Article

Single-trial detection in EEG and MEG: Keeping it linear.

Vision Technologies Laboratory, Sarnoff Corporation, Princeton, NJ 08540, USA; Department of Psychology and Department of Computer Science, University of New Mexico, Albuquerque, NM 87131, USA; Department of Psychology, Princeton University, Princeton, NJ 08544, USA; Department of Psychology, University of Pennsylvania, Philadelphia, PA 19104, USA; Department of Biomedical Engineering, 351 Engineering Terrace, MC 8904, Columbia University, New York, NY 10027, USA
Neurocomputing 01/2003; 52-54:177-183. DOI: 10.1016/S0925-2312(02)00821-4
Source: DBLP

ABSTRACT Conventional electroencephalography (EEG) and magnetoencephalography (MEG) analysis often rely on averaging over multiple trials to extract statistically relevant differences between two or more experimental conditions. We demonstrate that by linearly integrating information over multiple spatially distributed sensors within a predefined time window, one can discriminate conditions on a trial-by-trial basis with high accuracy. We restrict ourselves to a linear integration as it allows the computation of a spatial distribution of the discriminating source activity. In the present set of experiments the resulting source activity distributions correspond to functional neuroanatomy consistent with the task (e.g. contralateral sensory-motor cortex and anterior cingulate).

0 Bookmarks
 · 
75 Views
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Neuroergonomics complements traditional ergonomic approaches to understanding many aspects of human–system performance, including the analysis of human error. This paper reviews the evidence on neural signals associated with different error types, specifically the error-related negativity (ERN), a brain potential that is extracted from the scalp electroencephalogram and that is generated in a medial prefrontal brain region, the anterior cingulate cortex. Studies of the ERN indicate that the brain has a specialised error monitoring and feedback system with close ties to brain networks involved in learning and decision making. The implications of this system for understanding how errors are made are discussed, as well as how errors are captured, learned from and used as feedback to enhance future performance. Applications of neuroergonomic studies of human error–in the development of brain–computer interfaces, selection and training design and procedures for system diagnosis–are also described. The current literature on the neural basis for human error processing is presented within the framework of neuroergonomics. This review relates the basic understanding of error processing to work-related problems. It is argued that better systems can be created provided the neural processes underlying behaviour are taken into account.
    Theoretical Issues in Ergonomics Science 09/2010; 11(5):402-421.
  • [Show abstract] [Hide abstract]
    ABSTRACT: In order to create reading support systems, the feasibility of using viewed size predictions was evaluated using features of EEG waveforms. Viewer's occipital single channel potentials were measured while six sizes of Landolt circles were shown. Features such as series of potentials, latencies of P1 and N1 peaks, and 5 vectors of Fourier descriptors (FD) were extracted from the EEG waveforms. The performances of 6-classes were classified and compared across sets of each feature and combinations of these. The EEG waveforms were transformed when accumulated. The five-fold cross validation rate for all sets of features increased with the number of EEGs accumulated. The accuracy of a feature combination for P1, N1 and FD is significant when the waveform is measured twice. This result suggests that the size of objects being viewed can be predicted using a set of features which can be extracted from a single channel EEG waveform recorded at an occipital electrode.
    01/2011;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Analyzing brain states that correspond to event related potentials (ERPs) on a single trial basis is a hard problem due to the high trial-to-trial variability and the unfavorable ratio between signal (ERP) and noise (artifacts and neural background activity). In this tutorial, we provide a comprehensive framework for decoding ERPs, elaborating on linear concepts, namely spatio-temporal patterns and filters as well as linear ERP classification. However, the bottleneck of these techniques is that they require an accurate covariance matrix estimation in high dimensional sensor spaces which is a highly intricate problem. As a remedy, we propose to use shrinkage estimators and show that appropriate regularization of linear discriminant analysis (LDA) by shrinkage yields excellent results for single-trial ERP classification that are far superior to classical LDA classification. Furthermore, we give practical hints on the interpretation of what classifiers learned from the data and demonstrate in particular that the trade-off between goodness-of-fit and model complexity in regularized LDA relates to a morphing between a difference pattern of ERPs and a spatial filter which cancels non task-related brain activity.
    NeuroImage 05/2011; 56(2):814-25. · 6.25 Impact Factor

Full-text (4 Sources)

View
20 Downloads
Available from
May 16, 2014