BMI for coding at the speed of thought

Let us layout a use for brain-machine interfaces (BMIs) that could have broad impact. The model project at this time is a BMI for controlling Vim, an extensible text editing environment used by software engineers. The motto of Vim is ‘precision editing at the speed of thought.’ I see Vim’s existing UI is uniquely ready to accept BMI input - literally delivering on the promise of the software’s motto.

§Specs

Vim uses a modal mechanism to contextualize actions and the range of text they act upon. The major modes are

  • selection (cursor)
  • visual selection
  • insertion
  • shell (CLI)
  • normal (command ready)

The number of gestures possible with a BMI will determine how many functions in each mode can be performed via brain gestures. Certainly a minimum number are needed for a useful interface. The speed at which the BMI can process gesture sequences could allow for a larger set of gestures via rapid repeating. The secondary objective is to sermons if using BMI gestures feels more natural than typing the keys related to Vim commands. The command keys in each mode are abstract and often non-intuitive. Regardless of how natural a BMI may feel, we must also determine how fast the BMI is relative to keyboard input.

§Meeting the Spec

First, let’s ask: How many degrees of measurement exist on a commercially available EEGs? (Or with any from kickstarters?) In part, the question addresses how many ‘cognitive gestures’ can be measured? When people use a BMI (brain machine interface) to control a mouse cursor - up, down, left, right, click1, click2, etc. These are all performed by doing a certain kind of ‘thought’ (gesture) that the EEG detects.

§Slight of Mind

EEGs have a training mode for a given user, where the person is instructed to do some things (gestures) and the BMI collects EEG data. The data from each of the things they are instructed to do is correlated into a brain ‘gesture’. If gestures can be performed simultaneously then the set of available gesture is the number of all permutations. Certain kinds of mental dexterity may be unavailable to most people. The median populace may only be able to control an even further limited set.

§Repetitive Stress Injuries

I am concerned about one aspect of the BMI: associating a certain pattern of cognition (a portion of wholistic brain use) with a ‘brain gesture’. Think of repetitive stress injuries. RSI happens because we use our body in a repeated, unnatural, way. For example we built a computer keyboard (machine interface) that isn’t ergonomic, and prolonged use of a keyboard causes RSI. I’m worried this will hurt people the same way that certain video game uses (WoW) get people so reward-addicted that they will malnourish themselves in an Internet cafe. A skillful BMI application has a similar problem on the horizon.

§Ethics

Morally speaking, we may not concern ourselves with worries over addictiveness nor RSI. Rather, addictive behavior is a lifestyle/situational choice that pre-exists the BMI. As designers, we need attention to the biophysical impact of this BMI use. Rather, we will not be paralyzed by this attentiveness, and will actively discuss the impacts with users during the development process.


§Reviews

§Emotiv EPOC EEG

This device is the Emotiv EPOC. 14 channel, which pragmatically means 14 ‘bits’ or actions your mind can signal to the machine. $300 or $750 for the research-model that gives unrestricted access to the live EEG waveforms for each channel. I wonder if, with the EPOC, you can activate 2 or more channels at once? 2^14 is over 16,000 combinations. If a person can have such full brain control. Enough for any purpose for now. But can people have that kind of control? What mental and emotional states are associated with those channels? By creating activity in each, are we ‘simulating’ anger or other unhealthy states.

§Necomimi Affective Expression

‘Necomimi’ is a hair band for girls that uses brain waves to control some neko ears on a headband.

§Games

A California based company advertised a gaming device that used a mix of hand gestures and brain waves to drive a ball across a very grated surface


© 2001-2023. All rights reserved.

Powered by Hydejack v9.1.6