Time-frequency analysis of single-sweep event-related potentials by means of fast wavelet transform

A time-frequency decomposition was applied to the event-related potentials (ERPs) elicited in an auditory oddball condition to assess differences in cognitive information processing. Analysis in the time domain has revealed that cognitive processes are reflected by various ERP components such as N1, P2, N2, P300, and late positive complex. However, the heterogeneous nature of these components has been strongly emphasized due to simultaneously occurring processes. The wavelet transform (WT), which decomposes the signal onto the time-frequency plane, allows the time-dependent and frequency-related information in ERPs to be captured and precisely measured. A four-octave quadratic B-spline wavelet transform was applied to single-sweep ERPs recorded in an auditory oddball paradigm. Frequency components in delta, theta, and alpha ranges reflected specific aspects of cognitive information processing. Furthermore, the temporal position of these components was related to specific cognitive processes.     

Wavelet analysis of neuroelectric waveforms: a conceptual tutorial

This paper presents a nontechnical, conceptually oriented introduction to wavelet analysis and its application to neuroelectric waveforms such as the EEG and event related potentials (ERP). Wavelet analysis refers to a growing class of signal processing techniques and transforms that use wavelets and wavelet packets to decompose and manipulate time-varying, nonstationary signals. Neuroelectric waveforms fall into this category of signals because they typically have frequency content that varies as a function of time and recording site. Wavelet techniques can optimize the analysis of such signals by providing excellent joint time-frequency resolution. The ability of wavelet analysis to accurately resolve neuroelectric waveforms into specific time and frequency components leads to several analysis applications. Some of these applications are time-varying filtering for denoising single trial ERPs, EEG spike and spindle detection, ERP component separation and measurement, hearing-threshold estimation via auditory brainstem evoked response measurements, isolation of specific EEG and ERP rhythms, scale-specific topographic analysis, and dense-sensor array data compression. The present tutorial describes the basic concepts of wavelet analysis that underlie these and other applications. In addition, the application of a recently developed method of custom designing Meyer wavelets to match the waveshapes of particular neuroelectric waveforms is illustrated. Matched wavelets are physiologically sensible pattern analyzers for EEG and ERP waveforms and their superior performance is illustrated with real data examples.     

Multiresolution Analysis of Event-Related Potentials by Wavelet Decomposition

Wavelet analysis is presented as a new tool for analyzing event-related potentials (ERPs). The wavelet transform expands ERPs into a time-scale representation, which allows the analyst to zoom in on the small scale, fine structure details of an ERP or zoom out to examine the large scale, global waveshape. The timescale representation is closely related to the more familiar time-frequency representation used in spectrograms of time-varying signals. However, time-scale representations have special properties that make them attractive for many ERP applications. In particular, time-scale representations permit theoretically unlimited time resolution for the detection of short-lived peaks and permit a flexible choice of wavelet basis functions for analyzing different types of ERPs. Generally, time-scale representations offer a formal basis for designing new, specialized filters for various ERP applications. Among recently explored applications of wavelet analysis to ERPs are (a) the precise identification of the time of occurrence of overlapping peaks in the auditory brainstem evoked response; (b) the extraction of single-trial ERPs from background EEG noise; (c) the decomposition of averaged ERP waveforms into orthogonal detail functions that isolate the waveforms experimental behavior in distinct, orthogonal frequency bands; and (d) the use of wavelet transform coefficients to concisely extract important information from ERPs that predicts human signal detection performance. In this tutorial we present an intuitive introduction to wavelets and the wavelet transform, concentrating on the multiresolution approach to wavelet analysis of ERP data. We then illustrate this approach with real data. Finally,we offer some speculations on future applications of wavelet analysis to ERP data.     

Oscillatory brain dynamics, wavelet analysis, and cognition

On the basis of a systems theoretical approach it was hypothesized that event-related potentials (ERPs) are superpositions of stimulus-evoked and time-locked EEG rhythms reflecting resonance properties of the brain (Ba?ar, 1980). This approach led to frequency analysis of ERPs as a way of analyzing evoked rhythms. The present article outlines the basic features of ERP frequency analysis in comparison to ERP wavelet analysis, a recently introduced method of time-frequency analysis. Both methods were used in an investigation of the functional correlates of evoked rhythms where auditory and visual ERPs were recorded from the cat brain. Intracranial electrodes were located in the primary auditory cortex and in the primary visual cortex thus permitting "cross-modality" experiments. Responses to adequate stimulation (e.g., visual ERP recorded from the visual cortex) were characterized by high amplitude alpha (8-16 Hz) responses which were not observed for inadequate stimulation. This result is interpreted as a hint at a special role of alpha responses in primary sensory processing. The results of frequency analysis and of wavelet analysis were quite similar, with possible advantages of wavelet methods for single-trial analysis. The results of frequency analysis as performed earlier were thus confirmed by wavelet analysis. This supports the view that ERP frequency components correspond to evoked rhythms with a distinct biological significance.     

Generate,maintain, manipulate? Exploring the multidimensional nature of motor imagery

Motor imagery (MI), the mental rehearsal of a motor task, is thought to be a synergistic interaction between processes that allow for the generation, maintenance, and manipulation, of motor images. While our understanding of the multidimensional nature of MI stems from research examining various methods to assess MI ability, limited research has been conducted employing multiple assessments across participants to probe the underlying dimensions of MI. Accordingly, the current study sought to explore the multidimensional nature of MI using an exploratory approach that would allow for the dimensions of imagery to be examined and linked to each assessment measure. Specifically, participants (N = 81) underwent a battery of MI-assessments (including questionnaires, mental chronometry, a hand laterality judgment task, and a MI-based learning task), and data was analyzed via principal component analysis (PCA). Three components resulted, which were named based on which outcome measures loaded on to the component: generation, maintenance, and manipulation of motor images. We further discuss a fourth component, named ‘temporal sequencing’ of motor images, identified via the initial component solution. Concordant with previous research, we highlight the importance of employing multiple measures when assessing imagery ability prior to its use in training. Notably, this work allowed us to link assessments of MI to the different dimensions of MI, informing on the nature of MI.     

利用时频分析研究非相位锁定脑电活动

时频分析技术自20世纪80年代被引入到心理学脑电数据分析领域以来, 克服了传统的时域ERP方法只能分析相位锁定成分的缺陷, 可以帮助研究者挖掘到脑电信号中非相位锁定的成分。在心理学领域, 应用最多的时频分析方法是小波变换和Hilbert变换, 而能量、相位一致性和耦合是三个最常用的分析指标。研究者利用不同的分析指标来揭示不同的心智过程。不同频段的能量被认为体现了不同的认知过程, 如α能量被发现与注意选择性有关, 而γ能量则与特征整合相关。相位一致性常被用于讨论ERP产生的机制。耦合则通常说明了长距离脑区之间的信息交流以及高级脑区对低级脑区的认知控制, 在完成各种复杂认知任务的时候会表现出不同的耦合模式。     

Information processing during sleep onset and sleep.

This Special Section examines the extent of information processing during sleep onset and sleep itself. It is generally agreed that, stimulus input is markedly inhibited during sleep, thus preventing conscious awareness of the external environment. Overt behavioural responses are rarely made within sleep. Two neurophysiological measures are therefore often used. The electrical activity of the brain (the EEG) can be employed to distinguish waking (conscious) from sleeping (unconscious) states. It is also possible to quantify the EEG prior to and following a detection (or a failure of a detection) of a stimulus. Such measures can thus be used to predict conscious awareness. A second measure that frequently has been employed is the brain's response to an external stimulus (the evoked potential). Different components of the evoked potential can be used to trace the extent of information processing during the different states of consciousness. Some are associated with a preconscious detection while others are associated with conscious awareness. Other evoked potentials may be unique to sleep.     

Semi-sparse PCA

It is well known that the classical exploratory factor analysis (EFA) of data with more observations than variables has several types of indeterminacy. We study the factor indeterminacy and show some new aspects of this problem by considering EFA as a specific data matrix decomposition. We adopt a new approach to the EFA estimation and achieve a new characterization of the factor indeterminacy problem. A new alternative model is proposed, which gives determinate factors and can be seen as a semi-sparse principal component analysis (PCA). An alternating algorithm is developed, where in each step a Procrustes problem is solved. It is demonstrated that the new model/algorithm can act as a specific sparse PCA and as a low-rank-plus-sparse matrix decomposition. Numerical examples with several large data sets illustrate the versatility of the new model, and the performance and behaviour of its algorithmic implementation.

     

Feature extraction of event-related potentials using wavelets: an application to human performance monitoring

This report describes the development and evaluation of mathematical models for predicting human performance from discrete wavelet transforms (DWT) of event-related potentials (ERP) elicited by task-relevant stimuli. The DWT was compared to principal components analysis (PCA) for representation of ERPs in linear regression and neural network models developed to predict a composite measure of human signal detection performance. Linear regression models based on coefficients of the decimated DWT predicted signal detection performance with half as many free parameters as comparable models based on PCA scores. In addition, the DWT-based models were more resistant to model degradation due to over-fitting than PCA-based models. Feed-forward neural networks were trained using the backpropagation algorithm to predict signal detection performance based on raw ERPs, PCA scores, or high-power coefficients of the DWT. Neural networks based on high-power DWT coefficients trained with fewer iterations, generalized to new data better, and were more resistant to overfitting than networks based on raw ERPs. Networks based on PCA scores did not generalize to new data as well as either the DWT network or the raw ERP network. The results show that wavelet expansions represent the ERP efficiently and extract behaviorally important features for use in linear regression or neural network models of human performance. The efficiency of the DWT is discussed in terms of its decorrelation and energy compaction properties. In addition, the DWT models provided evidence that a pattern of low-frequency activity (1 to 3.5 Hz) occurring at specific times and scalp locations is a reliable correlate of human signal detection performance.     

Dimensional analyses of taxonic data

Principal component analysis (PCA) and common factor analysis are often used to model latent data structures. Typically, such analyses assume a single population whose correlation or covariance matrix is modelled. However, data may sometimes be unwittingly sampled from mixed populations containing a taxon (nonarbitrary subpopulation) and its complement class. One derives relations between values of PCA parameters within subpopulations and their values in the mixed population. These results are then extended to factor analysis in mixed populations. As relationships between subpopulation and mixed-population principal components and factors sensitively depend on within-subpopulation structures and between-subpopulation differences, naive interpretation of PCA or factor analytic findings can potentially mislead. Several analyses, better suited to the dimensional analysis of admixture data structures, are presented and compared.     

Habituation and the human evoked potential

Habituation of human scalp-recorded cerebral evoked potentials was studied in response to auditory and visual repetitive stimuli of different intensities. Changes in magnitudes of evoked potentials with stimulus repetition were examined according to the parametric characteristics of habituation, generalization, and dishabituation. In addition, tests of the predictions of two theories of habituation were made regarding the degree and direction of intensity generalization of habituation. Both auditory and visual evoked potentials exhibited decrements in response magnitudes across the repetitive stimuli consistent with the parametric criteria of habituation. Early evoked potential peak components showed a pattern of intensity generalization of habituation consistent with the predictions of the dual-process theory of habituation. Intensity generalization of late evoked potential peak components occurred in a manner more consistent with the predictions of the stimulus comparator theory of habituation. These results provide further evidence that evoked potentials can be used as electrophysiological indexes of plasticity in humans.     

Hemispheric distribution of ERP components and word naming in preschool children

Event-related potentials (ERPs) were recorded at the left and right temporal and parietal sites during a word-naming task in 61 children ranging in age from 5.1 to 6.2 years. Principal component analysis (PCA) was performed on the average ERP waveforms. Analyses of the factor scores revealed that the P240, N380, and SW components were asymmetrically distributed over the hemispheres. Fast reading acquisition was associated with smaller N380 and larger SW activity than slow reading acquisition. Multiple regression analyses indicated that only right hemispheric ERP components were significantly associated with rate of reading acquisition. These findings are discussed in terms of right hemispheric involvement in early reading acquisition.     

Computational and performance aspects of PCA-based face-recognition algorithms

Algorithms based on principal component analysis (PCA) form the basis of numerous studies in the psychological and algorithmic face-recognition literature. PCA is a statistical technique and its incorporation into a face-recognition algorithm requires numerous design decisions. We explicitly state the design decisions by introducing a generic modular PCA-algorithm. This allows us to investigate these decisions, including those not documented in the literature. We experimented with different implementations of each module, and evaluated the different implementations using the September 1996 FERET evaluation protocol (the de facto standard for evaluating face-recognition algorithms). We experimented with (i) changing the illumination normalization procedure; (ii) studying effects on algorithm performance of compressing images with JPEG and wavelet compression algorithms; (iii) varying the number of eigenvectors in the representation; and (iv) changing the similarity measure in the classification process. We performed two experiments. In the first experiment, we obtained performance results on the standard September 1996 FERET large-gallery image sets. In the second experiment, we examined the variability in algorithm performance on different sets of facial images. The study was performed on 100 randomly generated image sets (galleries) of the same size. Our two most significant results are (i) changing the similarity measure produced the greatest change in performance, and (ii) that difference in performance of +/- 10% is needed to distinguish between algorithms.     

Spatial property of motion visual evoked potentials.

The cortical evoked response to drifting patterns (motion visual evoked potentials) was investigated. When the direction of motion of the stimulus pattern was reversed upward or downward at intervals, the cortical evoked response was triggered at the moment when the pattern changed direction. The polarity reversal of the main negative component occurred between upper and lower visual field stimulations as seen in pattern reversal visual evoked potentials. Our study indicates these potentials have a compound property reflecting the visual field.     

EP Component Identification and Measurement by Principal Components-Analysis

Between the acquisition of Evoked Potential (EP) data and their interpretation lies a major problem: What to measure? An approach to this kind of problem is outlined here in terms of Principal Components Analysis (PCA). An important second theme is that experimental manipulation is important to functional interpretation. It would be desirable to have a system of EP measurement with the following characteristics: (1) represent the data in a concise, parsimonous way; (2) determine EP components from the data without assuming in advance any particular waveforms for the components; (3) extract components which are independent of each other; (4) measure the amounts (contributions) of various components in observed EPs; (5) use measures that have greater reliability than measures at any single time point or peak; and (6) identify and measure conponents that overlap in time. PCA has these desirable characteristics. Simulations are illustrated. PCA′s beauty also has some warts that are discussed. In addition to discussing the usual two-mode model of PCA, an extension of PCA to a three-mode model is described that provides separate parameters for (1) waveforms over time, (2) coefficients for spatial distribution, and (3) scores telling the amount of each component in each EP. PCA is compared with more traditional approaches. Some biophysical considerations are briefly discussed. Choices to be made in applying PCA are considered. Other issues include misallocation of variance, overlapping components, validation, and latency changes.     

The semantic component of the evoked potential of differentiation

This work analyzes data from recordings of (occipital and temporal) cortical evoked potentials (called evoked potentials of differentiation (EPD) occurring in humans in response to an abrupt substitution of stimuli. As stimuli we used three groups of words: the names of the ten basic colors taken from Newton's color circle; the names of seven basic emotions forming Shlossberg's circle of emotions; and seven nonsense words comprised of random combinations of letters. Within each group of word stimuli we constructed a matrix of the differences between the amplitudes of mid-latency components of EPD for each pair of words. This matrix was analyzed using the method of multidimensional scaling. As a result of this analysis we were able to distinguish the semantic and configurational components of EPD amplitude. The semantic component of EPD amplitude was evaluated by comparing structure of the data obtained to the circular structures of emotion and color names. The configurational component was evaluated on the basis of the attribute of word length (number of letters). It was demonstrated that the semantic component of the EPD can only be detected in the left occipital lead at an interpeak amplitude of P120-N180. The configurational component is reflected in the occipital and temporal leads to an identical extent, but only in the amplitude of a later (N180-P230) component of the EPD. The results obtained are discussed in terms of the coding of categorized, configurational, and semantic attributes of a visual stimulus.     

Decision-related cortical potentials during an auditory signal detection task with cued observation intervals.

Cortical-evoked potentials were recorded from human subjects performing an auditory detection task with confidence rating responses. Unlike earlier studies that used similar procedures, the observation interval during which the auditory signal could occur was clearly marked by a visual cue light. By precisely defining the observation interval and, hence, syncrhonizing all perceptual decisions to the evoked potential averaging epoch, it was possible to demonstrate that high-confidence false alarms for accompanied by late-positive P3 components equivalent to those for equally confident hits. Moreover the hit and false alarm evoked potentials were found to covary similarily with variations in confidence rating and to have similar amplitude distributions over the scalp. In a second experiment wherein the signal intensity was increased to make signal presence and signal absence clearly discriminable and the a priori probability of signal presentation was varied from .5 to .9, it was demonstrated that correct rejections can be associated with a P3 component larger than that for hits. Thus it was possible to show, within the signal detection paradigm, how the two major factors of decision confidence and expectancy are reflected in the P3 component of the cortical-evoked potential.     

Eye movements and brain electric potentials during reading

The development of theories and computational models of reading requires an understanding of processing constraints, in particular of timelines related to word recognition and oculomotor control. Timelines of word recognition are usually determined with event-related potentials (ERPs) recorded under conditions of serial visual presentation (SVP) of words; timelines of oculomotor control are derived from parameters of eye movements (EMs) during natural reading. We describe two strategies to integrate these approaches. One is to collect ERPs and EMs in separate SVP and natural reading experiments for the same experimental material (but different subjects). The other strategy is to co-register EMs and ERPs during natural reading from the same subjects. Both strategies yield data that allow us to determine how lexical properties influence ERPs (e.g., the N400 component) and EMs (e.g., fixation durations) across neighboring words. We review our recent research on the effects of frequency and predictability of words on both EM and ERP measures with reference to current models of eye-movement control during reading. Results are in support of the proposition that lexical access is distributed across several fixations and across brain-electric potentials measured on neighboring words.     

Simulation of the response of a multichannel nerve array to pulse shapes produced by single-channel electrostimulation

Results of experiments on single-channel electrostimulation of the cochlea which throw light on the performance of the central auditory nervous system (CANS) have recently been reported by Dobie and Dillier. Trains of pulses with different rise times could be distinguished by subjects with cochlea implants, even though time differences involved were very small. It was suggested by the authors that the information is carried to the CANS by an array of nerve fibres with characteristic time differences. In the present paper, simulations produced by means of a nerve model are reported and used to compute the patterns of action potentials evoked on the nerve array by different pulse trains. The changes in the patterns of the nerve responses resulting from the shape variations which have to be perceived by the CANS are examined.     

Vertex potentials evoked during auditory signal detection: Relation to decision criteria

Vertex potentials were recorded from eight Ss performing in an auditory threshold detection task with rating scale responses. The amplitudes and latencies of both the N1 and the late positive (P3) components were found to vary systematically with the criterion level of the decision. These changes in the waveshape of the N1 component were comparable to those produced by varying the signal intensity in a passive condition, but the late positive component in the active task was not similarly related to the passively evoked P2 component. It was suggested that the N1 and P3 components represent distinctive aspects of the decision process, with N 1 signifying the quantity of signal information received and P3 reflecting the certainty of the decision based upon that information.