Pencil-and-Paper Neural Networks: An Undergraduate Laboratory Exercise in Computational Neuroscience

Area of Science:

• Neuroscience
• Computational Neuroscience
• Undergraduate Education

Background:

• Artificial neural networks (ANNs) are foundational to understanding brain function but are often confined to computer science curricula.
• Simple ANN models can bridge cellular neurobiology with higher-level cognitive processes like thought and behavior.

Purpose of the Study:

• To present a simple, paper-based laboratory exercise for constructing and training artificial neural networks.
• To demonstrate how ANNs can illustrate concepts in information processing relevant to neurobiology.
• To assess student learning gains in understanding brain function through practical application.

Main Methods:

• Students manually constructed, trained, and tested artificial neural networks on paper.
• The exercise involved exploring pattern recognition, pattern completion, noise elimination, and stimulus ambiguity.
• Learning was assessed by observing changes in students' written language regarding brain information processing.

Main Results:

• The paper-based ANN exercise effectively engaged students in learning core concepts.
• Students demonstrated an improved ability to articulate information processing in the brain.
• The exercise facilitated a connection between theoretical neurobiology and computational models.

Conclusions:

• Simple, hands-on artificial neural network models provide valuable pedagogical tools for undergraduate neuroscience education.
• Integrating computational exercises enhances students' comprehension of the link between neural mechanisms and cognitive functions.
• This approach makes complex concepts accessible and promotes deeper learning in neuroscience.