg's tire went flat. Peter chopped the onions for the salad. He wasn't sure how to prepare vegetables. He was helping his mom out. Peter couldn't stop crying. Peter chopped tomatoes for the salad. He wasn't sure how to prepare vegetables. He was helping his mom out. Peter couldn't stop crying. The scientist poured the powder into the boiling water. He made careful observations. He took perfect time measurements. After a few minutes the powder disappeared. The scientist watched the water closely. He made careful observations. He took perfect time measurements. After a few minutes the powder disappeared. 63 Text Probe Word Inference Probe Word Knowledge Verification Question Forgot Starved Do goldfish need to be fed? Sleep Wired Does caffeine make you sleepy? Flipped Dark Can lights be put out? Birthday Poked Bicycled Sharp Will glass pop tires? Salad Sting Do tomatoes make your eyes water? Scientist Mixed Will pins break cards? Will water dissolve metal? Type Inference Control Inference Control Inference Control Inference Control Inference Control Inference Control Inference Control Text Alice tripped on the stairs. The stairs were made of concrete. They were very steep. Alice's knee was bleeding. Alice climbed on the stairs. The stairs were made of concrete. They were very steep. Alice's knee was bleeding. The sugar was all used up. Sugar adds so much flavor. Sugar is the best. The coffee tasted bitter. The sugar was in the cup. Sugar adds so much flavor. Sugar is the best. The coffee tasted bitter. The basketball player hurt his knee. He had been practicing very hard. Basketball was his biggest obsession. He couldn't play in the big game. The basketball player used his skills. He had been practicing very hard. Basketball was his biggest obsession. He couldn't play in the big game. Phillip put water-filled trays in the freezer. He had many things to take care of. His mom had left him a list of chores. He was making ice. Phillip put water-filled trays on the counter. He had many things to take care of. His mom had left him a list of chores. He was making ice. The ringer on Marcy's phone was turned off. She took her phone everywhere. It usually played her favorite song. She had missed several of Sean's calls. The ringer on Marcy's phone was musical. She took her phone everywhere. It usually played her favorite song. She had missed several of Sean's calls. The earthquake shook the house. It was loud and frightening. The house was old and rickety. The house collapsed. The owners loved the house It was loud and frightening. The house was old and rickety. The house collapsed. The paint splattered on the furniture. The couch was brand new. The paint was bright red. The couch was ruined. It was Tuesday. The couch was brand new. The paint was bright red. The couch was ruined. 64 Text Probe Word Inference Probe Word Knowledge Verification Question Knee Fell Coffee Sweeten Is sugar sweet? Player Out Do injuries feel good? Trays Cubes Can you make ice on a counter? Ringer Hear Do phones make noise? Collapsed Destroyed Paint Spilled Can falling cause cuts to knees? Can noise damage a house? Do people paint their couches? Type Inference Control Inference Control Inference Control Inference Control Text Edward forgot to put on sunscreen. The sun was bright. There were barely any clouds in the sky. He got a sunburn. The water was cold. The sun was bright. There were barely any clouds in the sky. He got a sunburn. Alexis slept peacefully. Alexis had to be at work by nine. She had to wake up by eight. Alexis woke up late. The alarm clock was set. Alexis had to be at work by nine. She had to wake up by eight. Alexis woke up late. Stacy got a flu shot. She was an intelligent girl. She was always thinking ahead. Stacy did not catch the flu. Stacy had a good idea. She was an intelligent girl. She was always thinking ahead. Stacy did not catch the flu. The marathon was held on the hottest day of the year. It was held annually. Thousands of people came to watch. Many runners collapsed. The marathon was held on a Saturday. It was held annually. Thousands of people came to watch. Many runners collapsed. 65 Text Probe Word Inference Probe Word Knowledge Verification Question Forgot Apply Do people apply sunscreen? Alarm Morning Do broken alarm clocks work? Flu Prevent Is the flu good to catch? Held Exhausted Can runners get overheated? APPENDIX E DEMOGRAPHIC SURVEY Instructions: Please complete the following items by circling the appropriate option or filling in the blank. As with all aspects of this study, your responses will not be linked to your name, and you will remain anonymous. 1. Age (in years): __________ 2. Gender: __________ 3. Ethnicity: __________ 4. Is English your native language? Yes No If English is not your native language, what is your native language? ____________ 5. Please indicate your current level of study. Graduate 6. What is your college major? __________ 7. What year are you in your current program? _________ 8. Do you have a reading disability that is documented with Wichita State University? Yes 9. Please describe your reading disability: 66 Undergraduate No