The problem with the Openvibe solution is that Openvibe still works only with EPOC SDK Research Edition Version 1.0.0.4.  Since Emotiv actively supports only SDK Version 2 and later, the Openbvibe solution will not work unless you have the older SDK.  In my case, my EPOC shipped with 2.0.0.1, and I have been unable to get the older SDK from Emotiv.   I'm planning to try to work around this over the summer.

As far as I know the system transfers data via Bluetooth, which is to not precise enough to asset ERP data.

The sampling rate for the Emotiv looks suboptimal for ERP purposes. 256Hz is probably the absolute minimum that one would use.

The Emotiv EPOC is suboptimal throughout: the samplerate is bad (but there are ERP studies with 250 Hz), the channel placement is best suited for capturing facial emg rather than EEG, and so on.

Of course, it is a consumer product and this is how you should treat it - a very low quality device. Setting it up costs very little time, but this benefit is easily offset by the amount of extra noise. ERPs are quite small (think 0.5 to 8 uV) compared to raw EEG (e.g. 50 uV), which is strongly affected by artefacts (e.g. eye movements 200-800 uV). As a result,we already need many, many repetitions with high quality equipment to get reliable results (around 80 seems generally acceptable). Imagine you have an experiment that takes an hour - two hours in order to get a reasonable signal to noise ratio. Now, say your noise has doubled due to using bad electrodes, low grade conductivity, cheap amplifiers, few channels: a strong (I think exponential) increase in the number of trials should be predicted. I think the additional cost of this is easily more than using a "proper" EEG set. Funnily, most papers I review these days who advocate use of such and similar headsets seem to argue exactly the opposite: that a cheap headset allows for use of fewer trials, less experimental rigour, and little overall expertise.

However, if you insist on using the Emotiv (and it can be fun!), try Debener et al. (Psychophysiology), "How about taking a low-cost, small, and wireless EEG for a walk?" http://onlinelibrary.wiley.com/doi/10.1111/j.1469-8986.2012.01471.x/abstract It features a hacked Emotiv and seems one of the very few attempts at doing real ERP with the Emotiv.

Thanks everyone for your great analyses of the relative weaknesses of the Emotiv EPOC system.  I especially appreciated everyone warning me about the sample rate issue which I wasn't aware of.  My desire was to start using this system to collect pilot data that could get me back into the swing of things with ERPs.  I did ERP research in the 1980s with the University of IL group (Michael Coles, Art Kramer, etc.) but I've mainly been doing other types of research and clinical work with autonomic measures of psychophysiology since then. 

I was hoping to find a starter system that I could afford to buy myself until I could write a grant to get a better system.  So my second question still remains.  Is there anything under $5,000 that would serve my basic purposes?  I will mainly be doing auditory and possibly visual ERPs of people engaging in advanced meditation practices such as gtumo. 

You may be interested by these two papers, which attempted to compare EEG/ERP signal quality between the Emotiv headset and research-grade devices:

http://dx.doi.org/10.7717/peerj.38

http://dx.doi.org/10.1166/jnsne.2014.1092

Note that the first of these papers takes the interesting approach of sacrificing a pair of channels on the Emotiv headset in order to gain a precise marker of stimulus onset (as the Emotiv device crucially lacks a TTL input for this purpose). A precise event time marker is essential if you plan on doing ERP research, as the signal will otherwise be blurred when you try averaging trials.

In terms of low cost devices, I think one of the most interesting options available right now is OpenBCI (http://www.openbci.com). It is a very barebones system, so may not be ideal if you are looking for an "out of the box" solution; however, it offers up to 16 input channels with conventional paste-type electrodes. The setup will not be as convenient, but dry-electrodes—such as those used in most consumer-grade devices currently on the market—are still not nearly as reliable as the wet electrodes used by nearly all clinical- and research-grade devices (e.g. see this review: http://dx.doi.org/10.1109/RBME.2010.2084078). This EGI device may also be worth a look, though I have not heard much about it yet: http://www.egi.com/research-division/research-division-research-products/avatar

Hi, I am one of the lead PIs on one of the above linked papers (Ries et al).

We have a lot of experience w/ the Emotiv Epoc. As shown in the paper, it is indeed quite possible to get good-looking ERPs, and for classifiers to work OK with it. So the good news is that it is indeed a viable low-cost solution, even with the poor sample rate etc.

Note that the poor sample rate can be overcome by using stimulus epoch which are jittered/asynchronous to the sample rate, so when you average you get several points on the curve.

The bad news is that it does require some extra work. For example we have found teh timing to be wholly unreliable... however, the saving grace is that it can be resolved wit ha fairly simple post-hoc integration of data if you plan ahead. See here, and the above paper for an explanation:

http://oai.dtic.mil/oai/oai?verb=getRecord&metadataPrefix=html&identifier=ADA561715

Good luck

Dave

It seems we agree that, with enough hardware hacks, you may get something out of the Emotiv Epoc. Ideally, one would 1) remove the original 16 electrodes and replace them with Ag/Cl ones, 2) but sacrifice two for TTL input, Beyond that, I think their marketing department is so good, they don't need any extra from me :)

As for cheaper, but research grade systems. Surely with 5000 USD, you should get somewhere even with the "traditional" companies. I think most of them have some product out there that is significantly cheaper than the others. EGI is already mentioned and, of course, very respectable and will likely answer your enquiries as to the cost, benefit and problems with this "Avatar" unit linked above (I only have experience with their excellent high density solutions). Given the low current value of the Euro, several European companies could also give good value for money. BrainProducts offers a cheaper amp: http://www.brainproducts.com/productdetails.php?id=15 . G.tec likewise has a mobile unit: http://www.gtec.at/Products/(l)/1

Sadly, research grade EEG websites seldom put prices on their websites, so I'm not sure whether the products above will not be closer to the 10k end than the 5k. But most of them can be easily contacted and will be eager to give you free quote.

Personally, I like the Mistar 201 system.  http://bio-medical.com/products/mitsareeg201.html, yes it's a little pricey, but it has 21 channels and will do EPs right out the box because it comes with WinEEG software, and it is a very extensible system.  If I were going to throw $5K of my own money at something, for the $2K extra, this is what I would get.  Whatever you do buy, though, make sure it is compatible with WinEEG.  More clinically oriented units such as the Brainmaster Discovery 24E ($5900) are not.  If FGCU assigns me to a permanent full-time position, I definitely will self-purchase the Mistar, because of their quality for research work.  As far as I can tell, the Mistar systems are the most inexpensive way to reliably collect the  ERP (e.g., ERD/ERS) data I'm actually interested in.

Yes, basically you get what you pay for, regarding finesse and refinement for research applications. There really is nothing that is yet accepted by the community as "good" for research in this price range. 

Note that probably 90% of all new EEG systems coming out on the market today are, or will be, based on the same chipset, the TI ADS1299, which is basically an all-in-one very hi-fi DAQ intended for EEG... the difference really is packaging or how refined it is. The problem is that chipset is fairly power hungry and not cheap (on teh small scale that EEG systems are sold) so by the time a MFR makes a system aroudn it, and has to get paid for their time & R&D... it's expensive.

And as mentioned, this is the heart of the OpenEEG boards, which can be purchased very affordable but then you're on your own to finish teh system..

To get back to the original question - we do not use Matlab for data acquisition with Emotiv, but we do all analyses in Matlab. I have never had a problem with data conversion, translation, importing, etc. We generally use EEGLab (free toolbox in ML), and it will import the .edf data files just fine. After that point the source is irrelevant.

Also note that BCILab and other SCCN-based DAQ programs support Emotiv for input.

We used it long time ago in my former lab. It was horrible. Neither "Emotion" recognition over face movement worked nor true ERPs could be detected. We tried a P300 experiment but in the averaged signals there was neither a P300 nor did the signal processing find any pattern for classification. There was also no alpha waves to be seen when the subject closed the eyes. So not working at all and contacting Emotiv was also useless. They completely ignored us. Even worse, we had the much more expensive research license...

I had the same experience the Emotive support. Never answered me a question correctly and the support had no experience the Matlab. They all programming with C++...

The Problem with Emotiv EPOC and ERP's is the EPOC jitter. It destroys all ERP experiments. I tested it once. The loss of sample-time was 2 sec in a 5 min experiment.

The trick is to use the clock of the computer for the recorded data.

In the old emotive reseach DLL library there was a well-working Matlab example (Unfortunately not for ERP experiments).

I have some experience with Emotive.

Pros: cheap and mobile, Cons: limited channels for ERP, poor sampling rate and new versions require payment for sofware licence.

However, I have achieved classifying ERPs with this devise. I used only 6 trials.

@Apdullah Yayik

Are you using community SDK or purchased Licensed software ?

I've had nightmare experiences with Emotiv products and customer support for both research and then using some cheaper products to use in neuroscience lab.

I am going to back to raw EEG just using gold cap EEG leads and conductive paste. Yes, more work, but it's like a miracle after dealing with Emotiv. I've heard good things about Open BCI but have never used their products.

Here's a paper using the newly released Emotiv EXTENDER to provide hardware trigger events:

https://contattafiles.s3.us-west-1.amazonaws.com/tnt3851/6XNmjhLngkiGIeN/NWilliams_ACNS2019_FlexVal_poster.pdf

https://contattafiles.s3.us-west-1.amazonaws.com/tnt3851/7Ls47YNYRKl6r0u/eventMark_report_200320.pdf

Also one using a modified Emotiv Epoc+ for ERP trials: Article Validation of Emotiv EPOC+ for extracting ERP correlates of ...

Does Emotiv let you borrow the instrument to try first?