Chapter 1 Advanced Cognitive Modeling

These course notes support the Advanced Cognitive Modeling course taught in the Master’s program in Cognitive Science at Aarhus University. The course represents a journey into how we can understand cognitive processes through the formalization and implementation of hypothesized mechanisms, their testing and validation.

1.1 Course Philosophy and Approach

Advanced cognitive modeling focuses on three interrelated objectives that shape how we approach the modeling of cognitive processes:

The first objective centers on understanding the thought process behind model development. Rather than simply providing a toolbox of existing scripts, we explore how cognitive models are conceptualized and constructed from the ground up. This approach ensures you develop the skills to create novel models for unique research questions.

The second objective emphasizes mastering the Bayesian workflow essential for robust model development. This workflow encompasses simulation design, prior assessment, parameter recovery testing, and thorough model fit evaluation. These skills ensure your models are not just theoretically sound but also practically reliable and generalize way beyond cognitive modeling.

The third objective focuses on developing advanced probabilistic modeling capabilities. Through hands-on experience with Stan, you will learn to implement increasingly sophisticated models while maintaining scientific rigor.

1.2 Course Structure and Learning Path

The course follows a carefully structured progression that builds your modeling capabilities step by step:

After a deepdive into the physics of pizza ovens, we begin with simple scenarios that introduce fundamental modeling concepts. Each subsequent chapter introduces new modeling techniques while building upon previous knowledge. This cumulative approach ensures you develop a deep understanding of both basic principles and advanced applications.

The chapters include theoretical discussions paired with practical coding exercises. During practical sessions, we work with real datasets, design models collaboratively, and implement them using modern statistical tools. This hands-on approach provides ample opportunity for questions and exploration.

The course schedule maintains flexibility to adapt to the collective learning pace of each cohort. While we have clear learning objectives, we ensure everyone develops a solid foundation before moving to more advanced topics.

1.3 Prerequisites and Preparation

To make the most of this course, students should prepare their technical environment and review fundamental concepts:

Software Requirements: - R (version 4.4 or above) - RStudio (version 2024.12.0 or above) - brms package with proper configuration - cmdstanr package with complete installation

Technical Prerequisites: - Working knowledge of R programming - Basic understanding of Bayesian statistics - Familiarity with cognitive science fundamentals

Additional Resources: - Introduction to R and tidyverse: https://r4ds.had.co.nz/ - A condensed Bayesian statistics primer (by Chris Cox and me): https://4ccoxau.github.io/PriorsWorkshop/

1.4 Course Resources

The course materials include: - Lecture notes and presentations - Practical exercise guides - Example code and solutions - Additional readings and references

For comprehensive information: - Course syllabus: [TBA] - Lecture videos: [TBA]

1.5 About These Notes

These notes represent an evolving resource that builds upon previous iterations of the course while incorporating new developments in the field. They are designed to serve both as a learning guide during the course and as a reference for your future research endeavors.

knitr::opts_chunk$set(
  warning = FALSE,        # Suppress warnings
  message = FALSE,        # Suppress package loading messages
  echo = TRUE,           # Show R code
  fig.width = 8,         # Set default figure width
  fig.height = 5,        # Set default figure height
  fig.align = 'center',  # Center figures
  out.width = "80%",     # Make figures 80% of text width
  dpi = 300             # Set high resolution for figures
)

04-ModelQualityChecks