Significant New Breakthrough for Handicapped Individuals’ Communication

This research presents a practical technology to allow paralyzed individuals who have lost the ability to communicate again, with a high degree of accuracy.

It has been understood for some time that after paralysis, the neural codes for gross movements, such as grasping objects or moving a computer mouse, are still encoded in the brain. But the ability to access and utilize fine motor skills presents a significant breakthrough in the quality of life, independence, and communication for paralyzed individuals.

According to a recent paper published in Nature, researchers have discovered that the neural codes for highly dexterous motor skills, such as handwriting or typing, remain in the brain even after paralysis. The brain remembers how to write or type, even if the connections to perform that function are broken. With new computer technologies, called Brain-Computer Interfaces (BCI), it is possible to connect to this neural network, allowing paralyzed individuals direct written communication. The implications for individuals who have lost the ability to move or speak is life-changing. 

Brain-Computer Connections

Researchers developed an intracortical BCI (meaning a computer connection directly to the brain) that decodes handwriting attempts and translates them into real-time text. The results show participants achieving typing speeds comparable to non-handicapped averages and a great degree of accuracy. One participant, who was paalyzed with a spinal cord injury, “achieved typing speeds of 90 characters per minute with 94.1% raw accuracy”. 

These performances exceed previous BCI technologies, presenting a new paradigm of freedom for handicapped individuals. The results open new opportunities to approach BCIs as a feasible method to accurately decode neural activity related to rapid dexterous movement, even years after paralysis. 

Research Methodology

In testing the brain-computer interface technology, researchers requested participants to pretend they were writing as if they were not paralyzed, and imagine grasping a pen. They then collected data while the study participants attempted to both copy sentences and to answer open-ended questions at their own pace.

As a part of this, the researchers developed new decoding methods to overcome two key challenges:

  1. The participant’s hands never moved, so there was no perceptible activity.
  2. Limited amount of decoder training data had the potential to reduce accuracy.

The techniques used to overcome these hurdles adapt neural training data from automatic speech recognition to work with neural activity. The application of these techniques, beyond handwriting or typing, will work for any sequential behavior that cannot be directly observed. This includes speaking or writing for those who can no longer speak. 

Accuracy and Speed

In both the sentence copying test and in self-responses, accuracy was extremely high. When copying sentences, the error rate was 0.17% character error rate. When writing self-generated sentences with on-screen prompts, the average typing speed was 73.8 characters per minute with 8.54% character error rate. This more than doubles previous BCI models, which use a point-and-click model of typing and peak at 40 characters per minute. 

Researchers looked at why the results were nearly twice as fast typing speed with similar levels of accuracy. They theorize this is because letters’ shapes are more distinctive than clicking motions or lines. This was confirmed when looking at neuronal activity. 

The neural distance of activity for letters was almost twice that of point and click. This suggests that the varying patterns of movement such as handwritten letters make it easier for a decoder to distinguish differences than in point-to-point movements. 

Further Developments

This system is an initial model. It proves that the concept of Brain-Computer Integration is possible for high performance handwriting and fine motor communication. While it is not yet clinically viable, the potential for implementation and further development is enormous.

Further refinements to this system can include:

  • Expanded character sets to include capital letters.
  • Ability to edit or delete text.
  • Shorter calibration time with a high degree of accuracy.

The researchers have made their dataset available to facilitate further investigation and refinement by other researchers. The core data set includes more than 40,000 characters collected over 10 days. The hope is that this will provide a basis for testing and developing new decoding approaches to further enhance the technology. 


The ability to communicate wants, needs, feelings, and thoughts is a fundamental human need. This research presents a practical technology to allow paralyzed individuals who have lost the ability to communicate again, with a high degree of accuracy. The typing speeds and accuracy achieved are comparable to average phone typing speeds for non-handicapped individuals. 

This utilization of BCI for high-performance handwriting and communication will continue to be refined and developed until it is a clinically viable option to open new paradigms and build bridges of communication. This real-time BCI technology has implications across medicine and science, but the most important outcome is the improvement in quality of life of handicapped individuals. 


The Importance of Data Validation for Data Science

Back to Technology

Guide to Remote Learning for Parents