I stumbled upon the consumer EEG Melon (at Kickstarter). It has three electrodes and is advertised as measuring how "focused" you are. In the FAQ it says:

The Melon headband has three electrodes. Our primary electrode is on the forehead region known as FP1, where Melon can monitor brainwave activity from the prefrontal cortex.

I wonder: What can I measure with such an EEG? What applications would be unlikely?

  • How can "focus" be quantified?
  • Would it be possible to identify (different) states of sleep?
  • Could it be possible to control a cursor/game (left/right movements)?
  • 3
    $\begingroup$ Traditionally, sleep stages (according to Academy of Sleep Medicine criteria) are scored based on C3/C4 and O1/O2. Information from the frontal pole would provide some insight, but, as with any consumer EEG product, you're going to get mainly the muscle activity (EMG)of the forehead anyway. $\endgroup$ Jun 9, 2013 at 23:24
  • 3
    $\begingroup$ I've built a prototype of pong game using another commercial EEG headband, but for the most part the cursor moved in response to facial movements - if I twitch my eyebrows, it will go up. It's very hard to play like that for longer than a minute. I found "attention" metric to be fairly useless and random. $\endgroup$
    – Alex Stone
    Jun 10, 2013 at 10:05
  • 2
    $\begingroup$ FP1 is a prime blink electrode. The prefrontal cortex projects a bit of theta, but it's usually very hard to isolate. On the other hand, you're also going to see plenty of your reference (... and potentially, ground) electrode, and based on their layout, that one's most likely going to be somewhere where you get a lot of alpha ... so this thing is necessarily a mess. I'll see if I can come up with a more detailed answer ... $\endgroup$
    – jona
    Aug 16, 2014 at 19:18

2 Answers 2


Point by point:

The Melon headband has three electrodes. Our primary electrode is on the forehead region known as FP1, where Melon can monitor brainwave activity from the prefrontal cortex.

The problem with this is that electricity doesn't work like that. Current always flows between two points, and our electrodes measure the potential between two locations. After its discovery by Tönnies somewhere in 1938, most EEG amplifiers have been of the differential amplifier design. That means every electrode recording is made up of 3 different sites: the Ground electrode, the Reference, and the target site. What is being reported as activity at FPz is truly ((FPz-Ground) - (Reference-Ground)).
What this entails is that our measurement at the so-called active electrode is not an isolated measure of the activity directly under that electrode alone, but a mix of the activity underlying multiple sites. So what you observe in such a setup is highly dependent on where your Reference is located. I couldn't find where the Melon's Reference sits, but it has to be somewhere on the headband. The headband encircles the skull, and all of the positions allowed by the headband are over cortical areas.
Typical locations for Reference are Cz, Nose or the average, neither of which are possible or meaningful in a 3-electrode headband context. Possibly, the reference is located over the temporal or occipital lobes.

So what the Melon will report as activity at FPz will in truth be a mix between activity at FPz, and activity at some further site also above the cortex. (It is possible they placed their Ground very close to the Reference, which is a very unusual setup ...)

Furthermore, individual electrodes never pick up isolated brain sites. Electrical fields, as those projected by cortical clusters firing in synchrony, are propagated field-like. Every recording site therefore picks up a distinct mix of all of the brain. The activity of sites closer to the electrode will be somewhat stronger since field strength falls off with the square of distance, but the part of the cortex directly under the electrode will still be overwhelmed by the summed activity of everything next to and below it.
All the more so as the strongest activity patterns of the cortex come from the occipital lobes - the so-called alpha rhythm. Alpha usually dominates the cortical activity, as it is a simple synchronized firing.

Moreover, alpha will most likely be artificially reenforced on the Melon by the site of reference, since depending on where the reference is placed, it will necessarily be closer to the occipital lobe than FPz, FPz receives alpha with the inverted polarity of most of the scalp, and subtracting from the residual inverted alpha at FPz the alpha from the reference will lead to substantial alpha effects.

Both of these problems are well known since basically the beginning of the EEG, since Hans Berger himself, discoverer of the EEG, already stumbled on the phenomenon that regardless of where he put his two electrodes (Berger never really got around to appreciating Tönnies' proposal), he'd see very similar activity. Berger's interpretation was that all of the brain partially does basically the same thing, everything in the brain is partially united by one huge common shared oscillatory pattern - alpha.
The activity associated specifically with frontal sites, especially the theta rhythm, is much weaker. Extracting just prefrontal theta activity can be a very hard task, as one can tell from the elaborate methods employed by papers trying to isolate just that, such as this and this.

What especially the second paper will show you however is the most dominant activity at FPz: eye blinks. Here I'm showing you activity from FPz in an experiment of mine.

FPz ERPimage

What you mostly see in the top right image are the red dots. Each of these is a blink by our subject. You see that the dots totally overwhelm the overall activity. The spectrum below gives the same information: it's a simple power-law spectrum with almost all power in very low frequencies, without a specific peak in a frequency range such as theta or alpha.

I wonder: What can I measure with such an EEG? What applications would be unlikely? How can "focus" be quantified?

Interestingly, this might even be possible, since one of the best EEG predictors of focus is alpha amplitude. The distinct prefrontal cortex signals however are not very important here. Rather, the alpha signal is.
Alpha is a very rough indicator of attentiveness, and you'll always get less information out of a single sensor than by simple introspection ("do I feel droopy?"). But you might get a rough correlate of current awareness from looking at alpha power, assuming you position the thing so you get as much alpha out of it as possible (ie., not over FPz). The research in this regard is still in the explorative stage, much before you'd rely on the systems. See for a few examples: a three-channel system like the Melon; a 4-channel system; 14 electrodes.
When I say "it is possible" to do that using a 3-channel system, I do not mean it is possible right now. Nobody has that technology yet. Rather, it might become possible in the future. Although even then I'd rather hope for a larger sensor array. Note also that these are still very rough guesses - they can tell you if you're drowsy or not, if you are currently falling asleep or not. They won't be able to distinguish more subtle states.

Would it be possible to identify (different) states of sleep?

I'll have to pass on this one. I'm not a sleep expert. I think the location of the Melon is wrong though.

Could it be possible to control a cursor/game (left/right movements)?

Possibly - if you place it so that your eye movements reflect as strong positive and negative currents. From brain activity? No. Movement decoding requires much more sensors over very different sites. Motor areas are closer to the middle of the brain, roughly speaking, and that's where you'd need very dense electrode coverage to see anything.

Postscript edit: a few references for decoding movement intentions using (multichannel) EEG. As you see, even with multichannel, high-quality EEG, performance is not especially good. http://journal.frontiersin.org/Journal/10.3389/fnins.2014.00222/abstract http://journal.frontiersin.org/Journal/10.3389/fnhum.2013.00124/full

  • $\begingroup$ +1 great answer... $\endgroup$
    – draks ...
    Aug 21, 2014 at 22:10
  • $\begingroup$ What an amazing answer! $\endgroup$
    – Ana
    Aug 22, 2014 at 14:55

This type of sensor is hardly revolutionary, it seems that the integration and miniaturization are the key differences between say Melon, or this, or a 7-point sensor such as Muse, which supposedly monitors alpha and beta waves, and is already in production. Note that the 10/20 System is as old as I am: 30+ years. Additionally, it is an analog, dumb device, without interpretation or adaptive algorithms.

The most obvious non-listed application for this device is to train yourself to be a better video gamer. Seriously, the pre-frontal cortex, real time executive function, and a "focused state" would have all kinds of useful applications, especially for real-time activities involving executive function: video games, race car driving, piloting under extreme weather, or 1-on-1 sports such as tennis. This may be considered the "future" in some neuroscience textbook like the one copied and pasted above. However, this is very much reality, as the military, MIT, and other serious research institutions have been using this tech for many years for that exact purpose.

I am not a neuroscientist; I have a background in computing and liberal arts/research. However, according to: http://www.brainm.com/help/Positions_and_brain_function.htm Fp1 controls:

  • Logical Attention
  • Orchestrate network
  • Interactions planning
  • Decision making
  • Task completion
  • Working memory

It appears that they are attempting to integrate a tilt sensor, which could be used to control the left-right axis, such as in a video game. The documentation and specifications are lacking or vague.

  • 1
    $\begingroup$ That first page is an impressive array of pseudoscience. $\endgroup$
    – jona
    Aug 20, 2014 at 21:06
  • $\begingroup$ How so? Here is the original source, note that it pertains to EEG related materials, granted that this area of "mind-mapping" is still not entirely explored, however, my time machine is broken, so I imagine this should do: tandfonline.com/doi/abs/10.1300/J184v11n01_03 Note that the source material is cited from this man: neurotherapydallas.com Also, the author of the articles background: en.wikipedia.org/wiki/Thomas_F._Collura $\endgroup$
    – Cbaker510
    Aug 20, 2014 at 22:29
  • 1
    $\begingroup$ To begin with, the assignment of distinct cortical functions to individual 10-20 electrodes is a joke considering the volume conduction underlying the EEG and individual head and brain anatomy variability. $\endgroup$
    – jona
    Aug 20, 2014 at 22:54
  • $\begingroup$ I'm more inclined to believe this published, Neuroscience researcher with a 30+ year track record, over someone with a lack of respect for the basic syntax, and grammar of the English language. You seem to be mindlessly repeating jargon from, I'm guessing, a professor, or lecturer of sorts. In terms of MindMaster, they offer a product that is similar, to the Melon, in that it is useful for taking readings while performing certain tasks. The Melon is, obviously designed for personal development, and not serious Medical research. See: brainmaster.com/page/about $\endgroup$
    – Cbaker510
    Aug 20, 2014 at 23:28
  • 1
    $\begingroup$ Yes, the device Bjorn was asking about is a consumer, mini version of this, that measures the same areas, with a focus on the frontal lobe yes? So, unless you were born with say, FASD, or other form of brain deformity, the output from this data should be similar with this portable, consumer electronic. Note that there is already an existing product with nearly identical function: choosemuse.com $\endgroup$
    – Cbaker510
    Aug 21, 2014 at 0:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.