Skip to main content
eScholarship
Open Access Publications from the University of California

Scholarly Works (24 results)

  • Article
Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 40 (2018)
The Predictive Performance Equation (PPE) is a mathematical model of learning and retention that uses regularities seenin human learning to predict future performance (Walsh, Gluck, Gunzelmann, Jastrzembski, & Krusmark, in press). Togenerate predictions, PPEs free parameters must be calibrated to historical performance data, with generally inaccuratepredictions for initial performance events. Prior research (Collins, Gluck, Walsh, Krusmark & Gunzelmann, 2016; Collins,Gluck, & Walsh, 2017) explored the use of aggregate prior data to inform PPEs free parameters for initial performance pre-dictions. Here we report an extension of our prior research, using Bayesian hierarchical modeling to integrate informationfrom the historical performance of both prior data and an individual student to generate future performance predictionsover an entire instructional period. Data are sourced from DataShop an online educational data repository (Koedinger et al.2010). Adding Bayesian hierarchical modeling to the PPE will improve PPEs application in both education and trainingscenarios.
    Cover page: Using Bayesian Hierarchical Modeling and DataShop to Inform Parameter Estimation with the Predictive Performance Equation
    • Article
    • Peer Reviewed
    Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 41 (2019)
    The initial performance of individuals is often difficult for models of learning and retention to predict. One such modelis the predictive performance equation (PPE) a mathematical model of learning and retention that uses regularities seenin human learning to predict future performance.To generate predictions, PPEs free parameters must be calibrated to aminimum amount of historical performance data, preventing valid predictions for initial learning events.Prior research(Collins, Gluck, Walsh, Krusmark & Gunzelmann, 2016; Collins, Gluck, & Walsh, 2017), has shown that the generaliza-tion of best fitting parameters from prior data can improve initial performance predictions.Here we build on that research,using Bayesian hierarchical modeling to estimate free parameters from various sources of prior data. Bayesian hierarchalmodeling allows an opportunity to improve and add structure to the parameters used by PPE, improving its application tocognitive technology in education and training.
      Cover page: Integrating Methods to Improve Model-based Performance Prediction
      • Thesis
      • Peer Reviewed
      UCLA Electronic Theses and Dissertations (2018)

      Endocrine-disrupting compounds (EDCs) are found in numerous products and due to their structural similarity to endogenous hormones, they can interfere with many aspects of the endocrine system. Bisphenol A (BPA) is one such EDC commonly used as an intermediate in the production of several plastic polymers, including polycarbonate and epoxy resins. Here, we examined the role of BPA as a reproductive toxicant using three strains of Caenorhabditis elegans: N2 (wild-type), Strl-1 F52F12.7(ok3347) I (StAR) and tspo-1 C41G7.9(tm5526) (TSPO). The StAR and TSPO mutants have a mutation that inactivated the StAR or TSPO mitochondrial membrane transporter respectively. A combination of one cholesterol concentration (0 μg/ml cholesterol, 0.5 μg/ml cholesterol or 5 μg/ml cholesterol) with a BPA concentration (0 �M BPA, 100 �M BPa or 500 �M BPA) allowed us to analyze how varying cholesterol and BPA concentrations affect fertility, the germline and cholesterol uptake in the mitochondria. We found that cholesterol antagonizes the reproductive toxicity of BPA in wild-type (N2) C. elegans and BPA may induce its reproductive toxicity through the StAR transporter as the StAR mutants did not show signs of reproductive toxicity with increased exposure to BPA. These results indicate that cholesterol rescues the reproductive effects of BPA and that StAR may be an important component of this mechanism.

        Cover page: Bisphenol A induces reproductive toxicity by antagonizing cholesterol uptake at the StAR mitochondrial transmembrane transporter in C. elegans
        • Thesis
        • Peer Reviewed
        UCLA Electronic Theses and Dissertations (2018)

        The potential health impact of Bisphenol A (BPA) exposure remains a hotly debated issue. To explore the molecular actions of BPA in a data-driven manner, which may inform on pathway perturbations and health consequences, a systematic analysis of publicly available transcriptomic data was conducted. Species- and tissue-specific differential gene expression, biological pathways and processes, as well as transcriptional were reviewed across 19 datasets of BPA exposed tissues from rat, mouse, and human studies. A comparison of the data within and between species or tissues was conducted, revealing tissue-/species-specific molecular perturbation such as IL7 in human uterine tissue, or Wnt4 in mouse testis, or Tgfbr2 in rat breast tissues in response to exposure to BPA. In addition to such unique annotations, there were a few responses that were shared between tissues and between species such as common CYP26b1 or ALDH1a7 gene dysregulation, oxidative stress and insulin sensitivity related pathways in liver, and cytokine receptor activity in breast tissue, with differentially regulated genes such as IL1A being common across species that may point to a more conserved molecular response to BPA. In general, our results not only demonstrate complex species- and tissue-specific molecular activities of BPA, but also highlight some consistent transcription factors like SUZ12 and other PcG associated transcription factors as potential BPA targets across tissues and species. It is also apparent that there is a need for more comprehensive and systematic molecular investigations of BPA to better understand its mechanisms of action.

          Cover page: Multi-species, Multi-tissue Meta-analysis of Bisphenol A Transcriptome Studies Reveals Species-and Tissue-specific Molecular Perturbations
          • Article
          • Peer Reviewed
          Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 37 (2015)
          In multiple-cue probabilistic inference, people choose between alternatives based on several cues, each of which is differentially associated with an alternative’s overall value. Various strategies have been proposed for probabilistic inference. These include heuristics, simple strategies that ignore part of the available information to make decisions more quickly and with less effort. Heuristic models seek to explain the sequence of cognitive events that occur as people make decisions. Validating these models involves evaluating their predictions concerning both outcomes and process measures. In this study, we gathered verbal protocols from participants as they performed multiple-cue probabilistic inference. We find converging evidence across decisions, search behavior, and verbal reports that many participants use a simplifying heuristic, take-the-best. These results provide novel evidence for take-the-best as a process model of human decision behavior in multiple-cue probabilistic inference.
            Cover page: Verbal Reports Reveal Strategies in Multiple-Cue Probabilistic Inference
            • Article
            • Peer Reviewed
            Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 44 (2022)
            Adaptive scheduling systems aim to estimate the ability of an individual in order to prescribe a personalized training schedule. These adaptive systems are often founded on regularities of human memory such as a learning, forgetting, and the spacing effect. One such model which has been developed to both account for regularities of memory and be used in applied contexts is the Predictive Performance Equation (PPE). One limitation of the PPE is that it is only able to account for and incorporate information about a participant’s accuracy on a task and cannot take into account additional performance measures such as reaction time. To expand the PPE, we propose a simple extension to the model, allowing it to account for both accuracy and reaction time measures. Our paper reports the extension to the PPE as well as a formal model comparison to another model of learning and retention (Pavlik and Anderson, 2005). The results of our model comparison reveal that the extended PPE can both better account and predict an individual’s performance than Pavlik and Anderson (2005) model.
              Cover page: Extending the Predictive Performance Equation to Account for Multivariate PerformanceCreative Commons 'BY' version 4.0 license
              • Article
              • Peer Reviewed
              Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 43 (2021)
              Machine learning offers techniques for predicting behavior, but has limited capability to use behavior, context, and domain knowledge to infer intentions and predict behavior trajectory (i.e., sequence of directed actions towards reaching goal) of humans in real time to assist in strategic planning. Some existing models of human goal inference (Zhi-Xuan, Mann, Silver, Tenenbaum, & Manshinghka, 2020) possess some of these capabilities, however, these models are typically evaluated in toy worlds, which limits our understanding of how these approaches would generalize to real-world domains. Here, we introduce a novel integration of Bayesian goal inference and a deep learning model used for joint estimation (OpenPose) to address this generalizability problem and evaluate how this integrated approach can infer the goals of humans in real-world environments. This exploration provides an opportunity to evaluate capabilities required for real-time goal inference in real world environments and can highlight benefits and limitations informing future research.
                Cover page: Exploring Online Goal Inference in Real World EnvironmentsCreative Commons 'BY' version 4.0 license
                • Article
                • Peer Reviewed
                Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 42 (2020)
                Human learning has been characterized by three robust effects (i.e.power law of learning, power law of decay, and spacing), whichhave been validated across multiple domains and time intervals. Toaccount for these different effects mathematical model of learningand retention have been developed. These models hold a great dealof potential for application a wide range of educational and trainingscenarios. However, many models are not validated according fortheir ability to make accurate predictions of human performance.The predictive demand of these models is made increasinglycomplex by the needs of training domain, needing both to predictboth skill decay and reacquisition from little historical data. In thispaper, we examine the predictive capability of the PredictivePerformance Equation (PPE) implemented in a Bayesianhierarchical model. Through a comparison of two Bayesianhierarchical models we show how hierarchical model fit to aparticipant’s performance across a set of items compared to only asingle item improves PPE’s predictive accuracy of both skill decayand reacquisition over multiple learning schedules
                  Cover page: Improving Predictive Accuracy of Models of Learning and Retention Through BayesianHierarchical Modeling: An Exploration with the Predictive Performance EquationCreative Commons 'BY' version 4.0 license
                  • Article
                  • Peer Reviewed
                  Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 38 (2016)
                  The predictive performance equation (PPE) is a mathematical model of learning and retention that attempts to capitalize on the regularities seen in human learning to predict future performance. To generate predictions, PPE’s free parameters must be calibrated to a minimum amount of historical performance data, leaving PPE unable to generate valid predictions for initial learning events. We examined the feasibility of using the data from other individuals, who performed the same task in the past, to inform PPE’s free parameters for new individuals (prior-informed predictions). This approach could enable earlier and more accurate performance predictions. To assess the predictive validity of this methodology, the accuracy of PPE’s individualized and prior-informed predictions before the point in time where PPE can be fully calibrated using an individual’s unique performance history. Our results show that the prior data can be used to inform PPE’s free parameters, allowing earlier performance predictions to be made.
                    Cover page: Using Prior Data to Inform Model Parameters in the Predictive Performance Equation
                    • Article
                    • Peer Reviewed
                    Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 39 (2017)
                    The predictive performance equation (PPE) is a mathematicalmodel of learning and retention that uses regularities seen inhuman learning to predict future performance. Previous research(Collins, Gluck, Walsh Krusmark & Gunzelmann, , 2016) foundthat prior data could be used to inform PPE’s free parameterswhen generating predictions of a group’s aggregate performance,allowing for more accurate initial performance predictions. Herewe investigate an extension of this methodology to predictperformance of individuals, rather than aggregate samples. Thispaper documents the results of that investigation, which is on thecritical path to the use of this cognitive technology in educationand training.
                      Cover page: Using Prior Data to Inform Initial Performance Predictions of Individual Students
                      Top