Downloads & Free Reading Options - Results

Social loafing research has spanned several decades and fields of study. Research has provided support for both the existence of social loafing and its antecedents within the laboratory, classroom, and work place. Studies regarding the perceptions of social loafing and its effects in the online learning environment, however, are largely non-existent. This study surveyed 227 online learning students who were participating in online learning groups. The study seeks to determine whether the perception of social loafing exists within online learning groups. In addition, several psychosocial factors identified in face-to-face environments are analyzed to determine their impact in online learning groups. Evidence supports both the perception of social loafing in online learning groups as well as similarities between social loafing antecedents in face-to-face groups and those in the online learning environment. (Contains 2 tables.)

This action research examines how focused organization, group roles, and gender grouping impact student participation when working in a cooperative group setting. Fifty-two sixth graders were studied for a period of nine weeks. Results show when students are organized in their cooperative groups, there will be an increase in student participation. Participation also increased when students were given assigned roles. Lastly, this research shows that my hypothesis was incorrect by thinking participation would increase when students work in same gender cooperative groups. To come to these results, data was collected using a triangular approach focusing on observations, change in grades, and questionnaires. The following are appended: (1) Data Collection Matrix; (2) Teacher Observation Log: Focused Organization; (3) Focused Organization; (4) Teacher Observation Log: Group Roles; (5) Student Roles; (6) Group Roles; (7) Teacher Observation Log: Gender Grouping; and (8) Gender Grouping. (Contains 5 figures.)

The article discusses the optimal types of tasks and exercises that help students to effectively enrich their vocabulary in the Russian language textbook. In addition to the use of traditional types of vocabulary work, innovative ways to increase vocabulary that develop speech and broaden teachers' worldviews are also demonstrated

The article discusses the optimal types of tasks and exercises that help students to effectively enrich their vocabulary in the Russian language textbook. In addition to the use of traditional types of vocabulary work, innovative ways to increase vocabulary that develop speech and broaden teachers' worldviews are also demonstrated

A study examined daytime education groups, a major phenomenon in Irish adult education. Results showed that 96 groups were operating, mostly in suburban areas, particularly Dublin. These groups had 8,723 adults participating in their courses and programs. What made them unique was their voluntary, locally based nature. The groups received little or no recognition, encouragement, or support from the Department of Education. They were generally dependent on local level support. More than half operated from private homes. Two-thirds rented premises. One in three did not have any insurance coverage for their activities. Two-thirds provided child care, which approximately one-fifth of the participants used. The groups provided a large number and wide range of educational courses and programs. In contrast to statutory providers, the groups had greater control over the whole process. Although crafts and skill-based learning were still emphasized, the main type of learning was oriented toward social and personal understanding. Most courses were run on a strict self-financing basis. More than half provided subsidies for the cost of fees. Courses were not linked to any overall system of certification and accreditation. Several problems were identified: funding, maintaining the voluntary commitment and motivation of group members, lack of permanent premises, and lack of facilities. (Appendixes include the daytime education groups, listed alphabetically by county, and questionnaire.) (YLB)

A study examined daytime education groups, a major phenomenon in Irish adult education. Results showed that 96 groups were operating, mostly in suburban areas, particularly Dublin. These groups had 8,723 adults participating in their courses and programs. What made them unique was their voluntary, locally based nature. The groups received little or no recognition, encouragement, or support from the Department of Education. They were generally dependent on local level support. More than half operated from private homes. Two-thirds rented premises. One in three did not have any insurance coverage for their activities. Two-thirds provided child care, which approximately one-fifth of the participants used. The groups provided a large number and wide range of educational courses and programs. In contrast to statutory providers, the groups had greater control over the whole process. Although crafts and skill-based learning were still emphasized, the main type of learning was oriented toward social and personal understanding. Most courses were run on a strict self-financing basis. More than half provided subsidies for the cost of fees. Courses were not linked to any overall system of certification and accreditation. Several problems were identified: funding, maintaining the voluntary commitment and motivation of group members, lack of permanent premises, and lack of facilities. (Appendixes include the daytime education groups, listed alphabetically by county, and questionnaire.) (YLB)

A form of group learning, cohorts, has become increasingly attractive to administrators, instructors, and participants in adult education. Basic academic skills cohort learning supports three types of knowing: instrumental, socializing, and self-authoring; whereas, in higher and adult education cohort learning, the development of critical reflection and knowledge construction is the focus. Cohort learners report such positive effects as increased critical thinking skills, greater individual development, enhanced knowledge base, and learning motivation. Some research has revealed that not all cohorts work well, due to characteristics and behaviors of group members. Actual measurement of cohort learning outcomes has been inadequately studied. Educators can enhance the cohort experience by doing the following: (1) developing group relationships at the beginning; (2) balancing group and individual development; (3) providing an environment that both supports and challenges; and 4) acknowledging and addressing group and individual tensions. (Contains 12 references.) (AJ)

Money laundering is a major global problem, enabling criminal organisations to hide their ill-gotten gains and to finance further operations. Prevention of money laundering is seen as a high priority by many governments, however detection of money laundering without prior knowledge of predicate crimes remains a significant challenge. Previous detection systems have tended to focus on individuals, considering transaction histories and applying anomaly detection to identify suspicious behaviour. However, money laundering involves groups of collaborating individuals, and evidence of money laundering may only be apparent when the collective behaviour of these groups is considered. In this paper we describe a detection system that is capable of analysing group behaviour, using a combination of network analysis and supervised learning. This system is designed for real-world application and operates on networks consisting of millions of interacting parties. Evaluation of the system using real-world data indicates that suspicious activity is successfully detected. Importantly, the system exhibits a low rate of false positives, and is therefore suitable for use in a live intelligence environment.

Money laundering is a major global problem, enabling criminal organisations to hide their ill-gotten gains and to finance further operations. Prevention of money laundering is seen as a high priority by many governments, however detection of money laundering without prior knowledge of predicate crimes remains a significant challenge. Previous detection systems have tended to focus on individuals, considering transaction histories and applying anomaly detection to identify suspicious behaviour. However, money laundering involves groups of collaborating individuals, and evidence of money laundering may only be apparent when the collective behaviour of these groups is considered. In this paper we describe a detection system that is capable of analysing group behaviour, using a combination of network analysis and supervised learning. This system is designed for real-world application and operates on networks consisting of millions of interacting parties. Evaluation of the system using real-world data indicates that suspicious activity is successfully detected. Importantly, the system exhibits a low rate of false positives, and is therefore suitable for use in a live intelligence environment.

Money laundering is a major global problem, enabling criminal organisations to hide their ill-gotten gains and to finance further operations. Prevention of money laundering is seen as a high priority by many governments, however detection of money laundering without prior knowledge of predicate crimes remains a significant challenge. Previous detection systems have tended to focus on individuals, considering transaction histories and applying anomaly detection to identify suspicious behaviour. However, money laundering involves groups of collaborating individuals, and evidence of money laundering may only be apparent when the collective behaviour of these groups is considered. In this paper we describe a detection system that is capable of analysing group behaviour, using a combination of network analysis and supervised learning. This system is designed for real-world application and operates on networks consisting of millions of interacting parties. Evaluation of the system using real-world data indicates that suspicious activity is successfully detected. Importantly, the system exhibits a low rate of false positives, and is therefore suitable for use in a live intelligence environment.

In a study of school readiness, 90 American born, middle class Jewish children were tested before entering the first grade and divided into two groups. The groups were well-matched with one difference: children were either Ashkenazic (of European descent) or Sephardic (of Syrian descent). Families of both of these groups, however, had been in the United States for at least 25 years. Cognitive measures such as the Stanford-Binet, Columbia Mental Maturity Scale, Peabody Picture Vocabulary Test, and the Bender Visual Motor Gestalt Test were given to the children; parent attitude scales were given to the mothers. Additional data were collected on the educational and financial aspirations of the mothers for their children. Analysis of the data showed significant school readiness differences. The Ashkenazic children, whose cultural background possibly supported academic achievement, were more prepared than the Sephardic children, whose training seemed to stress financial success. Study findings suggest that implicit cultural factors, aspirations, and mores affect children's school readiness, even when poverty and other disadvantages are absent. (MS)

"A publication of the Horace Mann-Lincoln Institute of School Experimentation, Teachers College, Columbia University."

Includes bibliographies and index

preregistration

The use of cooperative learning at the college level was studied by investigating whether different instructor intervention approaches would affect the achievement of college students using cooperative learning methods. Subjects were 97 undergraduate elementary education majors enrolled in three sections of a curriculum course at the University of South Florida (Tampa) College of Education during the second semester of the 1990-91 academic year. Nineteen learning teams were formed and randomly assigned to treatment conditions as follows: (1) directions and materials provided, but no instructor assistance (control group); (2) instructors provided advance organizers; and (3) instructors met with groups after sessions to answer questions and provide assistance. All treatment groups read the same materials and followed a similar schedule of activities. A pretest was followed by a posttest and an attitude questionnaire after the 2-week period. On the posttest, the control and follow-up discussion groups had nearly identical mean scores, with the mean for the advance organizer group more than two points higher. Students liked the cooperative learning situation, believed they learned the materials well, and preferred instructor assistance to the control condition. Results suggest that advance organizers may yield greater learning than do follow-up discussions. Five tables present study findings, and a 17-item list of references is included. (SLD)

Modern crowd theories agree that collective behavior is the result of the underlying interactions among small groups of individuals. In this work, we propose a novel algorithm for detecting social groups in crowds by means of a Correlation Clustering procedure on people trajectories. The affinity between crowd members is learned through an online formulation of the Structural SVM framework and a set of specifically designed features characterizing both their physical and social identity, inspired by Proxemic theory, Granger causality, DTW and Heat-maps. To adhere to sociological observations, we introduce a loss function (G-MITRE) able to deal with the complexity of evaluating group detection performances. We show our algorithm achieves state-of-the-art results when relying on both ground truth trajectories and tracklets previously extracted by available detector/tracker systems.

Bulletin (West Virginia University. Agricultural Experiment Station)

It was a pleasure to welcome @GW University College Republicans to the Embassy of Israel! Thank you for visiting and engaging in meaningful discussions about the U.S.- Israel relationship. We appreciate your interest and look forward to continuing the dialogue. We are always happy to host any student groups, interested in learning more about our embassy, don't hesitate to reach out if you want to arrange a visit!

It was a pleasure to welcome @GW University College Republicans to the Embassy of Israel! Thank you for visiting and engaging in meaningful discussions about the U.S.- Israel relationship. We appreciate your interest and look forward to continuing the dialogue. We are always happy to host any student groups, interested in learning more about our embassy, don't hesitate to reach out if you want to arrange a visit!

This final report describes Project BLEND (Beginning Learning Experiences in Developmentally Inclusive Groups and at Home), a project designed to develop, implement, evaluate, and disseminate an ecological model for early intervention for children with developmental delays. Project BLEND included the following components: (a) the partnership (family, child care, BLEND) that was the context in which all activities for supporting each child's development were designed and implemented; (b) service coordination that assisted parents in establishing child care services and coordinating services from other agencies; and transition planning that assisted the child and family in making the transition to their next environment. In the first phase of the project, this model was developed and refined. In the second phase, the model was fully implemented with an existing early intervention center and community child care programs in middle Tennessee. In the third phase, the model was replicated in two sites: a school system in an urban setting that serves young children with disabilities (ages 3-5) and a second site that serves infants and toddlers with developmental delays and their families in a rural community. The report describes the project's activities and includes project implementation and replication checklists in the appendices. (Contains 12 references.) (Author/CR)

The Peer Assisted Learning (PAL) program at the University of Minnesota is a primary academic support program for historically difficult, introductory college courses that serve as gatekeepers to academic degree programs. Based upon operating principles of other academic support programs and educational theories, PAL is integrated into the courses it serves. The PAL groups review essential course content, model cognitive learning strategies to deepen understanding, and promote metacognitive awareness so students are autonomous learners in courses without academic support services. The PAL approach operates at the confluence of collaborative learning, cooperative learning groups, and learning communities. This article provides a detailed overview of the PAL model, educational theories upon which it is based, and how variations of it are implemented at the institution. Quantitative and qualitative studies reveal academic and personal benefits for participating students and those serving as PAL facilitators. The studies validate the role of PAL with closing the achievement gap between students of different ethnicities and levels of academic preparedness for rigorous college courses in mathematics and science.

The aim of this paper is to present "the three steps method" of the self-formation of collaborative groups in a problem-based learning environment. The self-formation of collaborative groups is based on sharing of accountability among students for solving instructional problems. The steps of the method are planning collaborative problem solving, self-evaluation of students, and building collaborative groups. The planning comprises determination of the nomenclature of higher order thinking (HOT) skills, defining the instructional problems and their complexity levels, creating problem groups according to the complexity levels, setting the problem-relevant HOT skills, determining the accountability measure and the assessments of accountability for solving the problems. The self-evaluation includes self-detection of personal HOT skills, measurement of the diversity between the personal HOT skills and the problem-relevant skills based on the proposed diversity measure, and self-evaluation of willingness and desire of a student to take accountability for solving the instructional problems. The personal willingness is evaluated by the diversity measure. The desire is guided by the accountability assessments for problem solving. Coordination of the self-evaluation outcomes allows building collaborative groups. A group's composition is adjusted by the specific requirements of an instructor.

We begin the article with a brief discussion of why groups are recommended in language teaching. After that, we describe what Cooperative Learning (CL) is. Then, we suggest ways in which CL and Information Technology (IT) fit well together. This is followed by some examples of how to combine CL with IT. [This article was published in Teaching of English Language and Literature," v13 n2 p5-8 1997.]

56p

56p

Blended learning can facilitate many opportunities about usage of online discussion for different purposes. As a tool in blended learning, online discussions widen the walls of classroom discussions in a virtual manner. Especially, in teacher education programs case based discussions are very helpful for pre-service teachers to practice real-life situations. By this way, the link between theory and practice can be established. This study aims to examine the pre-service teachers' experiences about three different discussion structures in a case based blended learning environment. This study also explored their thoughts toward the effects of online discussion on their professional development and their suggestions to increase the effectiveness of online discussion. The participants of this study were 14 pre-service teachers from the department of Early Childhood Education (ECE) of a funded university. The six open-ended questions were asked in order to explore the opinions of the pre-service teachers. The qualitative data were analyzed through the matrix displays. (Contains 1 figure.)

This study investigated the effect of using differentiated instruction by integrating multiple intelligences and learning styles on solving problems, achievement in, and attitudes towards math in six graders with learning disabilities in cooperative groups. A total of 60 students identified with LD were invited to participate. The sample was randomly divided into two groups; experimental (n = 30 boys) and control (n = 30 boys). ANCOVA and T.test were employed for data analysis. Findings from this study indicated the effectiveness of differentiated instruction by integrating multiple intelligences and learning styles on solving problems, achievement in, and attitudes towards math in the target students. On the basis of the findings, the study advocated for the effectiveness of using differentiated instruction by integrating multiple intelligences and learning styles on solving problems, achievement in, and attitudes towards math in learning disabled students.

This paper presents a methodological discussion of the potential and challenges of involving mobile eye tracking technology in studies of knowledge generation and learning in a science centre context. The methodological exploration is based on eye-tracking studies of audience interaction and knowledge generation in the technology-enhanced health promotion exhibition PULSE at a science centre in Copenhagen, Denmark. The current study is part of the larger PULSE project, which aims to develop innovative health promotion activities where a science centre exhibition is a key setting. The primary target groups were families with children age 6-12 years and school classes with students from 4th to 6th grade. The main purpose of the study was to understand the methodological potential and challenges mobile eye tracking comprises during the different stages of research on informal e-learning in a science centre context utilising digital platforms to enhance informal learning and interaction. The paper presents how eye-tracking methods influence research on: 1) an interventional level: what role eye tracking and eye-tracking equipment plays in interventions; 2) a data level: what new types of data eye-tracking methods specifically contribute; and 3) an analytical level: how analysis of eye tracking can supplement and contribute to other analytical approaches. Finally, the article discusses how the methodological approach presented invites consideration of other ways of understanding how users experience technology-enhanced exhibitions.

Rogers (1970) suggests that group therapy may be both safer and far more effective than individual counseling and therapy in achieving conditions which encourage learning and personal growth. There are many "how to's" and "do's and don'ts" which will ameliorate conditions for creating learning, or problem solving, groups. The author delineates many of those conditions in this paper along with some requisite dimensions of leader behavior, number behavior, and group process. (Author/HMV)

In this paper the design of a workflow support tool for competence based distance learning in a group setting is discussed. The design is based on a stakeholder analysis and crash-tested in an actual course setting. Preliminary findings suggest that some well-known problems have been solved, but further more in depth research is needed to assess the quality of the design with respect to more subtle issues.

Pedagogical shifts favouring collaborative learning and findings of recent studies have raised concerns regarding the claim that gifted students prefer to learn alone. The purpose of this study was to further investigate if, when and how high ability learners want to work with or without others. The distributions of 416 high ability students (n = 416; Gr. 3-8) responses to survey items were analyzed. Items assessed their general feelings about working alone and in a group and the appeal of specific conditions. Although a majority indicated they enjoyed learning alone, more also enjoyed group work-under certain conditions. Age differences were found but none related to gender. More of the younger students enjoyed teaching their peers while more of the older students were eager to contribute to group discussions and be seated in clusters. Sitting alone became increasingly unpopular with older students. The broad variability in the distribution of students' ratings across conditions demonstrated the preferences of high ability learners are sensitive to many factors in the setting, not just the involvement of others. High ability learners may prefer to work alone when attractive conditions for working in groups are not available. Evidence-based guidelines for group work are offered.

The impact on achievement of (a) cooperative learning in which members discussed how well their group was functioning and how they could improve its effectiveness, (b) cooperative learning without any group processing, and (c) individualistic learning were compared on daily achievement, post-instructional achievement, and retention. Eighty-four third-grade students were randomly assigned to the three conditions stratifying for sex and ability level. The results indicate that the high-, medium-, and low-achieving students in the cooperation with group processing condition achieved higher on all three measures than did the students in the other two conditions. Students in the cooperation without group processing condition, furthermore, achieved higher on all three measures than did the students in the individualistic condition. Keywords include: Cooperation; Achievement; and Group Processing.

Collective intelligence, which aggregates the shared information from large crowds, is often negatively impacted by unreliable information sources with the low quality data. This becomes a barrier to the effective use of collective intelligence in a variety of applications. In order to address this issue, we propose a probabilistic model to jointly assess the reliability of sources and find the true data. We observe that different sources are often not independent of each other. Instead, sources are prone to be mutually influenced, which makes them dependent when sharing information with each other. High dependency between sources makes collective intelligence vulnerable to the overuse of redundant (and possibly incorrect) information from the dependent sources. Thus, we reveal the latent group structure among dependent sources, and aggregate the information at the group level rather than from individual sources directly. This can prevent the collective intelligence from being inappropriately dominated by dependent sources. We will also explicitly reveal the reliability of groups, and minimize the negative impacts of unreliable groups. Experimental results on real-world data sets show the effectiveness of the proposed approach with respect to existing algorithms.

Continuing conflicts between violent groups and states generate an ever-present demand for higher quality and more timely information to support operations to combat terrorism. In particular, better ways are needed to understand how terrorist and insurgent groups adapt over time into more-effective organizations and increasingly dangerous threats. To adapt, terrorist organizations must learn. A group's ability to learn determines its chance of success, since learning is the link between what the group wants to do and its ability to gather the needed information and resources to actually do it. Despite the importance of terrorist group learning, comparatively little focused research effort has been directed at understanding this process and identifying the factors that influence group learning ability. While relevant data and insights can be found in the literature on terrorism and terrorist organizations, this information has not been collected and systematically analyzed to assess its importance from the perspective of efforts to combat terrorism. This study addresses that need in an effort to both analyze current understanding and stimulate further study and research in this area. The RAND Corporation conducted an analysis of organizational learning in terrorist groups and assessed its implications for efforts to combat terrorism. The work was performed between November 2003 and November 2004, a period during which the threat of international terrorism was high and concern about the capabilities of terrorist organizations and how they might change over time was a central focus of policy debate and U.S. homeland security planning. The study is described in this report and in Aptitude for Destruction, Volume 1: Organizational Learning in Terrorist Groups and Its Implications for Combating Terrorism, which applies the analytical framework described in the second part of this report to the practical demands of intelligence and law enforcement activities.

We propose a self-supervised framework that learns to group visual entities based on their rate of co-occurrence in space and time. To model statistical dependencies between the entities, we set up a simple binary classification problem in which the goal is to predict if two visual primitives occur in the same spatial or temporal context. We apply this framework to three domains: learning patch affinities from spatial adjacency in images, learning frame affinities from temporal adjacency in videos, and learning photo affinities from geospatial proximity in image collections. We demonstrate that in each case the learned affinities uncover meaningful semantic groupings. From patch affinities we generate object proposals that are competitive with state-of-the-art supervised methods. From frame affinities we generate movie scene segmentations that correlate well with DVD chapter structure. Finally, from geospatial affinities we learn groups that relate well to semantic place categories.

This practicum aimed to improve global awareness among middle school students. One specific goal was to increase the students' concept of the world and their global perspective. A second goal was to increase the students' geographic knowledge, enabling them to locate sites of current world events and affairs on a map or globe. Students were taught the geographic locational skills needed for learning the global issues and their localities in the world. The students synthesized facts about global interdependence. They wrote essays and reports derived from library research. Each student kept a current events notebook throughout the implementation of the practicum process. They were provided with 32 weeks of lessons, activities, and projects divided into 3 phases. Outcomes from this practicum experience were very positive. All six of the practicum's objectives were achieved and surpassed. The students gained knowledge of global interdependence. Their geographic skills improved. Finally, the students gained an interest in worldwide current events and affairs. The global awareness survey instrument and the geographic locational skills test are appended. (DB)

Individual and group assessments of quiz accuracy and students' discrimination of what they know and what they do not know regarding course material were examined using confidence ratings from 22 graduate students, 47 undergraduates, and their 23 heterogeneous learning groups over 6 quizzes. Students first took each multiple choice quiz as individuals and then as a group. Students received instruction regarding metamemory, confidence calibrations, and overconfidence after the first three quizzes. It was hypothesized that individuals and their groups would use this information to adjust their confidence ratings to discriminate appropriately between correct and wrong quiz answers. Within groups, students improved their accuracy, but did not appropriately adjust their confidence judgments. Moreover, the improved accuracy in groups came at a cost of increased confidence for wrong answers. Neither relevant information about metamemory nor assignment to structured learning groups was effective at improving students' assignments of confidence judgments, and may even have made it worse. Factors affecting group decision making appear to be high individual confidence and a majority effect, with educational status a marginally contributing component. There are six figures and two tables. (Contains 19 references.) (Author/SLD)

Individual and group assessments of quiz accuracy and students' discrimination of what they know and what they do not know regarding course material were examined using confidence ratings from 22 graduate students, 47 undergraduates, and their 23 heterogeneous learning groups over 6 quizzes. Students first took each multiple choice quiz as individuals and then as a group. Students received instruction regarding metamemory, confidence calibrations, and overconfidence after the first three quizzes. It was hypothesized that individuals and their groups would use this information to adjust their confidence ratings to discriminate appropriately between correct and wrong quiz answers. Within groups, students improved their accuracy, but did not appropriately adjust their confidence judgments. Moreover, the improved accuracy in groups came at a cost of increased confidence for wrong answers. Neither relevant information about metamemory nor assignment to structured learning groups was effective at improving students' assignments of confidence judgments, and may even have made it worse. Factors affecting group decision making appear to be high individual confidence and a majority effect, with educational status a marginally contributing component. There are six figures and two tables. (Contains 19 references.) (Author/SLD)

Simulations: Processing of chaos and memory storage. In view of our previously published findings showing that motor patterns represent adaptive behaviors may be generated by chaotic activity, we have used computer simulations to examine the ability of simple networks to learn to process chaotic signals and to perform complex operations on them. These studies have shown that even simple networks can be used to understand how networks store information, much of which information can not have been obtained from the more complex biological systems. As one example, an important and unexpected finding is that networks having trainable thresholds, in addition to trainable synapses, can performs computations that trainable synapses alone can not, regardless of the number synapses that may be included in the network. Another finding is that when networks must learn several tasks simultaneously, the effective size network is self-limiting, and probably does not require special algorithmic rules for limiting the size of successfully computing neural connections.

Simulations: Processing of chaos and memory storage. In view of our previously published findings showing that motor patterns represent adaptive behaviors may be generated by chaotic activity, we have used computer simulations to examine the ability of simple networks to learn to process chaotic signals and to perform complex operations on them. These studies have shown that even simple networks can be used to understand how networks store information, much of which information can not have been obtained from the more complex biological systems. As one example, an important and unexpected finding is that networks having trainable thresholds, in addition to trainable synapses, can performs computations that trainable synapses alone can not, regardless of the number synapses that may be included in the network. Another finding is that when networks must learn several tasks simultaneously, the effective size network is self-limiting, and probably does not require special algorithmic rules for limiting the size of successfully computing neural connections.

The effects of cooperative learning on students' perceptions of themselves and their roles in academic settings are explored. A group of 28 students from seven intermediate classrooms in an urban school system were selected to be videotaped while participating in a cooperative problem-solving lesson and were subsequently interviewed. The students were grouped heterogeneously by race and sex, and homogeneously by ability. The students' responses to the cooperative learning setting were analyzed within the following four major categories: (1) achievement level; (2) personal worth; (3) formation of friendships with students of different ethnic groups and sexes; and (4) enjoyment of school. Although responses in all four categories were positive for the majority of the students, significant differences in the perceptions of black and white, and of female and male students were found. These findings suggest that despite the demonstrated value of cooperative learning settings, they may subtly reinforce racial and sexual normative roles. The study concludes with the following three recommendations for future research: (1) teacher effect and experience should be considered and controlled; (2) individual group dynamics should be carefully studied; and (3) teacher role in structuring outcomes in cooperative groups should be investigated. A 13-item list of references is appended. (AF)

Peer-Led Team Learning (PLTL) is a structured learning model where a team leader is appointed to facilitate collaborative problem solving among students for Science, Technology, Engineering and Mathematics (STEM) courses. This paper presents an informed HMM-based speaker diarization system. The minimum duration of short conversationalturns and number of participating students were fed as side information to the HMM system. A modified form of Bayesian Information Criterion (BIC) was used for iterative merging and re-segmentation. Finally, we used the diarization output to compute a novel dominance score based on unsupervised acoustic analysis.

Peer-Led Team Learning (PLTL) is a structured learning model where a team leader is appointed to facilitate collaborative problem solving among students for Science, Technology, Engineering and Mathematics (STEM) courses. This paper presents an informed HMM-based speaker diarization system. The minimum duration of short conversationalturns and number of participating students were fed as side information to the HMM system. A modified form of Bayesian Information Criterion (BIC) was used for iterative merging and re-segmentation. Finally, we used the diarization output to compute a novel dominance score based on unsupervised acoustic analysis.

This paper identifies differences between the three ethnic groups, namely, Kadazans/Dusuns, Bajaus, and other minority ethnic groups on the beliefs about learning English as a second language based on the five variables, that is, language aptitude, language learning difficulty, language learning and communicating strategies, nature of language learning as well as learning motivation and expectation. A modified version of Horwitz's (1987) Beliefs about Language Learning Inventory (BALLI) was distributed to all the 254 form four students in Lahat Datu but only 193 students responded. This survey instrument consists of 34 items. Results of a One-way ANOVA test showed a significant difference among ethnic groups on motivation and expectation in learning English. Bajau students showed the highest motivation and expectation in learning English, followed by other ethnic minorities compared to Kadazan/Dusun students. Although participants scored high for motivation and expectation to learn English, they perceived that English is a difficult language to acquire. BALLI is used to identify the misconceptions or beliefs held among learners and find ways to reduce the negative impacts in learning English.

Workplace learning (WPL) is widely accepted in universities as a valuable component of educating for professional practices. Most often though, the focus of WPL is on helping students transition into the workforce, neglecting the role it can play in helping students transition into university. Using an online questionnaire and interviews, a study was conducted with undergraduate students enrolled in a regional Australian university to better understand their experiences of WPL in the first year of their studies. Findings from this study showed that although there are challenges associated with students undertaking WPL in the first year of university courses, WPL experiences were highly valued by students. Findings also highlighted that WPL had potential as a retention strategy for first year students in general, and students from under-represented groups in particular.

Changes in the environment require constant adaptation to optimize future behavior. Learning through reinforcement is essential for adaptation in response to changes in the environment; such ability is thought to play a key role in human cognitive development as well as in the emergence and maintenance of psychiatric disorders, such as ADHD, depression or anxiety. Uncertainty and valence are considered to be two important factors for cognitive adaptation during reinforcement learning (RL). Clinical symptoms have been shown to modulate RL with respect to uncertainty and valence conditions, but little is known how psychopathology affects RL in clinically diagnosed compared to typically developing children and adolescents. In this study, we used a probabilistic learning task in which we experimentally manipulated uncertainty and valence levels. Uncertainty was manipulated by two uncertainty conditions, one which included stochasticity, and another which added volatility. Stochasticity or expected uncertainty refers to learning action-outcome contingencies that were probabilistic and stable (80% contingent and 20% non-contingent feedback). Uncertainty due to the probabilistic outcomes becomes expected once the preferred action-outcome contingency had been learned. Volatility further adds uncertainty to the probabilistic action-outcome contingencies: at change points, the learned action-outcome contingencies have to be reversed. Volatility represents unexpected uncertainty, since the exact points of the change cannot be anticipated. Valence has been examined by effects of both valence condition and prediction error (PE) valence which makes it challenging to compare and generalize results. In our study, our task allowed us to examine both these valence effects. For PE effects, valence is determined by whether the received outcome relative to the expected outcome was positive or negative at each trial (Eckstein, Master, Dahl, Wilbrecht, & Collins, 2022; Rosenbaum, Grassie, & Hartley, 2022). Notably, PE valence represents trial-to-trial valence effects, and learning from a positive PE may lead to a momentary positive surprise that can change quickly from positive to negative, if the next trial comes with a negative PE. For valence as block-wise condition effects, valence is determined by whether the absolute outcome value was positive or negative, such as in reward learning and punishment learning, respectively (Palminteri, Kilford, Coricelli, & Blakemore, 2016). Here, valence may unfold over multiple trials and lead to more enduring valence effects. During reward learning, the preferred outcome was to receive 3 coins compared to 1 coin only; during punishment learning, the loss of 1 coin was preferred than the loss of 3 coins. We also added a third condition of both mixed reward and punishments, which was commonly used in studies that only examined PE valence effects. In this third condition, which we will refer to as mixed condition, the preferred outcome was to receive 1 coin compared to losing 1 coin. Notably, PE valence can be additionally examined in any of the block-wise valence conditions. A pilot study indicated that the task can be applied to both clinical and neurotypical 8-18-year-old children and adolescents, and that their adaptation performance was modulated by valence and uncertainty. Our sample of the main study will include 120 children and adolescents between the age of 8 and 18, one clinical group (n=60) with diagnoses of anxiety, depression or ADHD, and a neurotypical group (n=60). We aim to delineate differences between psychiatric conditions (ADHD, anxiety, depression) for cognitive adaptation in respect to valence and uncertainty. We also aim to explore potential transdiagnostic effects of internalizing and externalizing psychopathology during cognitive adaptation. We will use both behavioral measures, such as accuracy, switching behavior and reaction time, as well as computational modeling measures, with the aim to capture adaptations in latent prediction-related parameters. How one learns in the learning conditions that differ by uncertainty and valence might depend upon the current symptom severity, both in clinical as well as neurotypical groups. Therefore, our aim of this study is to examine individual differences in clinical symptoms and their effects on uncertainty and valence during learning. We will further explore the moderating roles of the individual’s environment, state and trait, such as socioeconomic disparities, motivational traits and momentary affective state. Research Questions (RQ) Effects of Depression Major depressive disorder is an affective disorder whose primary symptoms are low mood and loss of motivation and pleasure in daily life. Despite its primarily affective symptoms, there is robust evidence of cognitive deficits in depression (Gotlib & Joormann, 2010), which may be related to the hypoactivity of serotonin, dopamine and noradrenaline (Eshel & Roiser, 2010; Ruhé, Mason, & Schene, 2007). Of note, depression is considered an internalizing disorder, as opposed to externalizing disorders (Achenbach, 1966; Carver, Johnson, & Timpano, 2017), with a large overlap in symptoms such as with anxiety disorders, specifically within a distress-based or negative affect symptom cluster (Watson, 2005). The overlap in negative affect symptoms between depression and anxiety may contribute to shared differences during RL. Although previous studies have found similarities and differences in the learning effects of depression and anxiety, a direct comparison with a task that manipulates both valence and uncertainty during RL has not been done. DEP-RQ1: What are the effects of depression on uncertainty conditions during RL? Studies that used RL with stable environments, reflecting low uncertainty, found reduced learning performance (Elliott, Sahakian, Herrod, Robbins, & Paykel, 1997; Steele, Meyer, & Ebmeier, 2004). Two studies used computational models to examine learning processes in depression, with in inconsistent results. The studies examined various parameters, including learning rate and reward sensitivity. Learning rate serves as a metric for quantifying how individuals update future values, whereas reward sensitivity reflects an individual's responsiveness to perceived outcomes. Reward learning rates were negatively related to anhedonic depression symptoms (Brown et al., 2021), but unrelated in a meta-analysis (Huys, Pizzagalli, Bogdan, & Dayan, 2013). Reward sensitivity in turn was either reduced (Huys et al., 2013) or increased (Brown et al., 2021). These contrasting findings may partly be explained by the use of a pavlovian and an instrumental learning task, respectively, and our study will focus on instrumental learning. One study reported reduced choice sensitivity in a more depressed student group (Kunisato et al., 2012), suggesting that depression is associated with less value-dependent choice behavior. Because the parameters of reward sensitivity and choice sensitivity are computationally interchangeable, current models cannot answer whether the parameters captured decision-related or feedback-related effects of depression (Browning, Paulus, & Huys, 2022). Additional parameter differences have been reported, such as a more negative valuation of received outcomes during punishment learning in depression (Brown et al., 2021), suggesting valence effects, or a decreased attentional breadth in relation to higher trait rumination (Hitchcock et al., 2022), suggesting impaired attention and memory processes. Further studies are needed to establish robust links between depressive symptoms and computational parameters in low uncertainty learning conditions during RL. Consistent with tasks of relatively low uncertainty, studies with more volatile tasks such as probabilistic reversal learning have reported impaired learning performance in depression (Dombrovski et al., 2010; Dombrovski, Szanto, Clark, Reynolds, & Siegle, 2013; Mukherjee, Filipowicz, Vo, Satterthwaite, & Kable, 2020; Mukherjee, Lee, Kazinka, D Satterthwaite, & Kable, 2020; Must, Horvath, Nemeth, & Janka, 2013; Rupprechter, Stankevicius, Huys, Steele, & Seriès, 2018). Less optimal switching behavior in term of lower win-stay and higher lose-switch behavior (particularly after misleading feedback) was also observed (Dickstein et al., 2010; Dombrovski et al., 2015; Mukherjee, Filipowicz, et al., 2020; Murphy, Michael, Robbins, & Sahakian, 2003; Taylor Tavares et al., 2008). In contrast, only one study also reported no learning differences in relation to depression (Brolsma et al., 2020). Some studies applied computational models, with inconsistent findings, similar to studies with low uncertainty tasks. One study reported reduced learning rates in depressed individuals compared to healthy controls (Mukherjee, Filipowicz, et al., 2020), while another study did not find such differences (Brolsma et al., 2020). Additionally, reduced choice sensitivity was reported (Mukherjee, Filipowicz, et al., 2020; Rupprechter et al., 2018), as well as a decreased memory of observed rewards, putatively due to working memory constraints (Rupprechter et al., 2018). Overall, behavioral findings show a relatively consistent picture of reduced learning performance in depression under both low and high uncertainty learning. However, these studies mostly used tasks of mixed-valence or reward conditions only, and the picture regarding uncertainty under punishment learning is less clear. Further, robust links between computational parameters, depression and uncertainty are yet to be established. There is some evidence that learning rate and choice sensitivity estimates are lower across uncertainty conditions. DEP-RQ2: What are the effects of depression on valence conditions during RL? Past RL studies of depression have usually used either conditions for reward and punishment learning separately, or they used PE valence within a reward task to determine valence effects (i.e. gaining more or less than expected). While both PE valence as well as valence condition (learning from positive or from negative feedback) have been studied extensively in depression, these valence effects may represent distinct characteristics on learning which need to be disentangled. To shed light on the relevance of the way valence is manipulated in relation to depression, our study includes 3 valence conditions: reward learning, punishment learning and mixed valence learning. Of note, many studies that have examined reward learning used a mixed valence feedback scheme. For reward learning, past studies found reduced learning performance in depression (Forbes, Shaw, & Dahl, 2007; Herzallah et al., 2013; Morris, Bylsma, Yaroslavsky, Kovacs, & Rottenberg, 2015; Robinson, Cools, Carlisi, Sahakian, & Drevets, 2012). Reduced reward learning was related to reduced striatal activation following positive PE (Robinson, Cools, Carlisi, et al., 2012). Further, both reduced striatal activations and reduced reward learning were a predictor of future depressive symptoms during adolescence (Forbes et al., 2007; Morgan, Olino, McMakin, Ryan, & Forbes, 2013). In terms of positive PE effects on learning, a simulation meta-analysis found lower learning rates, while results from the conventional meta-analysis shows only showed lower learning rates at trend (Pike & Robinson, 2022). Estimation tasks found a reduced positivity bias (Garrett et al., 2014; Sharot, Korn, & Dolan, 2011), which may implicate a reduced positivity and confirmation bias in depression during RL as well. Here, learning rates from positive and confirmatory PE are usually higher than from negative and disconfirmatory PE, which leads to higher learning performance compared to symmetric learning rates (Palminteri & Lebreton, 2022). In contrast, a developmental study found no effect of depressive symptoms on learning rate asymmetry (Nussenbaum, Velez, Washington, Hamling, & Hartley, 2022). In brief, both reward condition and positive PE effects suggest that reward learning may be reduced in depression, but the underlying computational mechanisms during RL are less clear. For punishment learning, two studies found higher learning performance relative to that of reward learning (Herzallah et al., 2013; Timmer, Sescousse, Van Der Schaaf, Esselink, & Cools, 2017). An acute lowering of serotonin levels in healthy individuals through depleting its precursor tryptophan lead to enhanced punishment learning, which suggests that low serotonin increase punishment learning, but do not affect reward learning (Cools, Robinson, & Sahakian, 2008; Robinson, Cools, & Sahakian, 2012). Indeed, depressed individuals with SSRI medication showed lower punishment learning relative to unmedicated depression individuals, but no differences in reward leaning (Herzallah et al., 2013). Studies with computational models examined either punishment learning rates or negative PE learning rates. Punishment learning rates showed inconsistent findings, with higher learning rates in depressed and anxious individuals compared to healthy controls (Aylward et al., 2019), or lower learning rates in depressed compared to healthy controls (Mukherjee, Filipowicz, et al., 2020). In terms of negative PE valence, a meta-analysis found higher learning rates in a simulation approach, while results in the conventional approach showed no effect (Pike & Robinson, 2022). Another study did not find changed learning rates in relation to depression, but a higher shift towards perceiving punishments as larger (Brown et al., 2021). To summarize, despite some inconsistencies in the literature, both punishment condition and negative PE effects point towards enhanced punishment learning in depression. Overall, the finding of both reduced reward learning and enhanced punishment learning in depression has led to the idea that depressed individuals differ in their subjective valuation and exhibit an exaggerated loss aversion compared to what prospect theory has established in healthy individuals (Chen, Takahashi, Nakagawa, Inoue, & Kusumi, 2015). There is evidence of learning deviations under both valence conditions in depression, but it is yet unclear whether these emerge in earlier developmental phases. DEP-RQ3: Are there differential effects of the anhedonic and negative affect symptoms clusters during RL? Depression is a highly heterogeneous disorder, which might explain some of the inconsistencies found during learning in depression. A closer look at specific symptom effects might prove useful to shed new light on the effects on RL. Depression is often distinguished by the symptom clusters of anhedonia and negative affect. Anhedonia is the loss of pleasure or lack of reactivity to pleasurable stimuli. While negative affect in depression shows considerable overlap with anxiety disorders, anhedonia is more independent of anxiety, but overlapping considerably with other disorders such as schizophrenia or addiction (Pizzagalli, 2014). Negative affect or distress putatively is influenced by serotonergic function, whereas anhedonia depends more upon dopaminergic function (Nutt, 2008). Therefore, these two symptom clusters within depression may have distinct effects on learning behavior. Since serotonergic function has been shown to specifically target punishment learning (Cools et al., 2008; Robinson, Cools, & Sahakian, 2012), it is possible that enhanced punishment learning can be linked to negative affect. A direct link between negative affect and punishment learning in depression has been previously reported, but remains understudied (Brown et al., 2021). Anhedonia in turn has been studied extensively, and consistently identified related reduced striatal activations during reward anticipation and reward delivery in adolescents and adults (Gradin et al., 2011; Stringaris et al., 2015). Anhedonic symptoms were related to less optimal switching behavior (Pizzagalli, Iosifescu, Hallett, Ratner, & Fava, 2008), reduced learning rates from positive and negative PE (Chase et al., 2010), specifically to reduced reward learning rates (Brown et al., 2021), and to more exploratory decision-making during learning (Harlé, Guo, Zhang, Paulus, & Yu, 2017). Overall, this suggests that anhedonic and negative affect symptoms clusters may have differential effects during RL, and further studies are needed to examine how negative affect modulates RL. Effects of Anxiety Anxiety disorders can differ with regard to the situation (e.g. social anxiety) or the objects (e.g. specific phobia) that cause symptoms such as aversive affective state, somatic stress symptoms and perception of sustained threat. Anxiety is increasingly recognized as developmental disorder (Leonardo & Hen, 2008), and together with depression and obsessive-compulsive disorder, it has been categorized as internalizing disorder, as opposed to externalizing disorders, e.g. ADHD and conduct disorder (Achenbach, 1966; Carver et al., 2017). Among internalizing disorders, conditions can be further divided into fear-based or distress-based symptom clusters (Watson, 2005), however, other symptom clusters have been distinguished as well (e.g. somatic and cognitive anxiety symptoms). Maladaptive uncertainty learning has been found for both the fear-based (Brown, Price, & Dombrovski, 2023) and the distress-based symptom clusters (Hammond, Xu, Ai, & Van Dam, 2023). A study that distinguished by somatic and cognitive symptom clusters found stronger influences of the somatic symptom cluster on RL (Fan, Gershman, & Phelps, 2022; Wise & Dolan, 2020). Neurally, individuals with anxiety disorders exhibited higher noradrenaline levels (Kalk, Nutt, & Lingford-Hughes, 2011) and dysregulated serotonergic modulation (Dayan & Huys, 2009), which may relate to the observed learning disruptions. Further, structural brain differences have been found to mediate the effect of unpredictability experienced during childhood on anxiety and depression symptoms in adults, which suggests that the uncertainty experienced as a child increases the risk for anxiety and depression and may lead to structural brain changes during development (Wang, Cao, Zheng, Chen, & Zhu, 2023). It remains unclear to what extent anxiety symptoms explain learning disruptions across both clinical and normative anxiety symptoms equally, and whether age modulates these effects during development. In this study, we will examine how effects of anxiety symptoms are present during RL under different levels of uncertainty and valence in both clinical and typically developing individuals. ANX-RQ1: What are the effects of anxiety on uncertainty conditions during RL? The current literature suggests that uncertainty processing is central to learning disruptions in anxiety symptoms (Brown et al., 2023). The aversive consequence of uncertainty in anxiety has been shown to impair behavioral learning performance already during low uncertainty, such as expected uncertainty, which is inherent in probabilistic but stable learning (LaFreniere & Newman, 2019). Under high uncertainty, such as in volatile environments, higher anxiety reduced learning performance and predicted aberrant switching behavior, such as reduced win-stay, increased or reduced lose-shift behavior (Dickstein et al., 2010; Hein, de Fockert, & Ruiz, 2021; Huang, Thompson, & Paulus, 2017; Piray, Ly, Roelofs, Cools, & Toni, 2019; Xia, Xu, Yang, Gu, & Zhang, 2021). Maladaptive uncertainty learning may be related to chronic underconfidence in individuals with anxiety and depression, as they showed disproportionally larger updates from low confidence decisions (Katyal, Huys, Dolan, & Fleming, 2023). One study also reported an inverted-u-shaped effect of trait anxiety on learning: both low and high trait anxiety were related to lower behavioral learning performance compared to intermediate trait anxiety (Aberg, Toren, & Paz, 2022). A nonlinear effect of anxiety on learning may be linked to the nonlinear effect of arousal on optimal task engagement that is attributed to noradrenaline function (Eckstein, Guerra-Carrillo, Miller Singley, & Bunge, 2017). Computational studies suggest that individuals with higher trait anxiety or with internalizing symptoms have difficulty adjusting their learning rate to the learning environments. Specifically, higher trait anxiety or internalizing symptoms predicted smaller learning rate adjustments between stable and volatile learning environments, which reflects less flexible learning (Browning, Behrens, Jocham, O’Reilly, & Bishop, 2015; Gagne, Zika, Dayan, & Bishop, 2020). The less flexible updating in relation to high internalizing symptoms was further characterized by a reduced updating after positive PE when action-outcome contingencies where changing throughout the learning block (Gagne et al., 2020). Beyond effects on learning rates, there is evidence that anxiety modulates decision-making during RL. Trait anxiety predicted the strategy by which participants explored alternative choices: higher trait anxiety participants exhibited a shift from value-based exploration towards uncertainty-related exploration (Aberg et al., 2022). This is in line with another study which reported that among individuals that were best described by a value-free choice strategy win-stay-lose-shift, higher state anxiety individuals exploited this value-free strategy more. Overall, anxiety may enhance less optimal choice strategies at the cost of value-based decision-making. To summarize, the currently literature found that anxiety has disruptive effects on learning under both low and high levels of uncertainty. Behavioral studies found reduced learning performance and less optimal switching behavior, but learning performance effects may also be nonlinearly related to anxiety symptoms. Computational studies have focused on anxiety effects between high and low uncertainty, but it is unclear how parameters of each uncertainty level are related to anxiety. These studies reported maladaptive learning rate adjustments to the statistics of the environment, as well as a shift from value-based to value-free decision behavior, but no studies have reported both effects of learning rate and decision behavior. Further, some studies reported combined effects of anxiety and depression, and it is not clear to what extent anxiety and depression show distinct effects on learning. Despite its relevance of anxiety disorders during development (Leonardo & Hen, 2008), it is unknown whether anxiety disrupts learning in children and adolescents similarly as in adults, since the only study with a pediatric sample found no behavioral effects of anxiety and did not apply computational models (Dickstein et al., 2010). This study will examine how anxiety symptoms in children and adolescents, both of clinical and normative symptoms range, show effects of uncertainty that are independent of depressive symptoms. ANX-RQ2: What are the effects of anxiety on valence conditions during RL? The effects of anxiety on uncertainty learning were most consistently found during punishment or threat learning, both behaviorally and computationally (Aberg et al., 2022; Browning et al., 2015; Gagne et al., 2020; LaFreniere & Newman, 2019; Piray et al., 2019). Some studies extended the found effects of uncertainty to both punishment and reward learning (Aberg et al., 2022; Gagne et al., 2020; LaFreniere & Newman, 2019) or did not find effects in either valence condition (Dickstein et al., 2010). Other studies did not examine reward and punishment separately, but they presented both rewards and punishments mixed within learning blocks; these studies found behaviorally reduced learning performance (Xia et al., 2021), and in a simulation meta-analysis higher negative PE learning rates and lower positive PE learning across depression and anxiety (Pike & Robinson, 2022). A recent study using naturalistic learning found that higher negative PE learning rates explained lower and less precise expectations about future academic outcomes and predicted the long-term development of anxiety (Villano et al., 2023). In brief, this new body of literature suggests that higher anxiety symptoms may predict aberrant processing of negative outcomes, both under punishment learning and from negative PE, and that this may be a risk factor for the development of anxiety. It is unknown whether anxiety modulates processing of negative outcomes already in children and adolescents. ANX-RQ3: Are there differential effects of anxiety symptom clusters during RL? The distinctions between fear-based and distress-based anxiety symptoms as well as somatic anxiety and cognitive anxiety symptoms have helped to examine learning disruptions in relation to more specific symptom clusters. Although it has been proposed that fear-based symptoms exert stronger effects on RL (Brown et al., 2023), the current literature suggests that multiple symptom clusters may have effects on RL. Studies that specified effects of symptom clusters found that somatic anxiety reduced uncertainty-directed exploration, whereas cognitive anxiety increased uncertainty-directed exploration (Fan et al., 2022). Similarly, divergent roles of these two symptoms clusters have been found during aversive learning (Wise & Dolan, 2020). The previously described maladaptive learning rate adjustment between stable and volatile environments was attributed to distress-based symptoms, which are less specific to anxiety and likely reflect internalizing symptoms more broadly (Gagne et al., 2020; Hammond et al., 2023). Importantly, the effects of fear-based symptoms have not been examined in these two studies. Therefore, it is unknown whether the effects on learning rate can be better attributed to fear-based symptoms rather than to distress-based symptoms. Since maladaptive learning rate adjustments were also linked to reduced pupil dilation changes, reflecting noradrenaline function and arousal, physiological anxiety symptoms may underlie uncertainty-related effects (Browning et al., 2015). The divergent effects of somatic and cognitive anxiety symptoms on decision behavior are difficult to interpret, but it seems that the role of somatic anxiety symptoms is more in line with the effect of a study that did not distinguish by symptom clusters (Aberg et al., 2022). The current definitions of symptoms clusters vary (eg. fear-based and distress-based symptoms, somatic and cognitive anxiety symptoms), which makes is difficult to integrate findings. Effects of ADHD Attention-deficit/hyperactivity disorder (ADHD) is commonly described as a developmental brain disorder, in which the hypofunctioning of noradrenaline and dopamine lead to reduced cognitive performance such as during RL (Plichta & Scheres, 2014; Ziegler, Pedersen, Mowinckel, & Biele, 2016). Notably, not just neurotransmitter hypofunction such as in ADHD, but also hyperfunction such as during stress (Arnsten, 1999, 2009) has been shown to impair performance. Therefore, an inverted-u-shaped relationship between neurotransmitter function and cognitive performance has been proposed (Biederman & Spencer, 1999; Del Campo, Chamberlain, Sahakian, & Robbins, 2011). Beyond functional differences, the effect of ADHD has been observed in the brain structure, likely as long-term consequence of brain neuroplasticity. Brain regions implicated in reward processing and value representation, such as the ventral striatum and the orbitofrontal cortex, exhibited smaller volumes in individuals with ADHD compared to controls (Carmona et al., 2009; Hesslinger et al., 2002). The well-established biological differences in ADHD have led to effective psychopharmacological treatments in ameliorating ADHD symptoms. Stimulants such as methylphenidate are used to target and normalize dopamine function. Some studies that controlled for methylphenidate use showed that it improved learning performance (Luman, Goos, & Oosterlaan, 2015; Pelham, Milich, & Walker, 1986), which suggests that medication use is a relevant confounding factor for examining effects of ADHD on learning. It remains unclear whether ADHD symptoms explain learning differences only by diagnostic categorization such as reported in case-control studies, or whether these learning effects also extend to subclinical ADHD symptoms in clinical and neurotypically developing groups. Further, it remains unclear whether the effects of ADHD on RL are more pronounced during sensitive developmental periods such as during childhood and adolescence. In this study, we will examine how effects of ADHD symptoms are present during RL under different levels of uncertainty and valence. ADHD-RQ1: What are the effects of ADHD on uncertainty conditions during RL? Several studies have examined the effects of ADHD diagnosis on RL under different levels of uncertainty (Hulsbosch et al., 2021). Under low uncertainty, studies showed mixed findings: some found no effects of ADHD (Luman et al., 2015; Oades & Müller, 1997; Wiesner, Molzow, Prehn-Kristensen, & Baving, 2017), while other found reduced learning (Frank, Santamaria, O’Reilly, & Willcutt, 2007; Gabay, Shahbari-Khateb, & Mendelsohn, 2018; Luman et al., 2021; Shephard, Jackson, & Groom, 2016). Additionally two studies reported slower and more variable reaction times (Frank et al., 2007; Gabay et al., 2018). Under high uncertainty, studies show similarly mixed evidence: two studies found no effects of ADHD (Chantiluke et al., 2015; Finger et al., 2008), one study reported computational effects (Hauser et al., 2014) in terms of a more explorative choice behavior, and two studies showed reduced learning performance in ADHD compared to a control group (Itami & Ca, 2002; Shephard et al., 2016). To summarize, the current literature shows mixed findings during RL under both low and high uncertainty. If an effect was found for ADHD, learning performance was reduced, choices were more erratic, and reaction times were slower and more variable. This study aims to shed new light on whether uncertainty modulates effects of ADHD symptoms during RL, and how these effects can be captured both behaviorally and computationally. ADHD-RQ2: What are the effects of ADHD on valence conditions during RL? No studies have examined the effects of punishment learning in ADHD so far. One study has reported differences in response to positive and negative feedback: negative feedback elicited stronger activations in children with ADHD compared to neurotypical children (Van Meel, Oosterlaan, Heslenfeld, & Sergeant, 2005). Given that the abovementioned brain deficits in ADHD were found in relation to reward learning, processing of negative feedback may be relatively spared and lead to an imbalance towards stronger processing of negative information compared to positive information. It is currently unknown whether the effects of ADHD might be less pronounced during punishment learning and learning from negative PE, compared to learning from reward and positive PE. ADHD-RQ3: Are there differential effects of the symptom clusters hyperactivity/impulsivity and inattentiveness during RL? ADHD consists of a set of symptoms, most notably the symptom clusters hyperactivity/impulsivity and inattentiveness. While previous studies reported correlations of the symptom clusters to learning deficits (Gabay et al., 2018; Luman et al., 2015), it is currently unclear whether the symptom clusters show separable contributions to RL. Computational models may help to identify separable contributions. For example, one theoretical model showed that aberrant learning parameters reproduced impulsive behavior in a delayed response time task (Williams & Dayan, 2005). Impulsive and hyperactive symptoms might be to learning parameters such as learning rate. No study has reported learning rate effects of ADHD, although all theoretical models of ADHD agree that dopamine hypofunction leads to aberrant learning parameters (Frank et al., 2007; Sagvolden, Johansen, Aase, & Russell, 2005; Tripp & Wickens, 2008). Thus, it is unknown whether hyperactivity/impulsivity symptoms and the underlying dopamine hypofunction specifically explain effects on learning rate. The attention deficit in ADHD has not been linked to computational parameters yet. However, noradrenaline function, which is fundamental in attentional processes, has been linked decision behavior during RL (Dubois et al., 2021, 2020; Frank et al., 2007). In contrast, one influential theoretical account of ADHD has linked attention deficits to the dopaminergic mesocortical pathway (Sagvolden et al., 2005). Therefore, it remains unclear whether inattentiveness in ADHD is specifically linked to noradrenergic function and attributed explorative decision behavior, as well as slower and more variable reaction times during RL.

Changes in the environment require constant adaptation to optimize future behavior. Learning through reinforcement is essential for adaptation in response to changes in the environment; such ability is thought to play a key role in human cognitive development as well as in the emergence and maintenance of psychiatric disorders, such as ADHD, depression or anxiety. Uncertainty and valence are considered to be two important factors for cognitive adaptation during reinforcement learning (RL). Clinical symptoms have been shown to modulate RL with respect to uncertainty and valence conditions, but little is known how psychopathology affects RL in clinically diagnosed compared to typically developing children and adolescents. In this study, we used a probabilistic learning task in which we experimentally manipulated uncertainty and valence levels. Uncertainty was manipulated by two uncertainty conditions, one which included stochasticity, and another which added volatility. Stochasticity or expected uncertainty refers to learning action-outcome contingencies that were probabilistic and stable (80% contingent and 20% non-contingent feedback). Uncertainty due to the probabilistic outcomes becomes expected once the preferred action-outcome contingency had been learned. Volatility further adds uncertainty to the probabilistic action-outcome contingencies: at change points, the learned action-outcome contingencies have to be reversed. Volatility represents unexpected uncertainty, since the exact points of the change cannot be anticipated. Valence has been examined by effects of both valence condition and prediction error (PE) valence which makes it challenging to compare and generalize results. In our study, our task allowed us to examine both these valence effects. For PE effects, valence is determined by whether the received outcome relative to the expected outcome was positive or negative at each trial (Eckstein, Master, Dahl, Wilbrecht, & Collins, 2022; Rosenbaum, Grassie, & Hartley, 2022). Notably, PE valence represents trial-to-trial valence effects, and learning from a positive PE may lead to a momentary positive surprise that can change quickly from positive to negative, if the next trial comes with a negative PE. For valence as block-wise condition effects, valence is determined by whether the absolute outcome value was positive or negative, such as in reward learning and punishment learning, respectively (Palminteri, Kilford, Coricelli, & Blakemore, 2016). Here, valence may unfold over multiple trials and lead to more enduring valence effects. During reward learning, the preferred outcome was to receive 3 coins compared to 1 coin only; during punishment learning, the loss of 1 coin was preferred than the loss of 3 coins. We also added a third condition of both mixed reward and punishments, which was commonly used in studies that only examined PE valence effects. In this third condition, which we will refer to as mixed condition, the preferred outcome was to receive 1 coin compared to losing 1 coin. Notably, PE valence can be additionally examined in any of the block-wise valence conditions. A pilot study indicated that the task can be applied to both clinical and neurotypical 8-18-year-old children and adolescents, and that their adaptation performance was modulated by valence and uncertainty. Our sample of the main study will include 120 children and adolescents between the age of 8 and 18, one clinical group (n=60) with diagnoses of anxiety, depression or ADHD, and a neurotypical group (n=60). We aim to delineate differences between psychiatric conditions (ADHD, anxiety, depression) for cognitive adaptation in respect to valence and uncertainty. We also aim to explore potential transdiagnostic effects of internalizing and externalizing psychopathology during cognitive adaptation. We will use both behavioral measures, such as accuracy, switching behavior and reaction time, as well as computational modeling measures, with the aim to capture adaptations in latent prediction-related parameters. How one learns in the learning conditions that differ by uncertainty and valence might depend upon the current symptom severity, both in clinical as well as neurotypical groups. Therefore, our aim of this study is to examine individual differences in clinical symptoms and their effects on uncertainty and valence during learning. We will further explore the moderating roles of the individual’s environment, state and trait, such as socioeconomic disparities, motivational traits and momentary affective state. Research Questions (RQ) Effects of Depression Major depressive disorder is an affective disorder whose primary symptoms are low mood and loss of motivation and pleasure in daily life. Despite its primarily affective symptoms, there is robust evidence of cognitive deficits in depression (Gotlib & Joormann, 2010), which may be related to the hypoactivity of serotonin, dopamine and noradrenaline (Eshel & Roiser, 2010; Ruhé, Mason, & Schene, 2007). Of note, depression is considered an internalizing disorder, as opposed to externalizing disorders (Achenbach, 1966; Carver, Johnson, & Timpano, 2017), with a large overlap in symptoms such as with anxiety disorders, specifically within a distress-based or negative affect symptom cluster (Watson, 2005). The overlap in negative affect symptoms between depression and anxiety may contribute to shared differences during RL. Although previous studies have found similarities and differences in the learning effects of depression and anxiety, a direct comparison with a task that manipulates both valence and uncertainty during RL has not been done. DEP-RQ1: What are the effects of depression on uncertainty conditions during RL? Studies that used RL with stable environments, reflecting low uncertainty, found reduced learning performance (Elliott, Sahakian, Herrod, Robbins, & Paykel, 1997; Steele, Meyer, & Ebmeier, 2004). Two studies used computational models to examine learning processes in depression, with in inconsistent results. The studies examined various parameters, including learning rate and reward sensitivity. Learning rate serves as a metric for quantifying how individuals update future values, whereas reward sensitivity reflects an individual's responsiveness to perceived outcomes. Reward learning rates were negatively related to anhedonic depression symptoms (Brown et al., 2021), but unrelated in a meta-analysis (Huys, Pizzagalli, Bogdan, & Dayan, 2013). Reward sensitivity in turn was either reduced (Huys et al., 2013) or increased (Brown et al., 2021). These contrasting findings may partly be explained by the use of a pavlovian and an instrumental learning task, respectively, and our study will focus on instrumental learning. One study reported reduced choice sensitivity in a more depressed student group (Kunisato et al., 2012), suggesting that depression is associated with less value-dependent choice behavior. Because the parameters of reward sensitivity and choice sensitivity are computationally interchangeable, current models cannot answer whether the parameters captured decision-related or feedback-related effects of depression (Browning, Paulus, & Huys, 2022). Additional parameter differences have been reported, such as a more negative valuation of received outcomes during punishment learning in depression (Brown et al., 2021), suggesting valence effects, or a decreased attentional breadth in relation to higher trait rumination (Hitchcock et al., 2022), suggesting impaired attention and memory processes. Further studies are needed to establish robust links between depressive symptoms and computational parameters in low uncertainty learning conditions during RL. Consistent with tasks of relatively low uncertainty, studies with more volatile tasks such as probabilistic reversal learning have reported impaired learning performance in depression (Dombrovski et al., 2010; Dombrovski, Szanto, Clark, Reynolds, & Siegle, 2013; Mukherjee, Filipowicz, Vo, Satterthwaite, & Kable, 2020; Mukherjee, Lee, Kazinka, D Satterthwaite, & Kable, 2020; Must, Horvath, Nemeth, & Janka, 2013; Rupprechter, Stankevicius, Huys, Steele, & Seriès, 2018). Less optimal switching behavior in term of lower win-stay and higher lose-switch behavior (particularly after misleading feedback) was also observed (Dickstein et al., 2010; Dombrovski et al., 2015; Mukherjee, Filipowicz, et al., 2020; Murphy, Michael, Robbins, & Sahakian, 2003; Taylor Tavares et al., 2008). In contrast, only one study also reported no learning differences in relation to depression (Brolsma et al., 2020). Some studies applied computational models, with inconsistent findings, similar to studies with low uncertainty tasks. One study reported reduced learning rates in depressed individuals compared to healthy controls (Mukherjee, Filipowicz, et al., 2020), while another study did not find such differences (Brolsma et al., 2020). Additionally, reduced choice sensitivity was reported (Mukherjee, Filipowicz, et al., 2020; Rupprechter et al., 2018), as well as a decreased memory of observed rewards, putatively due to working memory constraints (Rupprechter et al., 2018). Overall, behavioral findings show a relatively consistent picture of reduced learning performance in depression under both low and high uncertainty learning. However, these studies mostly used tasks of mixed-valence or reward conditions only, and the picture regarding uncertainty under punishment learning is less clear. Further, robust links between computational parameters, depression and uncertainty are yet to be established. There is some evidence that learning rate and choice sensitivity estimates are lower across uncertainty conditions. DEP-RQ2: What are the effects of depression on valence conditions during RL? Past RL studies of depression have usually used either conditions for reward and punishment learning separately, or they used PE valence within a reward task to determine valence effects (i.e. gaining more or less than expected). While both PE valence as well as valence condition (learning from positive or from negative feedback) have been studied extensively in depression, these valence effects may represent distinct characteristics on learning which need to be disentangled. To shed light on the relevance of the way valence is manipulated in relation to depression, our study includes 3 valence conditions: reward learning, punishment learning and mixed valence learning. Of note, many studies that have examined reward learning used a mixed valence feedback scheme. For reward learning, past studies found reduced learning performance in depression (Forbes, Shaw, & Dahl, 2007; Herzallah et al., 2013; Morris, Bylsma, Yaroslavsky, Kovacs, & Rottenberg, 2015; Robinson, Cools, Carlisi, Sahakian, & Drevets, 2012). Reduced reward learning was related to reduced striatal activation following positive PE (Robinson, Cools, Carlisi, et al., 2012). Further, both reduced striatal activations and reduced reward learning were a predictor of future depressive symptoms during adolescence (Forbes et al., 2007; Morgan, Olino, McMakin, Ryan, & Forbes, 2013). In terms of positive PE effects on learning, a simulation meta-analysis found lower learning rates, while results from the conventional meta-analysis shows only showed lower learning rates at trend (Pike & Robinson, 2022). Estimation tasks found a reduced positivity bias (Garrett et al., 2014; Sharot, Korn, & Dolan, 2011), which may implicate a reduced positivity and confirmation bias in depression during RL as well. Here, learning rates from positive and confirmatory PE are usually higher than from negative and disconfirmatory PE, which leads to higher learning performance compared to symmetric learning rates (Palminteri & Lebreton, 2022). In contrast, a developmental study found no effect of depressive symptoms on learning rate asymmetry (Nussenbaum, Velez, Washington, Hamling, & Hartley, 2022). In brief, both reward condition and positive PE effects suggest that reward learning may be reduced in depression, but the underlying computational mechanisms during RL are less clear. For punishment learning, two studies found higher learning performance relative to that of reward learning (Herzallah et al., 2013; Timmer, Sescousse, Van Der Schaaf, Esselink, & Cools, 2017). An acute lowering of serotonin levels in healthy individuals through depleting its precursor tryptophan lead to enhanced punishment learning, which suggests that low serotonin increase punishment learning, but do not affect reward learning (Cools, Robinson, & Sahakian, 2008; Robinson, Cools, & Sahakian, 2012). Indeed, depressed individuals with SSRI medication showed lower punishment learning relative to unmedicated depression individuals, but no differences in reward leaning (Herzallah et al., 2013). Studies with computational models examined either punishment learning rates or negative PE learning rates. Punishment learning rates showed inconsistent findings, with higher learning rates in depressed and anxious individuals compared to healthy controls (Aylward et al., 2019), or lower learning rates in depressed compared to healthy controls (Mukherjee, Filipowicz, et al., 2020). In terms of negative PE valence, a meta-analysis found higher learning rates in a simulation approach, while results in the conventional approach showed no effect (Pike & Robinson, 2022). Another study did not find changed learning rates in relation to depression, but a higher shift towards perceiving punishments as larger (Brown et al., 2021). To summarize, despite some inconsistencies in the literature, both punishment condition and negative PE effects point towards enhanced punishment learning in depression. Overall, the finding of both reduced reward learning and enhanced punishment learning in depression has led to the idea that depressed individuals differ in their subjective valuation and exhibit an exaggerated loss aversion compared to what prospect theory has established in healthy individuals (Chen, Takahashi, Nakagawa, Inoue, & Kusumi, 2015). There is evidence of learning deviations under both valence conditions in depression, but it is yet unclear whether these emerge in earlier developmental phases. DEP-RQ3: Are there differential effects of the anhedonic and negative affect symptoms clusters during RL? Depression is a highly heterogeneous disorder, which might explain some of the inconsistencies found during learning in depression. A closer look at specific symptom effects might prove useful to shed new light on the effects on RL. Depression is often distinguished by the symptom clusters of anhedonia and negative affect. Anhedonia is the loss of pleasure or lack of reactivity to pleasurable stimuli. While negative affect in depression shows considerable overlap with anxiety disorders, anhedonia is more independent of anxiety, but overlapping considerably with other disorders such as schizophrenia or addiction (Pizzagalli, 2014). Negative affect or distress putatively is influenced by serotonergic function, whereas anhedonia depends more upon dopaminergic function (Nutt, 2008). Therefore, these two symptom clusters within depression may have distinct effects on learning behavior. Since serotonergic function has been shown to specifically target punishment learning (Cools et al., 2008; Robinson, Cools, & Sahakian, 2012), it is possible that enhanced punishment learning can be linked to negative affect. A direct link between negative affect and punishment learning in depression has been previously reported, but remains understudied (Brown et al., 2021). Anhedonia in turn has been studied extensively, and consistently identified related reduced striatal activations during reward anticipation and reward delivery in adolescents and adults (Gradin et al., 2011; Stringaris et al., 2015). Anhedonic symptoms were related to less optimal switching behavior (Pizzagalli, Iosifescu, Hallett, Ratner, & Fava, 2008), reduced learning rates from positive and negative PE (Chase et al., 2010), specifically to reduced reward learning rates (Brown et al., 2021), and to more exploratory decision-making during learning (Harlé, Guo, Zhang, Paulus, & Yu, 2017). Overall, this suggests that anhedonic and negative affect symptoms clusters may have differential effects during RL, and further studies are needed to examine how negative affect modulates RL. Effects of Anxiety Anxiety disorders can differ with regard to the situation (e.g. social anxiety) or the objects (e.g. specific phobia) that cause symptoms such as aversive affective state, somatic stress symptoms and perception of sustained threat. Anxiety is increasingly recognized as developmental disorder (Leonardo & Hen, 2008), and together with depression and obsessive-compulsive disorder, it has been categorized as internalizing disorder, as opposed to externalizing disorders, e.g. ADHD and conduct disorder (Achenbach, 1966; Carver et al., 2017). Among internalizing disorders, conditions can be further divided into fear-based or distress-based symptom clusters (Watson, 2005), however, other symptom clusters have been distinguished as well (e.g. somatic and cognitive anxiety symptoms). Maladaptive uncertainty learning has been found for both the fear-based (Brown, Price, & Dombrovski, 2023) and the distress-based symptom clusters (Hammond, Xu, Ai, & Van Dam, 2023). A study that distinguished by somatic and cognitive symptom clusters found stronger influences of the somatic symptom cluster on RL (Fan, Gershman, & Phelps, 2022; Wise & Dolan, 2020). Neurally, individuals with anxiety disorders exhibited higher noradrenaline levels (Kalk, Nutt, & Lingford-Hughes, 2011) and dysregulated serotonergic modulation (Dayan & Huys, 2009), which may relate to the observed learning disruptions. Further, structural brain differences have been found to mediate the effect of unpredictability experienced during childhood on anxiety and depression symptoms in adults, which suggests that the uncertainty experienced as a child increases the risk for anxiety and depression and may lead to structural brain changes during development (Wang, Cao, Zheng, Chen, & Zhu, 2023). It remains unclear to what extent anxiety symptoms explain learning disruptions across both clinical and normative anxiety symptoms equally, and whether age modulates these effects during development. In this study, we will examine how effects of anxiety symptoms are present during RL under different levels of uncertainty and valence in both clinical and typically developing individuals. ANX-RQ1: What are the effects of anxiety on uncertainty conditions during RL? The current literature suggests that uncertainty processing is central to learning disruptions in anxiety symptoms (Brown et al., 2023). The aversive consequence of uncertainty in anxiety has been shown to impair behavioral learning performance already during low uncertainty, such as expected uncertainty, which is inherent in probabilistic but stable learning (LaFreniere & Newman, 2019). Under high uncertainty, such as in volatile environments, higher anxiety reduced learning performance and predicted aberrant switching behavior, such as reduced win-stay, increased or reduced lose-shift behavior (Dickstein et al., 2010; Hein, de Fockert, & Ruiz, 2021; Huang, Thompson, & Paulus, 2017; Piray, Ly, Roelofs, Cools, & Toni, 2019; Xia, Xu, Yang, Gu, & Zhang, 2021). Maladaptive uncertainty learning may be related to chronic underconfidence in individuals with anxiety and depression, as they showed disproportionally larger updates from low confidence decisions (Katyal, Huys, Dolan, & Fleming, 2023). One study also reported an inverted-u-shaped effect of trait anxiety on learning: both low and high trait anxiety were related to lower behavioral learning performance compared to intermediate trait anxiety (Aberg, Toren, & Paz, 2022). A nonlinear effect of anxiety on learning may be linked to the nonlinear effect of arousal on optimal task engagement that is attributed to noradrenaline function (Eckstein, Guerra-Carrillo, Miller Singley, & Bunge, 2017). Computational studies suggest that individuals with higher trait anxiety or with internalizing symptoms have difficulty adjusting their learning rate to the learning environments. Specifically, higher trait anxiety or internalizing symptoms predicted smaller learning rate adjustments between stable and volatile learning environments, which reflects less flexible learning (Browning, Behrens, Jocham, O’Reilly, & Bishop, 2015; Gagne, Zika, Dayan, & Bishop, 2020). The less flexible updating in relation to high internalizing symptoms was further characterized by a reduced updating after positive PE when action-outcome contingencies where changing throughout the learning block (Gagne et al., 2020). Beyond effects on learning rates, there is evidence that anxiety modulates decision-making during RL. Trait anxiety predicted the strategy by which participants explored alternative choices: higher trait anxiety participants exhibited a shift from value-based exploration towards uncertainty-related exploration (Aberg et al., 2022). This is in line with another study which reported that among individuals that were best described by a value-free choice strategy win-stay-lose-shift, higher state anxiety individuals exploited this value-free strategy more. Overall, anxiety may enhance less optimal choice strategies at the cost of value-based decision-making. To summarize, the currently literature found that anxiety has disruptive effects on learning under both low and high levels of uncertainty. Behavioral studies found reduced learning performance and less optimal switching behavior, but learning performance effects may also be nonlinearly related to anxiety symptoms. Computational studies have focused on anxiety effects between high and low uncertainty, but it is unclear how parameters of each uncertainty level are related to anxiety. These studies reported maladaptive learning rate adjustments to the statistics of the environment, as well as a shift from value-based to value-free decision behavior, but no studies have reported both effects of learning rate and decision behavior. Further, some studies reported combined effects of anxiety and depression, and it is not clear to what extent anxiety and depression show distinct effects on learning. Despite its relevance of anxiety disorders during development (Leonardo & Hen, 2008), it is unknown whether anxiety disrupts learning in children and adolescents similarly as in adults, since the only study with a pediatric sample found no behavioral effects of anxiety and did not apply computational models (Dickstein et al., 2010). This study will examine how anxiety symptoms in children and adolescents, both of clinical and normative symptoms range, show effects of uncertainty that are independent of depressive symptoms. ANX-RQ2: What are the effects of anxiety on valence conditions during RL? The effects of anxiety on uncertainty learning were most consistently found during punishment or threat learning, both behaviorally and computationally (Aberg et al., 2022; Browning et al., 2015; Gagne et al., 2020; LaFreniere & Newman, 2019; Piray et al., 2019). Some studies extended the found effects of uncertainty to both punishment and reward learning (Aberg et al., 2022; Gagne et al., 2020; LaFreniere & Newman, 2019) or did not find effects in either valence condition (Dickstein et al., 2010). Other studies did not examine reward and punishment separately, but they presented both rewards and punishments mixed within learning blocks; these studies found behaviorally reduced learning performance (Xia et al., 2021), and in a simulation meta-analysis higher negative PE learning rates and lower positive PE learning across depression and anxiety (Pike & Robinson, 2022). A recent study using naturalistic learning found that higher negative PE learning rates explained lower and less precise expectations about future academic outcomes and predicted the long-term development of anxiety (Villano et al., 2023). In brief, this new body of literature suggests that higher anxiety symptoms may predict aberrant processing of negative outcomes, both under punishment learning and from negative PE, and that this may be a risk factor for the development of anxiety. It is unknown whether anxiety modulates processing of negative outcomes already in children and adolescents. ANX-RQ3: Are there differential effects of anxiety symptom clusters during RL? The distinctions between fear-based and distress-based anxiety symptoms as well as somatic anxiety and cognitive anxiety symptoms have helped to examine learning disruptions in relation to more specific symptom clusters. Although it has been proposed that fear-based symptoms exert stronger effects on RL (Brown et al., 2023), the current literature suggests that multiple symptom clusters may have effects on RL. Studies that specified effects of symptom clusters found that somatic anxiety reduced uncertainty-directed exploration, whereas cognitive anxiety increased uncertainty-directed exploration (Fan et al., 2022). Similarly, divergent roles of these two symptoms clusters have been found during aversive learning (Wise & Dolan, 2020). The previously described maladaptive learning rate adjustment between stable and volatile environments was attributed to distress-based symptoms, which are less specific to anxiety and likely reflect internalizing symptoms more broadly (Gagne et al., 2020; Hammond et al., 2023). Importantly, the effects of fear-based symptoms have not been examined in these two studies. Therefore, it is unknown whether the effects on learning rate can be better attributed to fear-based symptoms rather than to distress-based symptoms. Since maladaptive learning rate adjustments were also linked to reduced pupil dilation changes, reflecting noradrenaline function and arousal, physiological anxiety symptoms may underlie uncertainty-related effects (Browning et al., 2015). The divergent effects of somatic and cognitive anxiety symptoms on decision behavior are difficult to interpret, but it seems that the role of somatic anxiety symptoms is more in line with the effect of a study that did not distinguish by symptom clusters (Aberg et al., 2022). The current definitions of symptoms clusters vary (eg. fear-based and distress-based symptoms, somatic and cognitive anxiety symptoms), which makes is difficult to integrate findings. Effects of ADHD Attention-deficit/hyperactivity disorder (ADHD) is commonly described as a developmental brain disorder, in which the hypofunctioning of noradrenaline and dopamine lead to reduced cognitive performance such as during RL (Plichta & Scheres, 2014; Ziegler, Pedersen, Mowinckel, & Biele, 2016). Notably, not just neurotransmitter hypofunction such as in ADHD, but also hyperfunction such as during stress (Arnsten, 1999, 2009) has been shown to impair performance. Therefore, an inverted-u-shaped relationship between neurotransmitter function and cognitive performance has been proposed (Biederman & Spencer, 1999; Del Campo, Chamberlain, Sahakian, & Robbins, 2011). Beyond functional differences, the effect of ADHD has been observed in the brain structure, likely as long-term consequence of brain neuroplasticity. Brain regions implicated in reward processing and value representation, such as the ventral striatum and the orbitofrontal cortex, exhibited smaller volumes in individuals with ADHD compared to controls (Carmona et al., 2009; Hesslinger et al., 2002). The well-established biological differences in ADHD have led to effective psychopharmacological treatments in ameliorating ADHD symptoms. Stimulants such as methylphenidate are used to target and normalize dopamine function. Some studies that controlled for methylphenidate use showed that it improved learning performance (Luman, Goos, & Oosterlaan, 2015; Pelham, Milich, & Walker, 1986), which suggests that medication use is a relevant confounding factor for examining effects of ADHD on learning. It remains unclear whether ADHD symptoms explain learning differences only by diagnostic categorization such as reported in case-control studies, or whether these learning effects also extend to subclinical ADHD symptoms in clinical and neurotypically developing groups. Further, it remains unclear whether the effects of ADHD on RL are more pronounced during sensitive developmental periods such as during childhood and adolescence. In this study, we will examine how effects of ADHD symptoms are present during RL under different levels of uncertainty and valence. ADHD-RQ1: What are the effects of ADHD on uncertainty conditions during RL? Several studies have examined the effects of ADHD diagnosis on RL under different levels of uncertainty (Hulsbosch et al., 2021). Under low uncertainty, studies showed mixed findings: some found no effects of ADHD (Luman et al., 2015; Oades & Müller, 1997; Wiesner, Molzow, Prehn-Kristensen, & Baving, 2017), while other found reduced learning (Frank, Santamaria, O’Reilly, & Willcutt, 2007; Gabay, Shahbari-Khateb, & Mendelsohn, 2018; Luman et al., 2021; Shephard, Jackson, & Groom, 2016). Additionally two studies reported slower and more variable reaction times (Frank et al., 2007; Gabay et al., 2018). Under high uncertainty, studies show similarly mixed evidence: two studies found no effects of ADHD (Chantiluke et al., 2015; Finger et al., 2008), one study reported computational effects (Hauser et al., 2014) in terms of a more explorative choice behavior, and two studies showed reduced learning performance in ADHD compared to a control group (Itami & Ca, 2002; Shephard et al., 2016). To summarize, the current literature shows mixed findings during RL under both low and high uncertainty. If an effect was found for ADHD, learning performance was reduced, choices were more erratic, and reaction times were slower and more variable. This study aims to shed new light on whether uncertainty modulates effects of ADHD symptoms during RL, and how these effects can be captured both behaviorally and computationally. ADHD-RQ2: What are the effects of ADHD on valence conditions during RL? No studies have examined the effects of punishment learning in ADHD so far. One study has reported differences in response to positive and negative feedback: negative feedback elicited stronger activations in children with ADHD compared to neurotypical children (Van Meel, Oosterlaan, Heslenfeld, & Sergeant, 2005). Given that the abovementioned brain deficits in ADHD were found in relation to reward learning, processing of negative feedback may be relatively spared and lead to an imbalance towards stronger processing of negative information compared to positive information. It is currently unknown whether the effects of ADHD might be less pronounced during punishment learning and learning from negative PE, compared to learning from reward and positive PE. ADHD-RQ3: Are there differential effects of the symptom clusters hyperactivity/impulsivity and inattentiveness during RL? ADHD consists of a set of symptoms, most notably the symptom clusters hyperactivity/impulsivity and inattentiveness. While previous studies reported correlations of the symptom clusters to learning deficits (Gabay et al., 2018; Luman et al., 2015), it is currently unclear whether the symptom clusters show separable contributions to RL. Computational models may help to identify separable contributions. For example, one theoretical model showed that aberrant learning parameters reproduced impulsive behavior in a delayed response time task (Williams & Dayan, 2005). Impulsive and hyperactive symptoms might be to learning parameters such as learning rate. No study has reported learning rate effects of ADHD, although all theoretical models of ADHD agree that dopamine hypofunction leads to aberrant learning parameters (Frank et al., 2007; Sagvolden, Johansen, Aase, & Russell, 2005; Tripp & Wickens, 2008). Thus, it is unknown whether hyperactivity/impulsivity symptoms and the underlying dopamine hypofunction specifically explain effects on learning rate. The attention deficit in ADHD has not been linked to computational parameters yet. However, noradrenaline function, which is fundamental in attentional processes, has been linked decision behavior during RL (Dubois et al., 2021, 2020; Frank et al., 2007). In contrast, one influential theoretical account of ADHD has linked attention deficits to the dopaminergic mesocortical pathway (Sagvolden et al., 2005). Therefore, it remains unclear whether inattentiveness in ADHD is specifically linked to noradrenergic function and attributed explorative decision behavior, as well as slower and more variable reaction times during RL.

Paired-associate learning efficiency was assessed within four low-SES ethnic populations (black, Chinese-American, Latino-American, and white) as a function of presentation conditions and method of measurement (verbal recall vs. pictorial recognition). A mixed-list paired-associate task was administered individually to 40 second grade children from each group. The results revealed substantial effects for presentation conditions, but not for populations. Nevertheless, the pattern of conditions effects differed as a function of both populations and measurement method. An explanation of the results was discussed in terms of the concept of differential memory coding as a function of specific subject characteristics. (References, tables, and figures are appended.) (Author)

Paired-associate learning efficiency was assessed within four low-SES ethnic populations (black, Chinese-American, Latino-American, and white) as a function of presentation conditions and method of measurement (verbal recall vs. pictorial recognition). A mixed-list paired-associate task was administered individually to 40 second grade children from each group. The results revealed substantial effects for presentation conditions, but not for populations. Nevertheless, the pattern of conditions effects differed as a function of both populations and measurement method. An explanation of the results was discussed in terms of the concept of differential memory coding as a function of specific subject characteristics. (References, tables, and figures are appended.) (Author)

This project seeks to find out whether giving students a choice in some aspect of their learning benefits them, to find out whether having a choice benefits certain groups of students, and to find out whether certain choices benefit certain groups of students.