Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SequentialSamplingModels.jl: Simulating and Evaluating Cognitive Models of Response Times in Julia

Kianté Fernandez, Dominique Makowski, Christopher Fisher

TL;DR

This paper introduces SequentialSamplingModels.jl (SSM.jl), a Julia package that unifies simulation, visualization, and Bayesian inference for sequential sampling models of response times within the Julia ecosystem. It presents a core design with a mixed-output abstract type and Distributions.jl integration, enabling DDM, RDM, and LBA implementations for sampling, plotting, and likelihood-based inference via Turing.jl. Through concrete examples, the work demonstrates end-to-end workflows for simulating RT distributions, visualizing accumulation dynamics, and performing Bayesian model analysis, highlighting practical integration with Julia tools and ecosystems. The authors position SSM.jl as a turnkey platform to lower barriers to applying SSMs in Julia and to enable scalable, likelihood-based or likelihood-free inference for complex models.

Abstract

Sequential sampling models (SSMs) are a widely used framework describing decision-making as a stochastic, dynamic process of evidence accumulation. SSMs popularity across cognitive science has driven the development of various software packages that lower the barrier for simulating, estimating, and comparing existing SSMs. Here, we present a software tool, SequentialSamplingModels.jl (SSM.jl), designed to make SSM simulations more accessible to Julia users, and to integrate with the Julia ecosystem. We demonstrate the basic use of SSM.jl for simulation, plotting, and Bayesian inference.

SequentialSamplingModels.jl: Simulating and Evaluating Cognitive Models of Response Times in Julia

TL;DR

This paper introduces SequentialSamplingModels.jl (SSM.jl), a Julia package that unifies simulation, visualization, and Bayesian inference for sequential sampling models of response times within the Julia ecosystem. It presents a core design with a mixed-output abstract type and Distributions.jl integration, enabling DDM, RDM, and LBA implementations for sampling, plotting, and likelihood-based inference via Turing.jl. Through concrete examples, the work demonstrates end-to-end workflows for simulating RT distributions, visualizing accumulation dynamics, and performing Bayesian model analysis, highlighting practical integration with Julia tools and ecosystems. The authors position SSM.jl as a turnkey platform to lower barriers to applying SSMs in Julia and to enable scalable, likelihood-based or likelihood-free inference for complex models.

Abstract

Sequential sampling models (SSMs) are a widely used framework describing decision-making as a stochastic, dynamic process of evidence accumulation. SSMs popularity across cognitive science has driven the development of various software packages that lower the barrier for simulating, estimating, and comparing existing SSMs. Here, we present a software tool, SequentialSamplingModels.jl (SSM.jl), designed to make SSM simulations more accessible to Julia users, and to integrate with the Julia ecosystem. We demonstrate the basic use of SSM.jl for simulation, plotting, and Bayesian inference.

Paper Structure

This paper contains 7 sections, 4 figures.

Table of Contents

  1. Introduction
  2. SequentialSamplingModels.jl
  3. Availability, Development, and Documentation
  4. Example one: simulating choices and response times
  5. Example two: plotting RT-choice distributions and simulated traces
  6. Example three: Bayesian inference
  7. Future Applications

Figures (4)

  • Figure 1: An example of applying the sequential sampling modeling framework to a choice between two pizzas. The decision-maker samples evidence in favor of both options until reaching a decision boundary.
  • Figure 2: RT distributions of the racing diffusion model based on Code Block \ref{['lst:example2']}
  • Figure 3: Five traces of the racing diffusion model based on Code Block \ref{['lst:example2']}. On each simulations, the accumulators race independently until the evidence of the fastest accumulator reaches its threshold (horizontal, dashed line)
  • Figure 4: Posterior distributions for the parameters of the Linear Ballistic Accumulator model based on Block \ref{['lst:example3']}. MCMC trace plots were removed due to space limitations.