Skip to main content

Web Search and Neuroscience

Recently, I realized the similarities of how hub and authority search algorithms mimic how real-life neurons function. This is no coincidence as one of my cognitive science professors mentioned to me that search algorithms for the web were most likely inspired by real neuroscience.

In the EdX lecture videos, Professor Kleinberg discusses an analogy to provide an intuitive sense for how the hubs and authorities search algorithm works. The example compared the algorithm to gathering restaurant recommendations, where when searching for a good restaurant you consult your friends (the hubs) and get various restaurant names (the authorities). Furthermore if some of your friends are restaurant connoisseurs you will weight their recommendations more. While this analogy does not describe every detail of the search algorithm, it provides an intuitive methodology for understanding how neurons work.


A neuron is comprised of several components including an axon, dendrite, body. The body can send an electric charge through the axon called an action potential. When neurons communicate, an axon of one neuron will receive input from the dendrites of other neurons. The neurons will fire depending on what is being received by the dendrites. Furthermore, some dendrites to a listened more than others. In a mathematical sense, the neurons are taking in various dendrite inputs, applying certain weights to them, and then passing them into a mathematical function, such that if the output crosses a specific threshold value, will cause the neuron to fire.

A key phrase in the above paragraph is “some dendrites are listened to more than others.” This fact ties back to the restaurant analogy, where one will weight the recommendations from restaurant connoisseurs higher than others. While it is difficult to map hubs and authorities directly to neuron counterparts, the main similarity lies in the behavior of how the algorithm and neurons using a weighting of inputs in order to list a search or fire an action potential. I find the crossover between neurological and computer sciences fascinating and hope to encounter more neuroscience inspired algorithms.


Leave a Reply

Blogging Calendar

October 2019
« Sep   Nov »