Updates in Honors

Fridays, 9:05-12:05, Bousfield A105
Instructor: Jim Magnuson <james.magnuson@uconn.edu>
Undergraduate section: PSYC 3884-002
Graduate section: PSYC 5570-002

This class is a hands-on introduction to several kinds of computational models used in the cognitive and neural sciences. Computational models are tools for testing and refining theories by forcing us to make our assumptions about inputs, representations, and/or processes underlying some cognitive domain concrete. Once a theory proposes even a few interacting components, it may be hard to determine the specific predictions that result without creating a model capable of simulating the interactions. 

 

Each week, I will provide you with a 'jupyter notebook' -- like a webpage where you can access the simulator for that week and run simulations without programming (though you can also explore and modify the code). We will explore a variety of modeling approaches, including: mathematical models (equations that correctly capture aspects of data patterns), agent-based models (where we simulate interactions among many ‘individuals’ following similar rules or algorithms), network science models (which can characterize local and global aspects of connected systems, from neurons to friendships, and interactions among elements), and neural network models (dynamic networks of artificial neurons that can be trained to simulate cognitive processes). We will compare models to a variety of theories in domains such as learning and memory, vision, and language. We will explore ways to analyze models and link them to theories, behavior, and neural activity. 

 

What you will learn in this course is directly applicable to modeling cognitive and/or neural processes, and could extend to related areas such as most domains of psychology, neuroscience, linguistics, sociology, etc. More importantly, you will learn to approach problems from a computational perspective. This perspective focuses on developing causal explanations and testing them via simulation (or sometimes just calculation). It emphasizes the potential for complexity to emerge from interactions among simple elements. You will develop critical thinking skills that are transferable to any domain. Also, while the course does not require any prior programming experience, you will have the opportunity to acquire some skills in programming / data science / machine learning.

 

There are grad and undergrad sections. There's a joint meeting, but then additional expectations and opportunities for additional tutoring for grad students who want to delve more deeply into how to modify models or develop a model relevant for their own research.