Alterations to the Neural Code Unlock Our "Inner Learning"

Alterations to the Neural Code Unlock Our "Inner Learning" ...

Our brains are extremely adept at learning patterns in the world and making sense of them. Throughout our lives, the brain continuously learns and adapts, and so many neurons that help develop learned behaviors, such as the daily walk to work, are changing.

These changes in the neural code have similarities to how languages change gradually over time, while faithfully communicating common ideas and concepts.

This representational drift occurs without any noticeable improvement in behaviour or task performance. Rather, everything appears to be routine and steady, e.g. you follow the same path to work, make the same plan, and take the same steps, but patterns of neural activity in certain areas of the brain are changing. A new study, published in the journalPNAS, proposes how the brain stays stable despite neural code changes.

Dr Michael E. RuleandDr Timothy O''Leary, a Cambridge neuroscientist, claims that neurons (the cells that make your brain work) may detect when some of their inputs change, and adjust the strength of influence that one neuron has on another, in order to compensate, thus supporting a form of internal learning.

According to Dr Rule, a Leverhulme Early Career Fellow in Engineering, these changes in the neural code have similarities to how languages evolve gradually over time.

While some parts of the brain are plastic, and they have evolved rapidly, others have long-term stability. So how do neural circuits talk to each other without constantly having to re-learn the lessons they have already learned? Even brain-machine interfaces which are becoming increasingly used as assisted living devices for individuals with cognitive or physical disabilities must deal with drift.

The researchers claim that homeostatic pathways within single cells can assist the brain to see themselves as it changes, and that internally-generated signals help stable neural populations learn how to identify the unstable ones. This theory combines knowledge and data to demonstrate living brain activity.

Researchers have demonstrated how engineers are currently developing machine learning algorithms to monitor neural representations as they change automatically the process progresses. Currently, they believe that something similar to these algorithms might also be at work in the brain, following well-known learning rules and homeostatic processes.

Dr Rule said that representational drift occurs in the hippocampus the part of the brain that has a major role in learning and memory, and in the parietal cortex the area responsible for sensory perception and integration. We propose several additional measures that might assist make this plasticity compatible with long-term stability through the brain.

According to Dr OLeary, an associate professor in the Department of Engineering, a study highlights the possibility that continuous learning might be triggered by drift.

According to the narrator, artificial intelligence poses a huge unanswered challenge, the problem of constructing algorithms that can learn continuously without corrupting previously learned information. This technique is a step in the direction of discovering algorithms that can do the same.

You may also like: