Skip to main body content

ATAB | Boston College

Main Content

file

PNEURAL: Build Your Own Neural Network

Joseph Burdo

Adjunct Professor, Biology

My proposal involves using rigid tubing designed to resemble neurons to build a physical model of a neural network, that I have named PNEURAL (silent P as in pneumatics) to help students grasp the underlying electrical principles of information processing in the brain. Simple components like battery powered solenoids and air pumps controlled with a microcontroller can be used to deliver positive and negative air pressure bursts at a safe level for the tubing that can simulate excitatory and inhibitory potential (voltage) changes within a neuron. If the pressure builds to some critical level in the model neuron, analogous to an action potential in a real neuron, the microcontroller can provide output indicative of that, which could be converted into electronic or physical action. Early in the BI481 Introduction to Neuroscience course that I teach, we discuss numerous key concepts in electrophysiology such as graded potentials, action potentials, length and time constants, excitatory and inhibitory postsynaptic potentials, and temporal and spatial summation. These concepts have historically been difficult for undergraduates to grasp. Student use of the proposed PNEURAL cells would allow them a hands-on way to explore these concepts.

My initial goal through ATIG support is to build two simulators during the summer of 2014, for use in my Fall 2014 BI481 Introduction to Neuroscience course. One is an individual PNEURAL cell, which will allow the students to experiment with the concepts of temporal and spatial summation of excitatory and inhibitory potentials that can lead to action potentials. The second simulator I will construct over the summer will model a basic neural network. Individual PNEURAL cells can be networked together to help the students learn about more complex neurophysiological phenomena.

My longer-term goals (Fall 2014 and Spring 2015) include construction of additional simulators to be used in my courses.  PNEURAL cells could be built to demonstrate how central pattern generator neurons in the spinal cord and brain stem produce rhythmic output such as walking and breathing, how synaptic plasticity in the brain may be the physical substrate for declarative (e.g., facts) and nondeclarative (e.g., muscle) memory, or how red, green and blue specific photo sensors built into PNEURAL cells could model how the retina detects and processes different wavelengths of light. Ultimately, what really excites me for students to build their own cells! I would love to offer a Biology/Computer Science interdisciplinary course for students with an interest in neuroscience and engineering/physical computing, but not necessarily any experience in any of those disciplines.