Tanmay Sinha

Last Name

Sinha

First Name

Tanmay

ORCID

0000-0003-3069-2899

Show all metadata

Search Results

Publications 1 - 10 of 33

From problem-solving to sensemaking: A comparative meta-analysis of preparatory approaches for future learning
Sinha, Tanmay; Kapur, Manu (2021)
EdArXiv
Against the backdrop of a growing body of research showing the effectiveness of problem-solving activities followed by instruction (PS-I), we report a meta-analysis of the effectiveness of three broad categories of preparatory activities on future learning from instruction: (a) problem-solving followed by instruction (PS-I), (b) scaffolded problem-solving followed by instruction (+PS-I), or (c) an alternative sensemaking activity followed by instruction (!PS-I)? We examined 118 experimental comparisons spanning 33 articles that compared PS-I with +PS-I and !PS-I designs. Although scaffolding was descriptively associated with a small effect size, there was no significant difference relative to PS-I (Hedge’s g -0.08 [95% CI -0.20, 0.04]). Additionally, PS-I exhibited a non-significant moderate effect (Hedge’s g 0.22 [95% CI -0.06, 0.51]) compared to !PS-I. Bayesian analyses strongly favored the null hypothesis for the comparison of PS-I with +PS-I (suggesting a 99% probability of the difference in effect between these designs being less than 0.2), while it suggested a 40.37% probability of at least a moderate effect favoring PS-I relative to !PS-I. Further, the estimation of true effect sizes after accounting for the publication bias suggested moderate effect sizes in favor of PS-I, when considering both comparison conditions +PS-I (Hedge’s g 0.55) and !PS-I (Hedge’s g 0.64).
Automatic Generation of Socratic Subquestions for Teaching Math Word Problems
Shridhar, Kumar; Mačina, Jakub; El-Assady, Mennatallah; et al. (2022)
Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing
Socratic questioning is an educational method that allows students to discover answers to complex problems by asking them a series of thoughtful questions. Generation of didactically sound questions is challenging, requiring understanding of the reasoning process involved in the problem. We hypothesize that such questioning strategy can not only enhance the human performance, but also assist the math word problem (MWP) solvers.In this work, we explore the ability of large language models (LMs) in generating sequential questions for guiding math word problem-solving. We propose various guided question generation schemes based on input conditioning and reinforcement learning.On both automatic and human quality evaluations, we find that LMs constrained with desirable question properties generate superior questions and improve the overall performance of a math word problem solver. We conduct a preliminary user study to examine the potential value of such question generation models in the education domain. Results suggest that the difficulty level of problems plays an important role in determining whether questioning improves or hinders human performance. We discuss the future of using such questioning strategies in education.
Differential benefits of explicit failure-driven and success-driven scaffolding in problem-solving prior to instruction
Sinha, Tanmay; Kapur, Manu; West, Robert; et al. (2021)
Journal of Educational Psychology
Unscaffolded problem-solving before receiving instruction can give students opportunities to entertain their exploratory hypotheses at the expense of experiencing initial failures. Prior literature has argued for the efficacy of such preparatory activities in preparing students to learn from instruction. Despite growing understanding of the underlying cognitive mechanisms, the pedagogical value of success or failure in initial problem-solving attempts is still unclear. We do not know yet whether some ways of succeeding or failing are more efficacious than others. We report empirical evidence from a classroom intervention (N = 221), where we designed scaffolds to explicitly push student problem-solving toward success via structuring, but also toward failure via problematizing. Our rationale for explicit failure scaffolding was rooted in facilitating problem-space exploration. We subsequently compared the differential preparatory effects of success-driven and failure-driven problem-solving on learning from follow-up instruction. Results suggested that failure-driven scaffolding (nudging students to generate suboptimal solutions) and success-driven scaffolding (nudging students to generate optimal solutions by giving them heuristics with low specificity) had similar outcomes on posttest assessments of conceptual understanding. Students exposed to failure-driven scaffolding, however, demonstrated higher quality of constructive reasoning. These trends were more salient for the learning concept with greater difficulty. (PsycInfo Database Record (c) 2021 APA, all rights reserved)
Which Bright Side to Look at? Cultivating Goal-oriented Shame Regulation Under Challenging Problem-Solving Contexts
Wang, Fan; Sinha, Tanmay; Kapur, Manu (2024)
Proceedings of the 18th International Conference of the Learning Sciences - ICLS 2024
Emotion regulation in authentic learning settings has traditionally been understudied. In the context of problem-solving, understanding how students perceive their emotional experiences and how teachers can effectively facilitate emotion regulation remains unclear. In this study, we introduce an iteratively developed coding framework to examine how students navigate their feelings of shame, a hedonically negative emotion, while attempting problemsolving tasks intentionally designed to be challenging. Our qualitative findings reveal a notable increase in goal-oriented behaviors in the condition where students receive emotion regulation tips to maintain and utilize their shame, compared to conditions where students are provided with hedonic regulation tips or no guidance.
Embodied Agents to Scaffold Data Science Education
Sinha, Tanmay; Malhotra, Shivam (2022)
Lecture Notes in Computer Science ~ Artificial Intelligence in Education. Posters and Late Breaking Results, Workshops and Tutorials, Industry and Innovation Tracks, Practitioners’ and Doctoral Consortium
Problem-Solving Followed by Instruction: State-of-the-Art and New Directions
Sinha, Tanmay (2020)
This thesis makes three inter-related contributions pertinent to problem-solving followed by instruction (PS-I), a learning design with high potential to influence students’ conceptual understanding and transfer. The first contribution aims at expanding the field meta-analytically by investigating factors affecting PS-I efficacy, relative to other widely practiced learning designs such as instruction followed by problem-solving (I-PS), scaffolded problem-solving followed by instruction (+PS-I), and alternative preparatory activity (e.g., worked example, problem-posing) followed by instruction (!PS-I). This strand of research is motivated by the fact that on the one hand, immense variability exists in PS-I implementations over the years, while on the other hand, attempts to synthesize this body of work are mostly qualitative in nature. The conducted meta-analyses therefore fill the research gap of identifying conditions under which preparatory problem-solving effects are best fostered. Results suggest a significant differential advantage of PS-I over I-PS, with additive effects when PS-I is implemented with high fidelity to the design principles of Productive Failure (PF). The learning advantage of PS-I still holds relative to !PS-I (albeit marginally significantly, and with partially additive effects for PF fidelity). However, existing work on scaffolding preparatory problem-solving (primarily towards problem-solving success) has no added learning benefits. The second contribution, via a classroom and follow-up lab study, therefore aims at experimentally investigating deliberate, guided failure as a novel scaffolding strategy within PS-I. This strand of research is motivated by the fact that within PS-I, there is a growing body of research on (i) unsuccessful attempts to explicitly scaffold problem-solving towards success (e.g., via cognitive and/or metacognitive support), and, (ii) mixed attempts to increase failure-likelihood by leaving preparatory problem-solving unscaffolded (that is, by not providing any explicit scaffolds). Creating failure opportunities in a learning design, however, does not automatically imply that students actually experience failure. The conducted experimental studies therefore fill the research gap of explicitly scaffolding preparatory problem-solving towards failure and examining its differential impact on learning. Results based on posttest scores and quality of student reasoning suggest that nudging students towards suboptimal solutions (via explicit failure-driven scaffolds) may lead to stronger conceptual understanding and the ability to transfer than nudging students towards more optimal solutions (via explicit success-driven scaffolds). Additionally, students exposed to both forms of scaffolding perform better than students who receive no explicit scaffolding during the problem-solving phase. The third contribution aims at expanding the explanatory basis of PS-I using process (and several retrospective) measures. This strand of research is motivated by the fact that despite growing work on certain retrospectively-administered measures (e.g., knowledge gap awareness, state curiosity) explaining differential advantages of PS-I across multiple studies, their impact altogether has not been thus far examined within a single study context. Mechanisms such as cognitive dissonance and metacognitive calibration, which are pertinent to the presence of scaffolding, have not yet been explored. Further, former educational psychology work on process measures of affect is limited to I-PS and/or guided discovery style contexts, with learning outcomes assessed independent of formal classroom instruction. The conducted multimodal learning analyses foregrounding the mechanisms of PS-I therefore fill the research gap of assessing fine-grained temporal dynamics of cognition and affect, and triangulating these dynamics with several retrospectively-administered measures. Results suggest abundance of unconventional negative emotions like shame, anger, disgust and contempt that differentially impact learning, and corroborate previously established work on the presence and positive influence of positive emotions like surprise, interest and happiness in generative problem-solving. The emotional roller-coaster ride experienced by students in preparatory problem-solving drives learning benefits. Taken together, the interdisciplinary contributions of this thesis advance state of the art in Educational Psychology, Learning Sciences and Affective Computing by the design of (and investigation of mechanisms underlying) failure-driven and success-driven pedagogy in PS-I.
Comparing the effectiveness of preparatory activities that help undergraduate students learn from instruction
Trninic, Dragan; Sinha, Tanmay; Kapur, Manu (2022)
Learning and Instruction
Students can learn better from instruction after first engaging in activities that prepare them to learn (Kapur, 2016; Loibl, Roll, & Rummel, 2017; Schwartz & Bransford, 1998). In this study, we compare the effectiveness of four activities that prepare university students to learn from instruction. We use productive failure, an established instructional design, as the baseline preparatory condition. In productive failure, students generate solutions to challenging but accessible problems, which serves as preparation for formal instruction. We compare this approach with three alternative preparatory activities: contrasting a correct and an incorrect solution, sensemaking of the correct solution only, and studying a fully worked-out example of the correct solution. Despite the differences in preparatory activities, participants on average performed nearly identically on most of the process and outcome measures. In universities, or with similarly advanced learners, a variety of activities may be equally effective at preparing students to learn from instruction.
Narratives and their effects on students' understanding of biology
Tobler, Samuel; Sinha, Tanmay; Köhler, Katja; et al. (2022)
When Productive Failure Fails
Sinha, Tanmay; Kapur, Manu (2019)
Proceedings of the 41st Annual Meeting of the Cognitive Science Society (CogSci 2019): Creativity + Cognition + Compulation
The Disappearing “Advantages of Abstract Examples in Learning Math”
Trninic, Dragan; Kapur, Manu; Sinha, Tanmay (2019)
Proceedings of the 41st Annual Conference of the Cognitive Science Society

Publications 1 - 10 of 33

Tanmay Sinha

Last Name

First Name

ORCID

Organisational unit

Filters

Search Results