Characteristics Of Event-related Potentials Research Articles

A clinical trial to validate event-related potential markers of Alzheimer's disease in outpatient settings

IntroductionWe investigated whether event-related potentials (ERP) collected in outpatient settings and analyzed with standardized methods can provide a sensitive and reliable measure of the cognitive deficits associated with early Alzheimer's disease (AD). MethodsA total of 103 subjects with probable mild AD and 101 healthy controls were recruited at seven clinical study sites. Subjects were tested using an auditory oddball ERP paradigm. ResultsSubjects with mild AD showed lower amplitude and increased latency for ERP features associated with attention, working memory, and executive function. These subjects also had decreased accuracy and longer reaction time in the target detection task associated with the ERP test. DiscussionAnalysis of ERP data showed significant changes in subjects with mild AD that are consistent with the cognitive deficits found in this population. The use of an integrated hardware/software system for data acquisition and automated data analysis methods make administration of ERP tests practical in outpatient settings.

Auditory evoked potentials in patients with major depressive disorder measured by Emotiv system.

In a previous study (unpublished), Emotiv headset was validated for capturing event-related potentials (ERPs) from normal subjects. In the present follow-up study, the signal quality of Emotiv headset was tested by the accuracy rate of discriminating Major Depressive Disorder (MDD) patients from the normal subjects. ERPs of 22 MDD patients and 15 normal subjects were induced by an auditory oddball task and the amplitude of N1, N2 and P3 of ERP components were specifically analyzed. The features of ERPs were statistically investigated. It is found that Emotiv headset is capable of discriminating the abnormal N1, N2 and P3 components in MDD patients. Relief-F algorithm was applied to all features for feature selection. The selected features were then input to a linear discriminant analysis (LDA) classifier with leave-one-out cross-validation to characterize the ERP features of MDD. 127 possible combinations out of the selected 7 ERP features were classified using LDA. The best classification accuracy was achieved to be 89.66%. These results suggest that MDD patients are identifiable from normal subjects by ERPs measured by Emotiv headset.

Motor imagery classification via combinatory decomposition of ERP and ERSP using sparse nonnegative matrix factorization

BackgroundBrain activities could be measured by devices like EEG, MEG, MRI etc. in terms of electric or magnetic signal, which could provide information from three domains, i.e., time, frequency and space. Combinatory analysis of these features could definitely help to improve the classification performance on brain activities. NMF (nonnegative matrix factorization) has been widely applied in pattern extraction tasks (e.g., face recognition, gene data analysis) which could provide physically meaningful explanation of the data. However, brain signals also take negative values, so only spectral feature has been employed in existing NMF studies for brain computer interface. In addition, sparsity is an intrinsic characteristic of electric signals. New methodTo incorporate sparsity constraint and enable analysis of time domain feature using NMF, a new solution for motor imagery classification is developed, which combinatorially analyzes the ERP (event related potential, time domain) and ERSP (event related spectral perturbation, frequency domain) features via a modified mixed alternating least square based NMF method (MALS-NMF for short). ResultsExtensive experiments have verified the effectivity the proposed method. The results also showed that imposing sparsity constraint on the coefficient matrix in ERP factorization and basis matrix in ERSP factorization could better improve the algorithm performance. Comparison with existing methodsComparisons with other eight representative methods have further verified the superiority of the proposed method. ConclusionsThe MALS-NMF method is an effective solution for motor imagery classification and has shed some new light into the field of brain dynamics pattern analysis.

P300-based brain-computer interface (BCI) event-related potentials (ERPs): People with amyotrophic lateral sclerosis (ALS) vs. age-matched controls

ObjectiveBrain-computer interfaces (BCIs) aimed at restoring communication to people with severe neuromuscular disabilities often use event-related potentials (ERPs) in scalp-recorded EEG activity. Up to the present, most research and development in this area has been done in the laboratory with young healthy control subjects. In order to facilitate the development of BCI most useful to people with disabilities, the present study set out to: (1) determine whether people with amyotrophic lateral sclerosis (ALS) and healthy, age-matched volunteers (HVs) differ in the speed and accuracy of their ERP-based BCI use; (2) compare the ERP characteristics of these two groups; and (3) identify ERP-related factors that might enable improvement in BCI performance for people with disabilities. MethodsSixteen EEG channels were recorded while people with ALS or healthy age-matched volunteers (HVs) used a P300-based BCI. The subjects with ALS had little or no remaining useful motor control (mean ALS Functional Rating Scale-Revised 9.4 (±9.5SD) (range 0–25)). Each subject attended to a target item as the items in a 6×6 visual matrix flashed. The BCI used a stepwise linear discriminant function (SWLDA) to determine the item the user wished to select (i.e., the target item). Offline analyses assessed the latencies, amplitudes, and locations of ERPs to the target and non-target items for people with ALS and age-matched control subjects. ResultsBCI accuracy and communication rate did not differ significantly between ALS users and HVs. Although ERP morphology was similar for the two groups, their target ERPs differed significantly in the location and amplitude of the late positivity (P300), the amplitude of the early negativity (N200), and the latency of the late negativity (LN). ConclusionsThe differences in target ERP components between people with ALS and age-matched HVs are consistent with the growing recognition that ALS may affect cortical function. The development of BCIs for use by this population may begin with studies in HVs but also needs to include studies in people with ALS. Their differences in ERP components may affect the selection of electrode montages, and might also affect the selection of presentation parameters (e.g., matrix design, stimulation rate). SignificanceP300-based BCI performance in people severely disabled by ALS is similar to that of age-matched control subjects. At the same time, their ERP components differ to some degree from those of controls. Attention to these differences could contribute to the development of BCIs useful to those with ALS and possibly to others with severe neuromuscular disabilities.

Classification of Familiar or Strange English Words using ERP

English vocabulary is the bricks of the language, and it’s important for English learners to handle accurately English words. Words recitation or dictation is the most widely used simple test form in vocabulary quiz. However, some students do less well in tests due to stress or even meltdown for exam. Here, we apply Brain fingerprinting (BF) in English vocabulary test. We studied the Event Related Potential (ERP) components elicited by familiar English words and strange English words, and used the ERP to decipher whether subject recognized the word presented in the screen. We analyzed the distinct ERP features related to different stimuli (familiar and strange words), and further speculated whether subjects recognized the presenting words from P300 by calculation correlations between different stimuli. The results of 8 subjects implied that familiar words recognized by subjects could be detected from P300 with accepted precision and recall. This means that BF provides a novel alternative for vocabulary quiz and opens the way for application in education, where concealed vocabulary related information stored in mind was detected by measuring brain signals, such as electroencephalography (EEG).

Detection of braking intention in diverse situations during simulated driving based on EEG feature combination

Objective. We developed a simulated driving environment for studying neural correlates of emergency braking in diversified driving situations. We further investigated to what extent these neural correlates can be used to detect a participantʼs braking intention prior to the behavioral response. Approach. We measured electroencephalographic (EEG) and electromyographic signals during simulated driving. Fifteen participants drove a virtual vehicle and were exposed to several kinds of traffic situations in a simulator system, while EEG signals were measured. After that, we extracted characteristic features to categorize whether the driver intended to brake or not. Main results. Our system shows excellent detection performance in a broad range of possible emergency situations. In particular, we were able to distinguish three different kinds of emergency situations (sudden stop of a preceding vehicle, sudden cutting-in of a vehicle from the side and unexpected appearance of a pedestrian) from non-emergency (soft) braking situations, as well as from situations in which no braking was required, but the sensory stimulation was similar to stimulations inducing an emergency situation (e.g., the sudden stop of a vehicle on a neighboring lane). Significance. We proposed a novel feature combination comprising movement-related potentials such as the readiness potential, event-related desynchronization features besides the event-related potentials (ERP) features used in a previous study. The performance of predicting braking intention based on our proposed feature combination was superior compared to using only ERP features. Our study suggests that emergency situations are characterized by specific neural patterns of sensory perception and processing, as well as motor preparation and execution, which can be utilized by neurotechnology based braking assistance systems.

Low-rank approximation based non-negative multi-way array decomposition on event-related potentials.

Non-negative tensor factorization (NTF) has been successfully applied to analyze event-related potentials (ERPs), and shown superiority in terms of capturing multi-domain features. However, the time-frequency representation of ERPs by higher-order tensors are usually large-scale, which prevents the popularity of most tensor factorization algorithms. To overcome this issue, we introduce a non-negative canonical polyadic decomposition (NCPD) based on low-rank approximation (LRA) and hierarchical alternating least square (HALS) techniques. We applied NCPD (LRAHALS and benchmark HALS) and CPD to extract multi-domain features of a visual ERP. The features and components extracted by LRAHALS NCPD and HALS NCPD were very similar, but LRAHALS NCPD was 70 times faster than HALS NCPD. Moreover, the desired multi-domain feature of the ERP by NCPD showed a significant group difference (control versus depressed participants) and a difference in emotion processing (fearful versus happy faces). This was more satisfactory than that by CPD, which revealed only a group difference.

Characteristics of event-related potentials in response to symbolical and alphabetical stimulation matrices used in a P300-based brain-computer interface

In order to create a P300-based brain-computer interface (BCI) (the so-called Farwell-Donchin paradigm, FD) with a symbol matrix used as a stimulus, we compared characteristics of event-related potentials (ERPs) in response to stimulation by 6 × 6 matrices composed of either pictogram symbols or Cyrillic alphabet characters. Nine healthy adults were examined in 18 experiments, during which 28-channel EEGs were recorded in the course of stimulation with matrices of these two types. The obtained ERP data, i.e., amplitudes and peak latencies of the ERP components N1, P3 (with the P3a and P3b sub-components), and N4 were compared and analyzed for different types of stimulation matrices. In at least seven out of nine subjects, P3a, P3b, and N4 ERP amplitudes were larger in response to the symbol matrix than to the character matrix, while N1 amplitudes were larger for the character matrix. For N1 and P3a, the ERP latencies were shorter for the symbol matrix, while for P3b and N4, they were longer for the character matrix. The topography of differential ERP responses to the two types of stimuli was analyzed using a series of paired t-tests. Differences of ERP component amplitudes were determined individually for each of the 28 channels; next, for each site, absolute t-test values were summed for all nine subjects. For all ERP components studied, the t-test for peak amplitudes in response to target and non-target letters identified two separate areas with distinct lateralization. ERP responses to target and non-target symbols differed most in transversely extended areas. Finally, the yield surface of differential response to target letters and target symbols had a complex topography.

Frequency characteristics of cortical activity associated with perturbations to upright stability

Cortical evoked potentials are evident in the control of whole-body balance reactions in response to transient instability. The focus of this work is to continue to advance understanding of the potential cortical contributions to bipedal balance control. Temporally unpredictable postural perturbations evoke a negative potential (N1), which has drawn parallels to error-related negativity (ERN) as well as visual and auditory evoked N1 responses. The mechanism underlying the generation of event-related potentials (ERPs) has been a matter of debate for the past few decades. While the evoked model proposes that ERPs are generated by the addition of fixed latency and fixed polarity responses, the phase reorganization model suggests that ERPs are the result of stimulus-induced phase reorganization of the ongoing oscillations. Previous studies have suggested phase reorganization as a possible mechanism in auditory N1, visual N1 and error-related negativity (ERN). The purpose of the current study was to explore the frequency characteristics of the cortical responses to whole-body balance perturbations. Perturbations were evoked using a lean and release protocol. The results revealed a significant power increase and phase-locking of delta, theta, alpha, and beta band activity during perturbation-evoked N1. This may suggest that the stimulus-induced phase reorganization of the ongoing electroencephalographic (EEG) activity could account for the features of cortical ERPs in response to perturbation of upright stability.

Is what goes in what comes out? Encoding and retrieval event-related potentials together determine memory outcome.

Understanding memory function amounts to identifying how events (cognitive and neural) at study eventually influence events at test. Many of the proposed cognitive correlates of memory-related event-related potentials (ERPs) at study resemble proposed cognitive correlates of other memory-related ERPs, recorded at test. We wondered whether a given known ERP feature at study might in fact reflect an effective-encoding process that is, in turn, tapped by another specific ERP feature, recorded at test. To this end, we asked which pairs of known memory-related ERP features explain common variance across a large sample of participants, while they perform a word-recognition task. Two early ERP features, the Late Positive Component (study) and the FN400 (test), covaried significantly. These features also correlated with memory success (d' and response time). Two later ERP features, the Slow Wave (study) and the Late Parietal Positivity (test), also covaried when lures were incorporated into the analysis. Interestingly, these later features were uncorrelated with memory outcome. This novel approach, exploiting naturally occurring subject variability (in strategy and ERP amplitudes), informs our understanding of the memory functions of ERP features in several ways. Specifically, they strengthen the argument that the earlier ERP features may drive old/new recognition (but perhaps not the later features). Our findings suggest the Late Positive Component at study, in some degree, may cause the FN400 to increase at test, together producing effective recognition memory. The Slow Wave at study appears to relate the Left Parietal Positivity at test, but these may play roles in more complex memory judgments and may be less critical for simple old/new recognition.

Neural correlates of inferring speaker sincerity from white lies: An event-related potential source localization study

During social interactions, listeners weigh the importance of linguistic and extra-linguistic speech cues (prosody) to infer the true intentions of the speaker in reference to what is actually said. In this study, we investigated what brain processes allow listeners to detect when a spoken compliment is meant to be sincere (true compliment) or not (“white lie”). Electroencephalograms of 29 participants were recorded while they listened to Question–Response pairs, where the response was expressed in either a sincere or insincere tone (e.g., “So, what did you think of my presentation?”/“I found it really interesting.”). Participants judged whether the response was sincere or not. Behavioral results showed that prosody could be effectively used to discern the intended sincerity of compliments. Analysis of temporal and spatial characteristics of event-related potentials (P200, N400, P600) uncovered significant effects of prosody on P600 amplitudes, which were greater in response to sincere versus insincere compliments. Using low resolution brain electromagnetic tomography (LORETA), we determined that the anatomical sources of this activity were likely located in the (left) insula, consistent with previous reports of insular activity in the perception of lies and concealments. These data extend knowledge of the neurocognitive mechanisms that permit context-appropriate inferences about speaker feelings and intentions during interpersonal communication.

EEG study on affective valence elicited by novel and familiar pictures using ERD/ERS and SVM-RFE

EEG signals have been widely explored in emotional processing analyses, both in time and frequency domains. However, in such studies, habituation phenomenon is barely considered in the discrimination of different emotional responses. In this work, spectral features of the event-related potentials (ERPs) are studied by means of event-related desynchronization/synchronization computation. In order to determine the most relevant ERP features for distinguishing how positive and negative affective valences are processed within the brain, support vector machine-recursive feature elimination is employed. The proposed approach was applied for investigating in which way the familiarity of stimuli affects the affective valence processing as well as which frequency bands and scalp regions are more involved in this process. In a group composed of young adult women, results prove that parietooccipital region and theta band are especially involved in the processing of novelty in emotional stimuli. Furthermore, the proposed method has shown to perform successfully using a moderated number of trials.

Association between waking electroencephalography and cognitive event-related potentials in patients with obstructive sleep apnea

Sleepiness, cognitive deficits, abnormal event-related potentials (ERP), and slowing of the waking electroencephalography (EEG) activity have been reported in patients with obstructive sleep apnea (OSA). Our study aimed at evaluating if an association exists between the severity of ERP abnormalities and EEG slowing to better understand cerebral dysfunctions in OSA. Twelve OSA patients and 12 age-matched controls underwent an overnight polysomnographic recording, an EEG recording of 10 min of wakefulness, and an auditory ERP protocol known to specifically recruit attention. P300 and P3a ERP components were measured as well as the spectral power in each frequency band of the waking EEG. Pearson product moment correlations were used to measure associations between ERP characteristics and EEG spectral power in OSA patients and control subjects. A positive correlation between the late P300 amplitude and θ power in the occipital region was observed in OSA subjects (P<.01). A positive correlation was also found between P3a amplitude and β1 power in central region in OSA subjects (P<.01). No correlation was observed for control subjects. ERP abnormalities observed in an attention task are associated with a slowing of the waking EEG recorded at rest in OSA.

Single-trial ERP analysis reveals facial expression category in a three-stage scheme

Emotional faces are salient stimuli that play a critical role in social interactions. Following up on previous research suggesting that the event-related potentials (ERPs) show differential amplitudes in response to various facial expressions, the current study used trial-to-trial variability assembled from six discriminating ERP components to predict the facial expression categories in individual trials. In an experiment involved 17 participants, fearful trials were differentiated from non-fearful trials as early as the intervals of N1 and P1, with a mean predictive accuracy of 87%. Single-trial features in the occurrence of N170 and vertex positive potential could distinguish between emotional and neutral expressions (accuracy=90%). Finally, the trials associated with fearful, happy, and neutral faces were completely separated during the window of N3 and P3 (accuracy=83%). These categorization findings elucidated the temporal evolution of facial expression extraction, and demonstrated that the spatio-temporal characteristics of single-trial ERPs can distinguish facial expressions according to a three-stage scheme, with “fear popup,” “emotional/unemotional discrimination,” and “complete separation” as processing stages. This work constitutes the first examination of neural processing dynamics beyond multitrial ERP averaging, and directly relates the prediction performance of single-trial classifiers to the progressive brain functions of emotional face discrimination.

A multi-signature brain–computer interface: use of transient and steady-state responses

Objective. The aim of this paper was to increase the information transfer in brain–computer interfaces (BCI). Therefore, a multi-signature BCI was developed and investigated. Stimuli were designed to simultaneously evoke transient somatosensory event-related potentials (ERPs) and steady-state somatosensory potentials (SSSEPs) and the ERPs and SSSEPs in isolation. Approach. Twelve subjects participated in two sessions. In the first session, the single and combined stimulation conditions were compared on these somatosensory responses and on the classification performance. In the second session the on-line performance with the combined stimulation was evaluated while subjects received feedback. Furthermore, in both sessions, the performance based on ERP and SSSEP features was compared. Main results. No difference was found in the ERPs and SSSEPs between stimulation conditions. The combination of ERP and SSSEP features did not perform better than with ERP features only. In both sessions, the classification performances based on ERP and combined features were higher than the classification based on SSSEP features. Significance. Although the multi-signature BCI did not increase performance, it also did not negatively impact it. Therefore, such stimuli could be used and the best performing feature set could then be chosen individually.

Evaluating the cognitive consequences of mild traumatic brain injury and concussion by using electrophysiology

Mild traumatic brain injury (MTBI), often referred to as concussion when it occurs in sports, produces persistent cognitive problems in at least 15% of patients. Unfortunately, conventional neuropsychological tests usually yield results within normal limits in this population. The main objective of this event-related potential (ERP) study was to understand brain functioning during the performance of a working memory (WM) task in patients who have sustained an MTBI, mostly due to motor vehicle accident or sports concussion. This study also aimed for a better understanding of the association between brain functioning as measured with ERP, behavioral performance on the WM task, postconcussion symptoms, type of injury (that is, sports concussion vs other types), and time since the injury. Forty-four patients with MTBI (7.6 ± 8.4 months postinjury) were tested on a visual WM task with simultaneous recording of ERP, and were compared with 40 control volunteers who were their equivalent for age and sex. Amplitude and latency of frontal (N200 and N350) and parietal (P200 and P300) ERP waves were measured and were compared between groups. Correlation analyses were also performed between ERP characteristics, clinical variables, and behavioral performance. A significant group difference was found for behavioral performance on the WM task, in which the MTBI group had a lower percentage of correct answers than the control group (p < 0.05). The patients with MTBI also had smaller amplitudes of both frontal N350 and parietal P300 ERP components when compared with control volunteers (p < 0.05). No changes were found for latency of ERP components. Smaller ERP amplitudes were associated with slower reaction times and worse accuracy on the WM task among patients with MTBI (p < 0.05). Types of injury (that is, sports concussion vs other mechanisms) were not associated with different ERP characteristics. Abnormal ERP results are observed in patients after MTBI or sports concussion, even for those in the nonacute stage after their injury. Current standard clinical evaluations most often fail to detect cerebral dysfunction after MTBI, even when patients or athletes report symptoms. Clinicians should be aware that patients with MTBI, including sports concussion, probably have underlying mild but persistent cerebral dysfunctions that require further investigation.

Ontogenetic Modifications of Evoked EEG Activity in 6- to 16-Year-Old Children and Teenagers

We studied peculiarities of the age-related dynamics of amplitude/time characteristics of evoked EEG potentials (EPs) and event-related potentials (ERPs) recorded within the framework of the go/no go paradigm in 6- to 16-year-old children and teenagers. We found a number of significant correlations between age of the tested persons and characteristics of EPs and ERPs. With age, latencies of the components Р1 (in the frontal and central regions), N1 (practically throughout the head surface), Р2 (in the frontal, temporal, and central leads of the left hemisphere, as well as in the temporal and occipital leads of the right hemisphere), and wave Р300 (in the central regions of both hemispheres) demonstrated clear trends toward shortening. On the whole, the amplitudes of Р2 (in the frontal, central, and parietal regions of both hemispheres and in the right temporal region), N1-Р2 (in the frontal, central, and parietal regions), and contingent negative variation (in the left central lead) increased with age; we also observed a decrease in the amplitude of the N2 component, which reached a significant value in the left occipital lead. In addition, the working efficiency increased, which was manifested in decreases in the mean delay of the sensorimotor reaction, number of errors of omission of significant stimuli, and number of erroneous pushes. Therefore, the amplitude/time characteristics of some components of EPs and ERPs in 6- to 16-year-old children and teenagers demonstrate obvious age-dependent modifications; these shifts reflect the processes of maturation of the brain and formation of cognitive functions.

Correlations of the Characteristics of Event-Related Potentials with the Contents of Some Chemical Elements in the Organism of Urban Teenager Residents (According to a Longitudinal Observation)

We examined correlations of the characteristics of event-related potentials (ERPs) recorded according to a two-stimulus go/no-go paradigm (with measurements of the reaction time) with the contents of seven essential and potentially harmful elements in a group of teenagers (15 boys and 15 girls). Electrophysiological examinations and measurements of the contents of chemical elements in samples of a biologically stable tissue (hairs) using X-ray fluorescence analysis were carried out during three years (age of the subjects increased from 12 to 14 years). Nonparametric correlation analysis by Spearman was used. In 12- and 13-year-old teenagers, the element balance was characterized by deficiencies of Ca, Fe, and Mn; in 14-year-old subjects, an excess of Ni and a noticeably excessive content of Pb in hairs were found. Statistically significant correlations (0.37 < r < 0.64) of the amplitudes and latencies of ERPs with the contents of the above elements were observed (20 and 7 cases, respectively). Correlations of the ERP parameters with the contents of chemical elements were more clearly expressed in the case of deficiencies of essential elements in 12- and 13-year-old teenagers; in older subjects (14 years), manifestations of such correlations demonstrated a trend toward limitation.

Characterization and Robust Classification of EEG Signal from Image RSVP Events with Independent Time-Frequency Features

This paper considers the problem of automatic characterization and detection of target images in a rapid serial visual presentation (RSVP) task based on EEG data. A novel method that aims to identify single-trial event-related potentials (ERPs) in time-frequency is proposed, and a robust classifier with feature clustering is developed to better utilize the correlated ERP features. The method is applied to EEG recordings of a RSVP experiment with multiple sessions and subjects.The results show that the target image events are mainly characterized by 3 distinct patterns in the time-frequency domain, i.e., a theta band (4.3 Hz) power boosting 300–700 ms after the target image onset, an alpha band (12 Hz) power boosting 500–1000 ms after the stimulus onset, and a delta band (2 Hz) power boosting after 500 ms. The most discriminant time-frequency features are power boosting and are relatively consistent among multiple sessions and subjects.Since the original discriminant time-frequency features are highly correlated, we constructed the uncorrelated features using hierarchical clustering for better classification of target and non-target images. With feature clustering, performance (area under ROC) improved from 0.85 to 0.89 on within-session tests, and from 0.76 to 0.84 on cross-subject tests. The constructed uncorrelated features were more robust than the original discriminant features and corresponded to a number of local regions on the time-frequency plane. Availability: The data and code are available at: http://compgenomics.cbi.utsa.edu/rsvp/index.html

Estimating workload using EEG spectral power and ERPs in the n-back task

Previous studies indicate that both electroencephalogram (EEG) spectral power (in particular the alpha and theta band) and event-related potentials (ERPs) (in particular the P300) can be used as a measure of mental work or memory load. We compare their ability to estimate workload level in a well-controlled task. In addition, we combine both types of measures in a single classification model to examine whether this results in higher classification accuracy than either one alone. Participants watched a sequence of visually presented letters and indicated whether or not the current letter was the same as the one (n instances) before. Workload was varied by varying n. We developed different classification models using ERP features, frequency power features or a combination (fusion). Training and testing of the models simulated an online workload estimation situation. All our ERP, power and fusion models provide classification accuracies between 80% and 90% when distinguishing between the highest and the lowest workload condition after 2 min. For 32 out of 35 participants, classification was significantly higher than chance level after 2.5 s (or one letter) as estimated by the fusion model. Differences between the models are rather small, though the fusion model performs better than the other models when only short data segments are available for estimating workload.