Monday, October 14, 2013

So I'm sitting in class the other day talking about the nervous system and the brain and how if you chop a brain in half some really interesting things happen when the one of the guys sitting next to me asks the following:

"So what the hell are brain waves, anyway?"

Even if you don't live in a place where a high percentage of the population is composed of hippies, the phrase "brain wave" isn't exactly a member of the "Actual Scientific Terminology"-Club. It makes sense that it could use a solid definition.


So what are they? Put simply:
 Neural oscillations ("brain waves"), are rhythmic, repetitive patterns of neural activity which can be measured by electroencephalography.

Let's break it down.

What is electroencephalography?
Electroencephalography is a method of monitoring neural activity by placing electrodes on someone's head and then recording the electrical activity along their scalp. An electroencephalogram is the record produced by recording the aforementioned electrical activity. Both of these are abbreviated as EEG. Clinically, EEGs are used primarily to diagnose epilepsy, comas, encephalopathies, sleep disorders, and brain death.

Look at this awesome hat. Someone should model this and put it in TF2.

Most EEG signals originate in the neocortex, the outer area of the brain, and generate electrical signals that only last somewhere between 10 to 100 milliseconds. At the moment, EEGs and MEGs (magnetoencephalography) are the only technologies we have that can even catch these signals since they happen so quickly. Other methods, like MRIs and PET scans, are simply too slow, even if they can produce much higher-resolution images of brain activity. It's a trade off.

What are "rhythmic, repetitive patterns of neural activity"?
Remember when we were talking about how neurons work? When a signal travels down a neuron, that's something we can actually measure by sticking an electrode somewhere where it can pick up the electrical signals that are constantly skittering about your scalp. That signal is then carried down a wire, through an amplifier, and then into some sort of recording device.

It's like... sticking a fork into an electrical outlet. Where the fork is the electrode, the arm is the wire, and the rest of the kid sticking it in there is our recording device. It's a way to gain information about what's happening inside the wall without actually breaking down the wall or sticking anything directly into the electrical system. We don't even have to make any holes in your head. (Though we could, if we wanted or needed to.)


Carrying this metaphor forward. Let's say we give little Vanya here a pen and have him rate the level of shock he's receiving over time, he'll move the pen up if he's getting more of a shock, and down if it's less. If that outlet were a spot on your scalp, he might end up drawing something like this:


When we see repeated patterns of activity over time (say, for example, ten spikes in one second - a 10 Hz signal) that's a neural oscillation. Neural oscillations have been observed throughout the body, not just in the brain, and are associated with both individual neurons, as well as groups of the same.

Using EEGs
Because our brains are always turned on and doing things, it can be difficult to tell which patterns of activity are a result of specific stimuli, or caused by a specific mental, sensory, or motor event. So we end up with two options:
  1. We could collect an enormous amount of data through repeated trials in which the same stimulus (something simple, like a flashing light) is presented over and over. We would then time-lock what we've recorded and average the results together to filter out the brain's normal background noise, thus leaving us with the relevant waveform of electrical activity which occurs after the presentation of that stimulus. This is how we calculate Event-Related Potentials (ERPs) and Evoked Potentials (EPs): By gathering hundreds of samples, then math.



  2. Forget about specific events and instead focus on studying the patterns of neural oscillations themselves and seeing whether or not they're correlated with specific mental states. This is the study of brainwaves.
Most of the neural oscillations we observe in EEGs fall between 1 and 20 Hz. i.e. We can see anywhere between 1 and 20 spike/fall cycles per second. (Also, anything below that might not even be real, because of how the technology involved works.) This range is then subdivided into frequency bands which are, in fact, correlated with specific mental states. 

This means that by observing the patterns of neural oscillations, even without any other data, we should be able to gain some idea of a person's mental state during the period in which a given EEG was recorded.

Correlations Between Wave-Types and Mental States
A very basic overview of the EEG bands is as follows.


Delta Waves: Up to 4 Hz. Most often seen in periods of non-REM sleep in adults, and quite often in babies, even when awake, up until kids are about 5 years old.


Theta Rhythms: 4-7 Hz. There are actually two types of Theta rhythms which have been studied in much detail: the hippocampal theta rhythm is a strong oscillation originating in the hippocampus (but which has not been studied in detail in humans, as such testing would be extremely invasive) and cortical theta rhythms, which do not seem to be related to activity in the hippocampus, but are something we can observe via EEG and are seen when people are sleepy, meditating, or sleeping (though not during deep sleep).


Alpha Waves: 7-14 Hz. Seen in both the posterior regions on the head on both sides. When you're awake, they seem to originate in the occipital lobe and are most often seen when people close their eyes, but are not sleepy or physically exhausted. They also show up during REM sleep, but from some place around the front-center.


Mu Rhythms: 8-13 Hz. In contrast the Alpha waves, this one shows up most often in the motor cortex, and is associated with physical rest.


Beta Waves: 15-30 Hz.  Both sides, equal distribution, but most evident in the front of your head. Linked to motor behavior and seen most often during deliberate physical activity and having your eyes open, as well as active thinking and concentration.


Gamma Waves: Ok this one is kind of a mess: 25-100 Hz, and may be related to why we're all... actually conscious. (See, there's this thing called "The Binding Problem" and what is boils down to is this: If everything in our head seems to be working separately, why does it all seem so cohesive to us?) We know the presence of Gamma waves is correlated with conscious activity, and even active sleep, but there's no way to prove it's actually related to subjective consciousness at this point in time.

Practical Applications
As EEG technology becomes less expensive and more widely available, and our understanding of brain states improves, the observation of brain waves is becoming more and more appealing as not only a method of monitoring user experience, but also of actual human-computer interaction as well.

In the latter example, if we have an AI that is capable of identifying possible mental states based on EEG input, and we hook her up to a human wearing an EEG headset, then she would be able to use that information just as she would any other input. This has some obvious implications for medical monitoring technology, but would allow for much more responsive software in general. Video games alone would be an amazing testing environment!

Which actually ties in with the former potential application.

If I wanted to know how engaged people were while doing a certain activity, I might not want to stop them or interrupt the flow of what they were doing by having them talk-aloud or answer questions every few minutes, (more on the strengths and weaknesses of various testing methods later). So, as an alternative, I could just pop that same headset on their head and watch the output to see when/if they got bored, and use that to modify whatever I was testing out at the time.

(I could also totally use it to make sure people we're zoning out when they needed to be focusing, though that could get into some tricky ethical and privacy-related issues at some point, couldn't it?)


1 comments:

  1. you explain very simple and original complicated things. I like how you write. Do you use any resources for inspiration? I will be glad if you share. Thank you

    ReplyDelete