Patent Application
20120310105
Oscillations in the beta frequency band (13-30 Hz) range in sensorimotor cortex have long been associated with movement. The hallmark of beta activity in cortical motor areas is a pattern of peri-movement beta suppression (relative to the resting state) followed by an increase in beta activity post-movement. This pattern has been observed in local field potential (LFP) and EEG recordings from cortical motor areas in humans and monkeys performing simple single or repetitive movements, such as finger presses or wrist movements in response to a cue. Typically, beta activity fell to a minimum following cue presentation or the start of movement execution, and reached a maximum immediately following movement.
In view of this pattern, beta was labeled an “idling” frequency, with beta band oscillations theoretically representing a minimal energy state that the brain enters in the absence of processing. However, the finding of pathologically excessive beta band synchrony in patients with Parkinson's disease (PD) and animal models of Parkinsonism, led to the idea that beta might be more specifically anti-kinetic or anti-movement in nature. Rather than representing idling, this theory suggests that beta oscillations (or the mechanisms that give rise to them) actively determine if movement is to occur. The inverse correlation between beta activity and movement has been used to account for the increase in beta activity observed during sustained motor output, including the maintenance of a precision grip or the application of a constant force. These findings give further support to the notion of beta activity representing not the absence of motor output, but rather a decreased likelihood of changing the existing motor output. More recently, this connection between beta activity and preserving the status quo was extended to cognitive processes, where the role of beta oscillations in brain regions involved in cognitive processing is to preserve the current state. According to this idea, beta oscillations in a given brain region should be higher when that region does not anticipate an impending change in motor/cognitive output or set, and so maintains its current pattern of activity.
Despite their theoretical appeal, none of these interpretations fits well with the core phenomenon of maximal beta activity occurring in cortical motor areas immediately following movement. If beta activity reflects idling or conservation of the current state or output, why is it highest following movement, but relatively low during rest (behavioral “idling”)?
Excessive beta band (13-30 Hz) activity in cortico-basal ganglia circuits is recognized as a pathophysiological signature of Parkinson's disease. Yet, the function of beta oscillations in the healthy primate brain is not clear. Current theories link beta activity to idling or the preservation of the current state of cortical areas.
The application is based in part on the discovery that beta oscillations in the local field potentials (LFPs) in frontal cortex and striatum of monkeys performing arm movements are not directly related to movement, but to the completion of behavioral performance, and the disengagement of the brain region from direct involvement in that behavior. Beta oscillations occurred in brief, spatially localized bursts that were most pronounced following movement or task performance. Beta bursts were detected based on their relative contribution to the spectrum of the LFP signal, as opposed to their overall amplitude. The rates and power of beta bursts differed across brain regions studied. Post-performance beta burst rates and power tracked the details of the preceding task performance differently by brain region. Moreover, in striatum and prefrontal cortex, beta burst rates were higher following correct trials than errors, and the bursts at pairs of sites across these regions were coherent. Beta oscillations represent post-performance reinforcement of the network activity that led to the desired behavioral outcome obtained immediately prior to the post performance beta activity.
Moreover, the timing of beta bursts is substantially random during task performance, and become significantly more coherent (e.g., coordinated) across brain regions after task performance. Relative measurement of the timing of bursts therefore can be used to monitor task engagement in real time. In addition, burst rate can be used to monitor task engagement in real time. Because beta oscillations correlate to proper task completion, beta oscillations can be used to monitor an individual's engagement in or premature disengagement from a task. Similarly, beta oscillations can be used to diagnose learning disabilities where an individual fails to engage or to stay engaged in a cognitive task. The methods disclosed herein also can be used to diagnose cognitive conditions, such as but not limited to Parkinson's Disease, that are typified by abnormal beta oscillations (e.g., hyperactivity or hypoactivity).
The invention provides, in part, a method of assessing whether an individual is engaged in a task. The method includes the steps of measuring an EEG, for a plurality of frequencies, of the individual prior to, during and after task performance using a plurality of electrodes to generate a data set comprising frequencies and power; selecting, using a computer, frequencies in the beta frequency range from the EEG data set; comparing, using a computer, the power in the frequencies in the beta frequency range prior to, during and after task performance; and determining whether the person was engaged in the task in response to the comparison of the power in the beta frequency range prior to, during, and after task performance.
The invention also provides, in part, a method of assessing whether an individual is engaged in a task. The method includes the steps of measuring an EEG, for a plurality of frequencies, of the individual prior to, during and after task performance using a plurality of electrodes to generate a data set comprising frequencies and burst rate; selecting, using a computer, frequencies in the beta frequency range from the EEG data set; comparing, using a computer, the burst rate in the frequencies in the beta frequency range prior to, during and after task performance; and determining whether the person was engaged in the task in response to the comparison of the burst rate in the beta frequency range prior to, during, and after task performance.
The invention also provides, in part, a method of assessing whether a brain region is engaged in a task, the method comprising: monitoring, using a computer, brain activity in the beta frequency range prior to, during, and after performance of a task, wherein a burst in beta frequency activity after performance of the task is indicative that the brain region has disengaged from performance of the task.
Embodiments of the methods can include one or more of the following features:
The data set can be a sample obtained from the patient.
The comparison is made by plotting frequency and power against time.
The measurement of task engagement is made for a plurality of brain regions.
The method includes the additional step of measuring coherence in the beta frequency range between brain regions of the plurality of brain regions.
The comparison is performed by a computer by transforming the data. For example, the transformation can be performed by a filter. The transformation can be a HHT transform and/or a FFT transform.
The individual is considered to have been engaged in the task if beta frequencies prior to task performance during resting have decreased power relative to the beta frequencies after task performance.
The beta frequency range is about 13 Hz to about 30 Hz.
The brain region is selected from the group consisting of primary motor and dorsal premotor cortex, dorsolateral prefrontal cortex, caudate nucleus, and putamen.
The invention also provides, in part, a method of diagnosing a condition affecting movement or thought preparation and movement cessation in an individual. The method can include measuring, using an electroencephalographic computer, brain activity in the beta frequency range prior to, during, and post performance of a task, wherein a burst in beta frequency activity during the task completion time in a normal brain region is indicative that the brain region has prematurely disengaged from performance of the task.
In some embodiments, the method can include the step of measuring beta frequency activity in primary motor and dorsal premotor cortex, dorsolateral prefrontal cortex, caudate nucleus, and putamen.
The invention also provides, in part, a method of diagnosing a condition affecting movement or thought preparation and cessation in an individual. The method can include measuring, using a computer, brain activity in the beta frequency range before, during and after performance of a task, wherein bursts in beta frequency activity that are not coherent with bursts in other brain regions, as compared to normal brains, are indicative that brain region activity is abnormally coordinated.
In some embodiments, the method can include the step of measuring the coherence of beta frequency activity in at least two brain regions.
In some embodiments, the at least two brain regions comprise striatum and prefrontal cortex.
In some embodiments, the at least two brain regions comprise caudate nucleus and dorsolateral prefrontal cortex.
The invention also provides, in part, an apparatus for assessing whether an individual is engaged in a task. The apparatus includes: a plurality of electrodes for placement on the head of the individual; and a computer system. The computer system includes (e.g., on one or more suitably programmed mediums) an EEG module, in communication with the plurality of electrodes, for measuring an EEG, for a plurality of frequencies, of the individual prior, during and post task using a plurality of electrodes to generate a data set of frequencies and power; a selection module in communication with the EEG module for selecting frequencies in the beta frequency range from the EEG data set; a comparator for comparing the power in the frequencies in the beta frequency range before, during and post task. The system also can include a display, in electrical communication with the comparator, for showing whether the person was engaged in the task, in response the comparison of the power in the beta frequency range before, during, and post task.
The invention also provides, in part, a method of assessing the efficacy of a drug used to treat an individual with a cognitive disability typified by hyperactive beta oscillations. The method can include the steps of measuring an EEG at a first time, for a plurality of frequencies, of the individual prior to, during and after task performance using a plurality of electrodes to generate a first data set comprising frequencies and power, the first time being prior to administration of the drug; measuring an EEG at a second time, for a plurality of frequencies, of the individual prior to, during and after task performance using a plurality of electrodes to generate a second data set comprising frequencies and power, the second time being after administration of the drug; selecting, using a computer, frequencies in the beta frequency range from the first EEG data set and the second EEG data set; comparing, using a computer, the power in the frequencies in the beta frequency range prior to, during and after task performance in the first EEG data set and in the second EEG data set; and determining whether administration of the drug decreased beta frequency oscillations.
In some embodiments, cognitive disability is selected from the group consisting of: Parkinson's Disease, obsessive compulsive disorder, autism, attention deficit disorder, attention deficit hyperactivity disorder, and post traumatic stress disorder.
The figures are not necessarily to scale, emphasis instead generally being placed upon illustrative principles. The figures are to be considered illustrative in all aspects and are not intended to limit the invention, the scope of which is defined only by the claims.
Beta oscillations in the local field potentials (LFPs) in frontal cortex and striatum of monkeys performing arm movements are not directly related to movement, but to the offset of behavioral performance. In all brain regions, beta oscillations occurred in brief, spatially localized bursts that were most pronounced following movement or task performance. Beta bursts were detected based on their relative contribution to the spectrum of the LFP signal, as opposed to their overall amplitude. The rates and power of beta bursts differed across brain regions. Post-performance beta burst rates and power tracked the details of the preceding task performance differently by brain region. Moreover, in striatum and prefrontal cortex, beta burst rates were higher following correct trials than errors, and the bursts at pairs of sites across these regions were coherent. Based on our results, beta oscillations represent post-performance reinforcement of the network activity that led to the desired behavioral outcome, movement or result, obtained immediately prior to the beta activity.
Prominent post-movement increases in beta power may be a ubiquitous signature of post-performance processing in the brain. This hypothesis was tested in monkeys by recording LFPs in cortico-striatal regions known to be directly involved in motor control, movement sequencing and executive function. We found that LFPs in frontal cortex and striatum were characterized by brief, spatially localized episodes, during which beta frequencies disproportionately dominated the LFP spectrum. These beta bursts were most pronounced following behavioral performance, when they were modulated by specific features of the preceding behavior, as well as the outcome of that behavior. Based on our results, we propose that the role of beta oscillatory activity might not be to preserve the current state or behavioral output of individual brain regions, but to preserve or reinforce the cortico-striatal network dynamics that led to the desired outcome obtained immediately prior.
Based on human electroencephalogram (EEG) and magnetoencephalograpy (MEG), and based on monkey local field potential (LFP) experiments, using simple single movement tasks—such as wrist flexion and precision grip—activity in the beta band (roughly 13-30 Hz, depending on the species, subject and brain region) has been linked to brain idling. Studying visually cued single and sequential arm movements, as well as rest periods, it was discovered that across regions of the frontal cortex and striatum of monkeys, beta activity changes on a considerably faster time-scale than previously thought, appearing in brief (˜150 ms) bursts of oscillations, during which beta-band power dominates the spectrum. These changes in the EEG or LFP can be detected with high temporal resolution using a custom-modified version of the Hilbert-Huang transform (HHT), making it possible to track beta activity in response to individual behavioral events, without averaging the results over multiple trials. It is hypothesized that these beta bursts represent packets of communication or local processing. The timing of these bursts in a given brain region appears to correlate with the end of the brain's involvement in behavior. During these times, increased coherence was detected in the beta band between frontal cortex and striatum. Based on these findings, it is hypothesized that, in contrast to the idling interpretation, beta activity represents packets of communication or local processing, which become prominent when a brain region disengages from behavior. Beta bursts might thus function as “offline” data dumps and/or can facilitate periods of network tuning/updating following a brain region's engagement in a task.
Thus, bursts in the beta frequency band of signals from the human or animal brain can be used to detect the precise temporal boundaries of individual brain regions' engagement in a given task, as well as the depth (or level) of that engagement. As a result, the present invention can be used for a number of applications including, by way of non-limiting example:
In order to study beta burst phenomena, analytical methods were developed that are based on the Hilbert transform, rather than the Fourier transform. This new approach is based largely on recent advances in the application of the Hilbert transform to nonlinear data, known as the Hilbert-Huang transform (HHT) (Huang et al., 1998 & 2003). Using a process termed Emprical Mode Decomposition (EMD), the HHT constructs Intrinsic Mode Functions (IMFs) that are AM-FM analogues of Fourier components of the Fourier transform. In contrast to the potentially infinite number of Fourier components needed to recreate a waveform, the IMFs capture the intrinsic timescales of the raw data in compact form, enabling faithful reconstruction of the raw signal with only a handful of IMFs.
Among the strengths of the HHT is its definition of a meaningful instantaneous frequency, which gives rise to the possibility of analyzing the short bursts of beta activity that we have observed in the raw LFPs. The HHT makes it possible detect the timing, duration, frequency and amplitude of bursts much more accurately than can be done with traditional Fourier-based methods. One embodiment of the HHT was implemented using the Matlab (Mathworks, Natick, Mass.) software package, in order to construct IMFs from raw brain signals. A custom-made algorithm was developed to detect consistent results obtained from the HHT, by performing the EMD repeatedly with different parameter values and defining confidence limits. The entire processing pipeline was performed from end to end, including making composites, for each iteration of parameter values. We defined the CLs as +/−2*SEM away from the mean over all iterations. The mean over iterations then became our composite HHT waveform for subsequent analysis.
A separate algorithm was created for determining the contribution of the IMFs to the content of individual frequency bands, by summing IMFs, appropriately weighted by frequency, in order to construct an HHT composite (an analogue of the band-pass filtered signal). “Appropriately weighted by frequency” means weighted by
exp(−(f−mu)̂pow)/(2*sigmâpow))
where mu=mean(freqlim) and sigma=mu−freqlim(1).
Finally, an algorithm was developed for detecting the boundaries of bursts in the HHT composite, by assessing at each time point the goodness-of-fit of the HHT composite to the LFP, and imposing frequency and phase constraints on the HHT composite, as well as requiring it to behave in sinusoidal-like fashion (e.g., a zero-crossing must exist between successive extrema). The algorithm is described as follows:
The discovery of spatially localized, brief, discrete bursts of beta-frequency oscillations in the LFPs recorded from sites across frontal cortex and striatum led us to interpret the trial-averaged beta-band power as indicating the probability distribution of observing a burst of beta oscillations dominating the spectrum of the LFP at each time-point in a given trial. In all brain areas, beta bursts were suppressed (in terms of amplitude and rate of occurrence, relative to rest period levels) during the initial cue presentation period, prior to the monkey's initiation of arm movement. In sequential movement tasks, bursting activity tended to resume during the inter-movement hold periods (or inter trial intervals (ITI)), particularly when these were long (1.4 s), as opposed to short (0.7 s). This pattern of peri-movement burst suppression along with increased bursting activity during relatively long breaks between sequential movements suggests that the modulation of beta bursts can be used to track engagement in task performance. This idea is further supported by an analysis of the timing of peak bursting activity during task performance. In all brain regions, the peak rates and peak amplitudes of beta bursts were significantly higher than the average values during periods of rest. Whereas beta bursts in cortical motor areas occurred most frequently and were highest in amplitude immediately following the last movement in a trial, the rate and amplitude of beta bursts in dorsolateral prefrontal cortex (dlPFC) and striatum peaked in the period immediately following the trial. Given the presumed roles of these different brain regions in task performance, we hypothesize that the beta bursts in a given brain region increase in amplitude and rate after that brain region disengages from task performance.
The modulation of beta activity in the absence of movement also was studied, under two different conditions: first, during periods of rest, in which the monkey sat quietly for several minutes; and second, during trials in which visual cues that would normally instruct upcoming arm movements were presented in a different context, so that the monkey scanned the visual cues, but did not respond with any arm movement. It was found that the time-course of beta activity in the latter case differed from the beta activity during rest periods. In contrast to previous reports, the changes in beta in response to the visual cues in the absence of subsequent arm movement, and following reward delivery were similar to those during the movement tasks. Thus, these changes in beta activity correspond to periods of the brain's engagement with external stimuli and reward, representing cognitive processing regardless of motor output. Collectively, the changes in beta activity signal the engagement (low beta activity relative to rest) and subsequent disengagement (high beta activity relative to rest) from behavioral task performance, regardless of motor output.
It has been further found that across brain sites, there is a strong, positive correlation between the average rate of post-engagement beta bursts and the number of movements or visual cues in the task. This finding suggests that changes in beta bursts not only signal the end of an engagement in a task, as opposed to ongoing periods of rest, but are directly related to the cognitive load during task performance; the more demanding the task, the greater the number of bursts following task performance.
In support of this conclusion, two tasks were compared in which the monkey performed sequences of three arm movements each. The monkey performed identical movement sequences in both tasks. The tasks differed only with respect to the timing of the presentation of the instructional cues. In the first task, the cues instructing the entire sequence were presented simultaneously at trial start and remained unchanged throughout the trial. In the second, only the cue instructing the first movement appeared at trial start, with each successive cue (instructing the next movement in the sequence) appearing at the offset of the preceding movement. We found that the burst rates during the trial were higher in the simultaneously cued task than the sequentially cued task. This pattern was reversed in the interval immediately following the trial. In those trials in which the cues instructing subsequent arm movements were presented in advance of the movements, the brain regions in question beta-burst relatively more than they did during trials in which the cues instructing upcoming movements were not presented ahead of time. Remarkably, in the period immediately following the trials (when the brain regions presumably disengaged from task performance), there were more beta bursts in the sequentially cued task than in the simultaneously cued task.
This pattern of results suggests that the brain regions beta-burst opportunistically, depending on the information available to them and the ongoing demands of the task. If a region can disengage briefly from task performance, then it will take the opportunity to do so, as indicated by the occurrence of beta bursts. However, the fact that the brain regions switch from bursting more during the trial, when the cues are presented ahead of time, to bursting more following the trial, when the cues are not presented ahead of time, argues against the idea that, when the brain regions disengage, they are simply idling. Rather, the beta bursts that occur as a brain region disengages from task performance appear to be involved in the processing of that performance, potentially integrating the cues, movements and outcome, and coordinating these results across brain sites. This proposed function of beta bursts would be relevant to adaptive learning in a dynamic environment, in which the brain must monitor and optimize the relationship between stimuli and responses in order to maximize its chances of obtaining desired outcomes in the future.
The Single Movement Task (Sing) requires the monkey to perform a single center-out-center joystick movement in response to a single peripheral cue, following a short or long self-timed hold period (
The Sequentially Cued Sequential Movement Task (Seq) requires the monkey to perform sequences of 3 joystick movements (M1-3), each preceded by a short or long self-timed center hold period (H1-3). Each movement is cued, in turn, at the start of the preceding self-timed hold interval (
The Simultaneously Cued Sequential Movement Task (Sim) requires the monkey to perform sequences of 3 movements (
In each block of the Sim or Seq task the monkey must correctly complete 32 trials, each requiring a distinct spatiotemporal sequence of joystick movements. Each sequence can be broken down conceptually into spatial and temporal templates. There are eight possible temporal templates (H1: short, H2: short, H3: short, through H1: long, H2: long, H3: long) and 8×7×6=336 spatial templates. At the start of every recording session, a unique set of 32 sequences to be used that day is constructed according to a prescription that ensures that each of the 8 temporal templates occurs exactly 4 times in every block of trials. Since the monkeys will have been trained extensively on all combinations of spatial and temporal templates prior to the recording phase of the experiment, little learning is expected to occur within any given recording session.
The Simultaneously Cued Single Movement Task (SimSing) is visually identical to the Sim task, but the monkey must perform only the first of the three cued center-out-center joystick movements (
The Simultaneously Cued No Movement Task (NoGo) is also visually identical to the Sim task, but the monkey must withhold all joystick movement in order to receive reward (
Beta frequency patterns are correlated to task engagement.
In addition, LFPs were recorded simultaneously from multiple sites in primary motor and dorsal premotor cortex (M1PMC), dorsolateral prefrontal cortex (dlPFC), caudate nucleus (CN) and putamen (Put) of two Rhesus monkeys. The monkeys had been trained extensively to perform single and sequential joystick movements in response to visual cues (
Striking differences were observed between the time-course of LFP power in the beta band across brain regions (
The timing of the rebound in power in dlPFC and striatum was dramatically different from what it was in M1PMC. Instead of peaking in the post-movement period, as in M1PMC, beta power in dlPFC and striatum peaked in the post-trial period, following reward delivery and the subsequent offset of the visual cues (
These results argue against a simple relationship between movement and the modulation of beta activity, even in M1PMC. Nevertheless, they were obtained during the performance of tasks requiring overt motor responses. Does the suppression-rebound pattern of modulation in beta power occur only during trials involving movement? To answer this question, the timing of the peaks in beta power was analyzed during trials of a third task, 0M3T, in which the monkeys were presented with the same visual cues as in the 3M3T task (
The time-course of the trial-averaged beta power in relation to task performance suggests the existence of sustained beta oscillations either post-movement (in M1PMC) or post-trial (in dlPFC and striatum). Sustained oscillations, if present during periods of rest or “steady-state” behavior, would support current interpretations of beta activity as an indicator of idling or of state-preserving processes. However, the results so far have been based on trial averages. In order to detect the occurrence of sustained beta oscillations during task performance, individual trials were analyzed. Surprisingly, no single trial resembled the trial average, in that there were no periods of sustained high-amplitude beta oscillations. Rather, each trial was characterized by brief (˜150 ms) episodes during which beta frequency oscillations dominated the spectrum (
The bursts in the beta composite HHT were detected based on the relative contribution of the composite (at each time point) to the broadband LFP signal (5-50 or 5-100 Hz).
When the contribution rose above a given threshold, potential bursts were marked. Burst boundaries were detected by proceeding forward and backward in time from each supra-threshold local maximum in the contribution, until any of three constraints were violated: the phase of the composite HHT shifted at a rate above an allowable threshold; the composite HHT failed to cross zero between successive extrema; or the level of the contribution of the composite to the broadband LFP fell below an allowable threshold.
All thresholds were computed on a per electrode basis, electrode by electrode, based on comparing the recorded LFP on each electrode to the composite HHT of simulated pink noise.
Beta oscillations were not only temporally discrete, but were also localized in space. In each of the four brain regions we studied, the peak cross-covariance between the envelopes of beta bursts recorded at pairs of sites decreased significantly as the distance between the recording sites increased (
No evidence was found for waxing and waning beta oscillations, but rather, a time-dependent probability of beta burst occurrence that varied by brain region (
We tested whether beta burst rates and power during the post-trial period were modulated by specific aspects of the preceding trial. For this purpose, a fourth behavioral task, 1M3T, was added in which the monkeys were presented with the same visual cues as in the 3M3T task, but were required to perform only the first instructed movement in order to obtain reward. The rate and, separately, the power, of beta bursts were compared following trials of four behavioral tasks (performed in separate blocks of each experimental session), involving different numbers of movements (0M3T, 1M3T and 3M3T) or of visual cues indicating the potential spatial targets of movement (1M1T and 1M3T). In each brain region, the population average of normalized rates of beta bursts in the post-trial period were modulated significantly by the details of the preceding behavioral task performance (
Not only were the post-trial burst rates modulated by the preceding trial type, but the pattern of modulation differed across brain regions. In M1PMC, burst rates showed significant modulation between tasks with 0, 1 or 3 movements, regardless of the number of visual targets (
To determine whether the post-trial burst rates depended on the outcome of the trial, we tested whether beta burst rates following correct trials differed from those following error trials (
Given that, in all tasks, the rate and power of beta bursts in the dlPFC and striatum were highest during the post-trial period, we asked whether the during this period the bursts at pairs of striatal-prefrontal sites exhibited consistent temporal relationships, which they did. The population average coherence between all pairs of simultaneously recorded LFPs in the CN and dlPFC reached a peak in the beta frequency range during the post-trial period, following trials of the 1M1T and 3M3T tasks (FIGS. 18A,C). The coherence values at these post-trial peaks were significantly higher than the coherence in any other trial period. Given the prominent bursting at beta frequencies during the post-trial period, we asked whether the high coherence during this period might be due to elevated coherence specifically between beta bursts. Indeed, individual pairs of recording sites in the CN and dlPFC showed significantly higher post-trial coherence during periods in which both LFPs were bursting, than when neither was bursting (FIGS. 18B,D). The significantly non-zero phases of the coherence in either case rules out the possibility of volume conduction between the two sites or from a common third site. At the population level, the ratio of the magnitudes of coherence among pairs of LFPs that were bursting as opposed to not bursting was ˜3, and was significantly higher during the post-trial period in all tasks than in rest periods. This ratio differed significantly across tasks, and was inversely related to the number of movements performed in the preceding trial.
In addition to the temporal relationships between bursts in different brain regions, co-activation of bursts were studied across different sites within each region. These results are preliminary, based on data from a single experimental session. Our analysis took into account the entire population of simultaneously recorded LFPs within each region, as opposed to averaging results across all pairs of LFPs across regions. Beta bursts in the striatum were significantly more co-active than beta bursts in M1PMC and dlPFC. This fits well with the earlier results of the cross-covariance between bursts at pairs of increasingly distant sites (
LFP-LFP coherence between CN and dlPFC showed a significant peak during periods in the ITI. Coherence among coincident signals is a strong indicator of potential communication between sites. Therefore, each pair of simultaneously recorded LFPs were analyzed across the two brain regions (one in CN and the other in dlPFC), when either both electrodes of the pair were bursting or both were non-bursting. The majority of electrode pairs showed significantly greater coherence in the beta band when they were bursting than when they were non-bursting, suggesting that the bursts themselves were coherent with each other, as opposed to a general coherence between the LFP signals. This further supports the interpretation that beta bursts are related to post-task performance communication or concerted updating across widely distributed networks in the brain. This result, coupled with our earlier finding of increased bursting following correct, rewarded task performance, as opposed to erroneous, unrewarded performance, suggests that beta bursts represent concerted network activity following “successful” behavior.
Based on these results, the invention provides methods for tracking the changes in the beta frequency range of EEG, MEG, ECoG or LFP activity (or similar brain activity signals), recorded from any part of the human or animal brain, in order to detect the engagement and disengagement of individual brain regions during the performance of motor or non-motor tasks—that is, even tasks that are purely cognitive, without any overt motor output. In some embodiments, the method involves the following steps:
The detection of beta bursts in brain activity signals can be used to measure quantitatively the brain-site-specific depth of engagement, e.g., based on post-engagement burst rates and amplitudes, as well as the duration of engagement—the neural processing time each brain region devotes to the task—based on the timing of significant changes in burst parameters (with low and high burst rates corresponding to task engagement and disengagement, respectively). Both the depth and duration of brain-site-specific task engagement can be used as important physiological measures of performance and learning, complementing current behavioral measures such as reaction time and the percentage/frequency of errors. Clearly, this monitoring system can be used as a training tool, and can aid in focusing therapeutic interventions to the times that are relevant for such interventions, in conditions ranging from ADD, ADHD and autism spectrum disorders (ASDs) to conditions involving overt problems with movement or thought preparation and cessation, including OCD, PTSD and learning disabilities.
This detection system can be used to analyze individual differences in performance and learning. Individuals might break up complex problems into a series of smaller component tasks, which can be detected on the basis of beta power modulation as we have found in our non-human primate experiments. This detection would provide an independent measure of engagement and task-decomposition that can be useful in addition to reaction time measures, known to be inadequate markers in many circumstances. Detection of the beta burst episodes and their patterns can also be useful for detecting when a subject might be “zoning out” while performing a task of extended duration, and thus can have high relevance for contexts ranging from air traffic control to vehicle operators to student groups.
Beta bursts, though spatially localized, can be significantly coherent across particular brain regions, a finding highly suggestive of inter-area communication. Thus, the detection system can be used to detect abnormal communication during behavior, e.g., in autism or schizophrenia.
We have shown significant differences across brain regions in the timing of peak beta activity relative to behavioral events. The timing of the peak in each brain region was consistent with a role for beta oscillations in post-performance processing of behavior. Investigating the source of the peaks in the trial-averaged beta power, we discovered that beta activity was fundamentally characterized by brief, spatially localized bursts of oscillations, whose rate and power were modulated by the behavioral tasks in a manner similar to the modulation of trial-averaged beta power. The modulation of post-trial bursts by task features (numbers of movements or visual cues) differed across brain regions, as follows: M1PMC bursts tracked numbers of movements, striatal bursts tracked numbers of cues and prefrontal bursts tracked a combination of both. Post-trial beta burst rates in striatum and dlPFC were also modulated by the outcome of the preceding behavioral task performance—burst rates were significantly higher following rewarded correct trials, as opposed to unrewarded errors. During the post-trial period, beta band coherence between pairs of brain regions reached a maximum. Paired sites in CN and dlPFC exhibited increased coherence specifically during beta bursts, consistent with the view that the bursts may facilitate communication or coordination across brain regions. This idea is supported by computational work showing that beta oscillations are particularly well-suited to long-range interactions, as well as by recent physiological work on task-dependent changes in beta coherence across cortical regions.
Taken together, our results suggest a unifying interpretation of beta oscillations in the brain. Our findings argue against the view that beta activity prevents brain regions from changing their present activity, and hence, the current behavioral output. Rather, we propose that beta oscillations represent mechanisms for integrating the successful outcome of behavior with the details of the performance that led to that outcome, across cortico-striatal networks. The purpose of such mechanisms could be to increase the likelihood of achieving goals in future trials, in either of two ways: either by preventing changes to network activity that led to a rewarded outcome, or by actively reinforcing such network activity.
Such a role for beta oscillations is consistent with current ideas about their mechanistic origins. In frontal cortex and basal ganglia beta oscillations are thought to arise from the interplay of excitatory and inhibitory feedback. In the basal ganglia, computational work has identified the GPe-STN network as a likely candidate generator of beta oscillations, which can arise from the inhibition of STN by GPe, coupled with the excitation of GPe by the STN. In vitro slice work suggests that beta oscillations in sensorimotor cortex may reflect gap-junction-dependent firing of pyramidal cell layer neurons. Furthermore, recent computational work has demonstrated that beta oscillations in association cortex can sustain representations of stimuli in short-term memory for the purpose of integrating changing stimuli across time. It is possible that beta activity in other brains regions could similarly function to integrate behavioral outcomes with the details of the behavior that led to them, and then tune or update network activity accordingly.
Low beta activity following an unrewarded outcome may allow synaptic plasticity to occur, mediated by other mechanisms. The purpose of such plasticity could be to change the network dynamics in order to increase the likelihood that the desired outcome will be obtained in the future. This would be particularly useful in learning situations (exploration), and warrants further study of the potential evolution of beta activity during learning. On the other hand, in the performance of well-learned behaviors, low beta activity following occasional errors, as we have observed in our experiments, might prevent the inappropriate reinforcement of the network activity that led to the undesired outcome. In such a scenario, low beta would effectively prevent the occasional erroneous performance from modifying the monkey's typically successful behavior. Beta activity would thus maintain the learned network settings by making them less susceptible to change by infrequent errors (exploitation). Thus, rather than promoting the current behavioral output, as suggested by the status quo-preserving theory, beta oscillations could be involved in promoting the current network settings that generated the behavior that culminated in a successful outcome. This would serves to increase the likelihood that the same outcome would be obtained in similar situations in the future.
Patients with PD, a disease involving a loss of the major source of dopaminergic input to the striatum, exhibit pathologically high coherence in beta band activity in the cortex and basal ganglia. While the clinical consequences of elevated beta synchrony are not clear, evidence has been mounting that it can be reduced by the leading treatments for PD. Remarkably, recent studies have shown that decreases in beta activity in the subthalamic nucleus can correlate with decreases in motor symptoms of PD, specifically bradykinesia (slowness of movement) and rigidity. Other work recently established a link between high cortical beta power and bradykinesia, and some have proposed that the benefits of deep-brain stimulation therapy are mediated by reducing high beta synchrony. Patients with PD typically present with cognitive as well as motor symptoms. Our results indicate a possible causal relation between the abnormally high beta activity in the cortex and basal ganglia of PD patients and a specific type of learning deficit associated with PD. A recent study found that PD patients were more likely than normal subjects to perseverate in their choices, independently of reward history, and that this perseveration in choice decreased with dopaminergic therapy. Notably, beta activity in PD patients has also been shown to decrease with dopaminergic treatment. The indiscriminately high beta activity in PD patients could drive their choices by reinforcing the behavior that led to the previous choice, regardless of whether it was rewarded. Alternatively, the high beta activity might prevent the plasticity necessary for the brain to learn from an erroneous choice, modify the network settings accordingly, and increase the likelihood of choosing correctly in future trials. We predict that the direct manipulation of beta activity in PD patients would reduce both akinesia and the perseveration of choice errors.
Particularly interesting are the unique features of beta activity in the dlPFC. Despite the fact that across all tasks the dlPFC had the lowest rate of bursts among the four studied brain regions, during the post-trial period it exhibited the greatest change in burst rates relative to rest (
A measure of burst concurrence was developed to quantify the degree to which beta bursts tend to occur simultaneously on multiple electrodes, which is computed as follows. Individual shuffled pseudo-trials were constructed by choosing trials at random without replacement across the set of N electrodes being analyzed, so that no two channels contained data from the same trial. A large number of such shuffled pseudo-trials was constructed and used as a proxy data set with the same statistics as the original data set, but with any correlations across channels destroyed. Then, a histogram of the frequency was computed with which each possible number of bursts from 0 to N occurred simultaneously at each time point in the proxy data set, and the thresholds were found for the high and low tails of the histogram that included no more than 2.5% of the pseudo-trials in each tail. Since the thresholds were constrained to be integers, we computed the actual fraction of pseudo-trials contained in the bins beyond the thresholds. The same type of histogram was constructed for the original data, and “concurrence” was defined as the ratio of the fraction of beyond-threshold trials in the original data set to that in the randomized proxy data set. A concurrence value of 1 thus corresponds to chance (i.e. uncorrelated) simultaneous bursting and/or simultaneous non-bursting, and values greater than 1 indicate more simultaneous bursting and/or simultaneous non-bursting than is expected by chance. To generate confidence limits on the value of concurrence, a bootstrap analysis was performed across trials.
Within each of the four brain regions examined (dlPFC, M1, CN and Put), the concurrence of bursts was statistically significantly greater than chance during ITI, but during at least the post-cue period and in some cases throughout the entire task execution period the concurrence of bursts was not statistically significantly greater than chance. In sum, the timing of beta bursts is substantially random during task performance, and become significantly more coherent (e.g., coordinated) across brain regions after task performance. Relative measurement of the timing of bursts therefore can be used to monitor task engagement in real time. In addition, burst rate can be used to monitor task engagement in real time.
Within each brain structure, each time point was marked where there was a greater-than-threshold number of simultaneous bursts in progress as belonging to a concurrent burst for that structure, and repeated the concurrence analysis on concurrent bursts across structures. As in the within-structure analysis, concurrence of concurrent bursts across structures is significantly greater than chance during ITI but not during post-Cue period. In addition, cross-structure concurrence was significantly greater during the entire ITI than at any time during the entire post-Cue and Movement period, and cross-structure concurrence showed a significant peak during the final hold period followed by a marginally significant valley during and shortly after the reward delivery period.
The cross-covariance of the envelopes of the composite HHT beta-band signals between CN and dlPFC also was analyzed using the same time periods as coherence. The majority of electrode pairs showed greater envelope cross-covariance when both electrodes were bursting than when both were non-bursting. This last result shows that the detailed fine temporal structure of beta bursts is similar across sites, further indicating the possibility that the sites are involved in concerted network-level interactions.
Referring to
The present invention may be embodied in many different forms, including, but in no way limited to, computer program logic for use with a processor (e.g., a microprocessor, microcontroller, digital signal processor, or general purpose computer), programmable logic for use with a programmable logic device, (e.g., a Field Programmable Gate Array (FPGA) or other PLD), discrete components, integrated circuitry (e.g., an Application Specific Integrated Circuit (ASIC)), or any other means including any combination thereof. In one embodiment of the present invention, some or all of the processing of the data collected is implemented as a set of computer program instructions that is converted into a computer executable form, stored as such in a computer readable medium, and executed by a microprocessor under the control of an operating system.
In some embodiments, computer software (e.g., a computer program) is provided for capturing, visualizing, and/or analyzing beta bursts.
The computer program may be fixed in any form (e.g., source code form, computer executable form, or an intermediate form) either permanently or transitorily in a tangible storage medium, such as a semiconductor memory device (e.g., a RAM, ROM, PROM, EEPROM, or Flash-Programmable RAM), a magnetic memory device (e.g., a diskette or fixed disk), an optical memory device (e.g., a CD-ROM), a PC card (e.g., PCMCIA card), or other memory device. The computer program may be fixed in any form in a signal that is transmittable to a computer using any of various communication technologies, including, but in no way limited to, analog technologies, digital technologies, optical technologies, wireless technologies networking technologies, and internetworking technologies. The computer program may be distributed in any form as a removable storage medium with accompanying printed or electronic documentation (e.g., shrink-wrapped software), preloaded with a computer system (e.g., on system ROM or fixed disk), or distributed over a network.
Programmable logic may be fixed either permanently or transitorily in a tangible storage medium, such as a semiconductor memory device (e.g., a RAM, ROM, PROM, EEPROM, or Flash-Programmable RAM), a magnetic memory device (e.g., a diskette or fixed disk), an optical memory device (e.g., a CD-ROM), or other memory device. The programmable logic may be fixed in a signal that is transmittable to a computer using any of various communication technologies, including, but in no way limited to, analog technologies, digital technologies, optical technologies, wireless technologies (e.g., Bluetooth), networking technologies, and internetworking technologies. The programmable logic may be distributed as a removable storage medium with accompanying printed or electronic documentation (e.g., shrink-wrapped software), preloaded with a computer system (e.g., on system ROM or fixed disk), or distributed from a server or electronic bulletin board over the communication system (e.g., the Internet or World Wide Web).
Computers and computer systems described herein may include operatively associated computer-readable media such as memory for storing software applications used in obtaining, processing, storing and/or communicating data. It can be appreciated that such memory can be internal, external, remote or local with respect to its operatively associated computer or computer system.
The aspects, embodiments, features, and examples of the invention are to be considered illustrative in all respects and are not intended to limit the invention, the scope of which is defined only by the claims. Other embodiments, modifications, and usages will be apparent to those skilled in the art without departing from the spirit and scope of the claimed invention.
The use of headings and sections in the application is not meant to limit the invention; each section can apply to any aspect, embodiment, or feature of the invention.
Throughout the application, where processes are described as having, including or comprising specific process steps, it is contemplated that processes also consist essentially of, or consist of, the recited process steps.
In the application, where an element or component is said to be included in and/or selected from a list of recited elements or components, it should be understood that the element or component can be any one of the recited elements or components and can be selected from a group consisting of two or more of the recited elements or components.
The use of the terms “include,” “includes,” “including,” “have,” “has,” or “having” should be generally understood as open-ended and non-limiting unless specifically stated otherwise.
The use of the singular herein includes the plural (and vice versa) unless specifically stated otherwise. Moreover, the singular forms “a,” “an,” and “the” include plural forms unless the context clearly dictates otherwise.
It should be understood that the order of steps or order for performing certain actions is immaterial so long as the invention remains operable. Moreover, two or more steps or actions may be conducted simultaneously.
This application claims priority to and the benefit of U.S. Provisional Application No. 61/491,656, filed on May 31, 2011, and U.S. Provisional Application No. 61/541,047, filed on Jun. 24, 2011, the entire disclosures of each of which are incorporated by reference herein.
This invention was supported by grants N00014-07-1-0903 awarded by Office of Naval Research (ONR), NBCHC070105 awarded by Defense Advanced Research Projects Agency (DARPA), and R01NS025529 awarded by National Institute of Neurological Disorders and Stroke (NINDS). The government has certain rights in the invention.
| Number | Date | Country | |
|---|---|---|---|
| 61491656 | May 2011 | US | |
| 61501047 | Jun 2011 | US |
