Long-term test-retest reliability of the P3 NoGo wave and two independent components decomposed from the P3 NoGo wave in a visual Go/NoGo task Jan Ferenc Brunner a,b, ⁎, Tor Ivar Hansen b , Alexander Olsen a,c , Toril Skandsen a,b , Asta Håberg b , Juri Kropotov d,e,f a Department of Physical medicine and rehabilitation, St.Olav Hospital, Trondheim University Hospital, Norway b Department of Neuroscience, Norwegian University of Science and Technology (NTNU), Trondheim, Norway c MI Lab and Department of Circulation and Medical Imaging (NTNU), Norway d Department of Psychology, NTNU, Trondheim, Norway e Institute of the Human Brain of Russian Academy of Sciences, St. Petersburg, Russian Federation f Andrzej Frycz Modrzewski Krakow University, Krakow, Poland abstract article info Article history: Received 2 November 2012 Received in revised form 31 May 2013 Accepted 4 June 2013 Available online 12 June 2013 Keywords: ERP Independent component analysis P3 NoGo Reliability The objective of the current study was to determine long-term test-retest reliability of the P3 NoGo wave as well as two independent components (IC P3 NoGo early and IC P3 NoGo late) decomposed from this wave by independent component analysis (ICA). For this purpose 19-channel EEG was recorded during a cued visual Go/NoGo task. First, spatial filters of the two independent components (ICs) were obtained by application of ICA to ERPs of 102 healthy adults. Second, in 26 individuals, ERPs were recorded from the same task a second time 6–18 months after the first recording. Statistical analyses were performed on both the P3 NoGo waves and the activation curves of the two ICs from both recordings. Amplitude and latency were estimated by “peak” and “fractional area” (FA) methods for both the P3 NoGo wave and ICs. Intraclass correlations (ICC) for latency were excellent (ICC > .90) for both the P3 NoGo wave and the two ICs when measured with the FA method, and good (ICC > .75) for amplitude measured by both methods (FA or peak). We conclude that the long-term stability of P3 NoGo wave and the two IC makes them well suited for reliable assessment of cognitive control function in research and clinic. © 2013 Elsevier B.V. All rights reserved. 1. Introduction There is increased optimism for the application of ERPs for clinical purposes (Duncan et al., 2009; Dockree and Robertson, 2011; Folmer et al., 2011). However, to become a useful tool for clinical assessment, ERPs need to be valid and reliable at the individual level. Further- more, high test-retest reliability is important. It is suggested that for research purposes (i.e., exploring the differences between groups), test-retest reliability more than .80 is adequate, while for clinical assessment a test-retest reliability of .90 should be the minimum (Nunnally and Bernstein, 1994). Most of the previous studies on test-retest reliability in ERP research have focused on the P3b wave in the auditory oddball paradigm or the error-related negativity (ERN) in paradigms where errors are committed. Test-retest reliabil- ity varies in the different studies, from 0.50 to 0.86 for amplitude pa- rameters and from 0.40 to 0.88 for latency parameters (Segalowitz and Barnes, 1993; Mathalon et al., 2000; Sandman and Patterson, 2000; Walhovd and Fjell, 2002; Williams et al., 2005; Hall et al., 2006; Segalowitz et al., 2010; Cassidy et al., 2012) Test-retest reliability of ERPs depends on many factors, including fluctuations in the functional state of the examinee and measurement errors (Segalowitz and Barnes, 1993). The measurement errors partly depend on the method for measuring latency and amplitude of the ERP wave. The most commonly used method is to measure the ampli- tude and latency of the ERP wave at its maximums or minimum (peak measurements). For ERPs without a clear peak (e.g. P300) a better technique is provided by the fractional area (FA) approach (Hansen and Hillyard, 1980; Kiesel et al., 2008; Kappenman and Luck, 2012). A high number of trials seem important for latency measurement, and averaging of more than 70 trials has been recommended for assessment of wave like P300 (Kiesel et al., 2008). The P300 wave (usually P3b/P3 Go) often does not show a clear onset and its peak latency can be difficult to determine, because the wave has a wide temporal extension without a sharp peak. The Go/NoGo task is commonly used to study cognitive control functions. There are different versions of this paradigm, all consisting of sequential and random presentation of stimuli that a person is instructed to respond to (Go stimuli) or not respond to (NoGo International Journal of Psychophysiology 89 (2013) 106–114 ⁎ Corresponding author at: Fysikalsk Medisin, Lian, Vådanveien 39, 7024 Trondheim, Norway. Tel.: +47 97713663; fax: +47 72822601. E-mail address: jan.brunner@stolav.no (J.F. Brunner). 0167-8760/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ijpsycho.2013.06.005 Contents lists available at SciVerse ScienceDirect International Journal of Psychophysiology journal homepage: www.elsevier.com/locate/ijpsycho