Testing the validity of wireless EEG for cognitive research with auditory and visual paradigms

Publikation: Forskning - peer reviewKonferenceabstrakt til konference

One of the challenges in collecting ERP data is the time-consuming process of fitting caps and prepping electrodes with gel. This can be particularly true when working with clinical populations, where efficiency in data collection is important. Recently gel-free wireless headsets designed for brain computer interface have advanced to the point where they offer an attractive alternative to traditional EEG methods. However, although research exists suggesting that some of these devices can register the robust P300 component (e.g. Duvinage et al, 2012), little is known about their validity for earlier and smaller cognitive components. To test the feasibility of these headsets for cognitive research, we compared performance of the Emotiv Epoc wireless headset (EM) with Brain Products ActiCAP (BP) active electrodes on two well-studied components: the auditory mismatch negativity (MMN) and the visual face-sensitive N170. We collected ERP data from 32 healthy adult participants who performed passive listening (MMN) and passive viewing tasks (N170). Because of recording error, data from only 30 participants were available from the Brain Products MMN session, and 25 from the Emotiv N170 and the Brain Products N170 sessions. Data were collected first with Emotiv, and then with the Brain Products electrodes. For the MMN, participants heard 6-8 standard tones (440 Hz) followed by a deviant (340 Hz or 640 Hz). For the N170, participants viewed images of 20 unknown faces and 20 houses in randomized order. The MMN is typically measured at frontal midline electrodes (Fz and Cz). Because the electrode layout in the Emotiv does not allow this, we measured the MMN in the Emotiv at neighboring electrodes F3 and F4, and at F3, F4, and Fz in the Brain Products electrodes. The traditional N170 sites, P7 and P8, are available in both Emotiv and Brain Products layouts, so we used these in our analysis. Both MMN and N170 were measured as difference waves (MMN: deviants - standards, N170: faces - houses). Significance was measured with one-sample t-tests at each of the chosen electrodes. The MMN was visible in data collected from both systems. T-test results were as follows: F3 (BP: t(29) = -4.2, p = 0.0002; EM: t(31) = -2.4, p = 0.023) , F4 (BP: t(29) = -4.1, p = 0.0003; EM: t(31) = -3.6, p = 0.001) and Fz (BP: t(29) = -2.4, p = 0.022). The N170 was visible in the Brain Products data, but not in the Emotiv data. T-test results were as follows: P7 (BP: t(24) = -4.2, p = 0.0003; EM: t(24) = -0.9; p=0.34) P8 (BP: t(24) = -3.7, p = 0.001; EM: t(24) = -0.2(24), p = 0.86). We successfully measured a comparable MMN response in a wireless headset and a traditional electrode set. The N170 component, however, was only visible with traditional electrodes. Our data suggest that, depending on the paradigm used, wireless headsets like the Emotiv may be a viable alternative to traditional ERP collection methods. However, issues like low sampling rate and limited scalp coverage currently still limit their broad applicability.
OriginalsprogEngelsk
Udgivelsesår22 jun. 2014
StatusUdgivet - 22 jun. 2014
BegivenhedSNL 2014 - Society for the Neurobiology of Language - Amsterdam, Holland
Varighed: 28 aug. 201429 aug. 2014

Konference

KonferenceSNL 2014 - Society for the Neurobiology of Language
LandHolland
ByAmsterdam
Periode28/08/201429/08/2014

Se relationer på Aarhus Universitet Citationsformater

ID: 78874171