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

Scholarly Works (4 results)

  • Thesis
  • Peer Reviewed
UC Irvine Electronic Theses and Dissertations (2018)

The "wisdom of the crowd" phenomenon is when an aggregated group answer to a problem is more accurate than the answer of individuals in the group. Traditionally, aggregation is done with simple statistical methods such as the mean or median of all people's answers. While these methods can be effective, they don't allow us to learn about how people make their decisions. Such cognitive processes are not evident from summary statistics. However, we can use cognitive modeling to infer the mechanisms that lead to a person's decision. In this dissertation, I show how cognitive models can be a powerful tool in accounting for individual differences and biases as well as in helping generate better crowd decisions. I first discuss a hierarchical model for aggregating people's estimates of probabilities, in an environment for which we know the true answers as well as in a predictive environment for which we don't know the truth. I demonstrate how using this model, we can identify the experts in the crowd and account for biases in probability perception. I move on to applying the "wisdom of the crowd" idea to the field of category learning. First, I establish that taking the modal response in a categorization task is an effective and accurate crowd measure. I then implement two prominent cognitive models of categorization into a Bayesian framework and apply them to existing category learning data sets. I show how we can learn about individual differences in strategy use and apply these inferences to development of a novel, latent-mixture cognitive model that allows for people to use multiple types of categorization strategies within the same data set. I conclude by discussing the implications of this research for the study of aggregation in the "wisdom of the crowd" effect and in the study of individual differences in decision-making.

    Cover page: Using Bayesian Cognitive Models in Wisdom of the Crowd ApplicationsCreative Commons 'BY-SA' version 4.0 license
    • Article
    • Peer Reviewed
    UC Irvine Previously Published Works (2018)
    We apply the "wisdom of the crowd" idea to human category learning, using a simple approach that combines people's categorization decisions by taking the majority decision. We first show that the aggregated crowd category learning behavior found by this method performs well, learning categories more quickly than most or all individuals for 28 previously collected datasets. We then extend the approach so that it does not require people to categorize every stimulus. We do this using a model-based method that predicts the categorization behavior people would produce for new stimuli, based on their behavior with observed stimuli, and uses the majority of these predicted decisions. We demonstrate and evaluate the model-based approach in two case studies. In the first, we use the general recognition theory decision-bound model of categorization (Ashby & Townsend, ) to infer each person's decision boundary for two categories of perceptual stimuli, and we use these inferences to make aggregated predictions about new stimuli. In the second, we use the generalized context model exemplar model of categorization (Nosofsky, ) to infer each person's selective attention for face stimuli, and we use these inferences to make aggregated predictions about withheld stimuli. In both case studies, we show that our method successfully predicts the category of unobserved stimuli, and we emphasize that the aggregated crowd decisions arise from psychologically interpretable processes and parameters. We conclude by discussing extensions and potential real-world applications of the approach.
      Cover page: A Model‐Based Approach to the Wisdom of the Crowd in Category Learning
      • Article
      • Peer Reviewed
      Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 38 (2016)
      Different models of categorization are often treated as compet-ing accounts, but specific models are often used to understandindividual differences, by estimating individual-level param-eters. We develop an approach to understanding categoriza-tion that allows for individual differences both between andwithin models, using two prominent categorization models thatmake different theoretical assumptions: the Generalized Con-text Model (GCM) and General Recognition Theory (GRT).We develop a latent-mixture model for inferring whether anindividual uses the GCM or GRT, while simultaneously allow-ing for the use of special-case simpler strategies. The GCMsimple strategies involve attending to a single stimulus dimen-sion, while the GRT simple strategies involve using unidimen-sional decision bounds. Our model also allows for simple con-taminant strategies. We apply the model to four previouslypublished categorization experiments, finding large and inter-pretable individual differences in the use of both models andspecific strategies, depending on the nature of the stimuli andcategory structures.
        Cover page: Inferring Individual Differences Between and Within Exemplar andDecision-Bound Models of Categorization
        • Article
        • Peer Reviewed
        Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 37 (2015)
        Decision-bound models of categorization like General Recognition Theory (GRT: Ashby & Townsend, 1986) assume that people divide a stimulus space into different response regions, associated with different categorization decisions. These models have traditionally been applied to empirical data using standard model-fitting methods like maximum likelihood estimation. We implement the GRT as a Bayesian latent mixture model to infer both qualitative individual differences in the types of decision bounds people use, and quantitative differences in where they place the bounds. We apply this approach to a previous data set with two category structures tested under different cognitive loads. Our results show that different participants categorize by applying diagonal, vertical, or horizontal decision bounds. Various types of contaminant behavior are also found, depending on the category structures and presence or absence of load. We argue that our Bayesian latent mixture framework offers a powerful approach to studying individual differences in categorization.
          Cover page: A Bayesian Latent Mixture Approach to Modeling Individual Differences in Categorization Using General Recognition Theory
          Top