Supplementary Motor Area exerts Proactive and Reactive Control of Arm Movements

Transcription

1 Supplementary Material Supplementary Motor Area exerts Proactive and Reactive Control of Arm Movements Xiaomo Chen, Katherine Wilson Scangos 2 and Veit Stuphorn,2 Department of Psychological and Brain Sciences, Johns Hopkins University, Baltimore, MD, USA 2 Department of Neuroscience, Johns Hopkins University School of Medicine and Zanvyl Krieger Mind/Brain Institute, Baltimore, MD, USA Functional role of changes in LFP power in the beta-band The physiological processes underlying LFP activity, in general and in specific frequency bands, is a complicated issue that is not well-understood at present. A recent review advanced the hypothesis that beta-band activity throughout the brain might be related to the maintenance of the current sensorimotor or cognitive state (Engel and Fries, 2). This is an interesting hypothesis, and a number of studies provide evidence for this idea. This hypothesis is in agreement with the finding in our study that increases in beta-band power are correlated with longer RTs (Figure 7). However, this idea is not in agreement with the fact that activity in the beta range also increases when the monkeys exert proactive control following an error (Figure ), because in this case there was clearly a change of the sensorimotor and cognitive state. The increase of beta-range activity during successful cancelation is similar to the findings in the Swann and colleagues paper (Swann et al., 29). This finding has been cited by Engel and Fries as support for their hypothesis, but other interpretations are possible. The sudden appearance of the stop signal requires the activation of a new neuronal process that is necessary to cancel the movement that is in preparation. This represents not so much task maintenance as a phasic shift in task requirements. Furthermore, it is not clear what the underlying mechanism might be that would link this computational process (state maintenance) with the neurophysiological phenomenon (the power in the beta range). So in conclusion, we are not sure, to what degree the maintenance hypothesis explains our findings.

2 Our results are generally consistent with the idea that low-frequency LFPs reflect synchrony in the synaptic inputs, whereas the higher frequencies are more sensitive to spiking activity in the neuronal network surrounding the electrode tip (see Figure 4A). Thus, the activity in the beta range might represent a very general physiological signal, such as the synchronous synaptic inputs and resulting dendritic currents (Niessing et al., 25; Viswanathan and Freeman, 27; Ray et al., 28). The beta-range activity might reflect input from cortical and subcortical regions (Belitski et al., 28). In this case, the LFP activity in the beta-range in itself would not represent a specific computational process, such as task maintenance. Instead, the function of the beta-range activity would be dependent on the type of signals that are conveyed by the synaptic inputs into a specific cortical area and on the effect that they have on neuronal activity in the area. In the case of the SMA, the increased power in the beta range that comes before arm movements with longer RTs might reflect synaptic input from higher-order areas, such as dorsolateral prefrontal cortex, anterior cingulate cortex, or (in humans) inferior frontal gyrus. This input might initiate the change in threshold settings implemented by the SMA. This would explain why most of the effects in the multi-unit activity, which reflects more the output of the SMA, become apparent only after the changes in the beta-range LFP power. Comparison to other suggested classification schemes for behavioral control Both monkeys show history effects in their behavior and in the measures of neuronal activity in the SMA. In particular, monkey B showed stronger changes in reaction time following errors, both behaviorally (Figure 3), and in his LFP activity (Figure ). In contrast, monkey E showed larger adjustments following canceled trials (Figure 2). Thus, the adjustments in the countermanding task do not rely on errors per se. What instead seems important is the presence of a sensory cue that increases the motivation to wait and thereby adjusts the balance of motivations (to go or to wait) so that there is a different level of excitability. 2

3 However, the adjustment in motor readiness seems to come in response to some external event. One can therefore ask, if it is best to classify this type of behavioral control in the countermanding task as proactive. Hikosaka and Isoda have recently suggested a different classification of the mechanisms underlying behavioral switching (Hikosaka and Isoda, 2). They pointed out that a switch in behavior can occur either retroactively based on error feedback, or proactively by detecting a contextual cue. In particular the strong adjustment following errors, shown by monkey B might be classified as retroactive and not as proactive control according to the Hikosaka/Isoda classification scheme. However, this classification is along different lines than our classification into proactive and reactive control, which is related to the one proposed by Braver and colleagues (Braver, 27). Therefore, proactive is used in a slightly different meaning. Hikosaka and Isoda refer to phasic neuronal events as proactive control, whereas we hypothesize that tonic neuronal control signals that are active over long time periods underlie proactive control in our task. These control signals are thought to exert excitatory and inhibitory modulations of the motor system. The balance of inhibitory and excitatory signals sets the response threshold. Stop signal trials can lead to a shift in this balance and thus in the proactive control settings. This shift could be classified as retroactive in the Hikosaka/Isoda sense, since it follows an encounter with a stop signal. However, the shift could also be labeled proactive according to the same classification scheme, since it occurs before the next trial starts (i.e. before target onset). Furthermore, the shift does not always follow an error and occurs after the only sensory cue that could trigger such an adjustment in the countermanding task. Cited References: Belitski A, Gretton A, Magri C, Murayama Y, Montemurro MA, Logothetis NK, Panzeri S (28) Lowfrequency local field potentials and spikes in primary visual cortex convey independent visual information. J Neurosci 28: Engel AK, Fries P (2) Beta-band oscillations - signalling the status quo? Curr Opin Neurobiol 2: Hikosaka O, Isoda M Switching from automatic to controlled behavior: cortico-basal ganglia mechanisms. Trends Cogn Sci 4:

4 Niessing J, Ebisch B, Schmidt KE, Niessing M, Singer W, Galuske RA (25) Hemodynamic signals correlate tightly with synchronized gamma oscillations. Science 39: Ray S, Crone NE, Niebur E, Franaszczuk PJ, Hsiao SS (28) Neural correlates of high-gamma oscillations (6-2 Hz) in macaque local field potentials and their potential implications in electrocorticography. J Neurosci 28: Scangos KW, Stuphorn V (2) Medial Frontal Cortex Motivates but does not Control Movement Initiation in the Countermanding task. J Neurosci 3: Swann N, Tandon N, Canolty R, Ellmore TM, McEvoy LK, Dreyer S, DiSano M, Aron AR (29) Intracranial EEG reveals a time- and frequency-specific role for the right inferior frontal gyrus and primary motor cortex in stopping initiated responses. J Neurosci 29: Viswanathan A, Freeman RD (27) Neurometabolic coupling in cerebral cortex reflects synaptic more than spiking activity. Nat Neurosci :

5 A Left.6.8 Right.5 HighGamma Gamma 25 4 Hz B C Time from Target Onset (ms) Time (ms) Supplementary figure : Cross-correlation between LFP and multi-unit activity align on target presentation. The two columns on the left show the mean normalized firing rates (black line) and mean normalized LFP power in high gamma band (red line), gamma band (purple line) and 25-4 Hz band (blue line) for arm movement to the left (left) and to the right (right). The three columns on the right show the cross-correlation between mean normalized firing rates and mean normalized LFP power in each band for both movement to the left (right line) and movement to the right (black line). The time periods at which the

6 cross-correlation is significant (Spearman rank test, p <.5, Bonferroni-corrected for multiple comparisons) are denoted by thick horizontal lines at the bottom of each plot. A: Cross-correlation analysis across all recordings in SMA of monkey B. B: Crosscorrelation analysis across all recordings in SMA of monkey E.

7 A Left.6.8 Right.4 HighGamma Gamma 25 4 Hz B C n Spearman Rank Correlatio Time from Movement Onset (ms).2.2 Time (ms) Supplementary figure 2: Cross-correlation between LFP and multi-unit activity align on movement initiation. Conventions are as Supplementary Figure.

8 A Monkey B B Monkey E Number of recordings Number of recordings Number of recordings Difference Index Difference Index.5.5 Supplementary figure 3: Distribution of trial history effects on multi-unit recording. Distribution of sequential effect index that describe the relative difference of multi-unit activity for Ca-Go and Go-Go trials (top) and for E-Go and Go-Go trials (bottom) during baseline (left), target onset (middle) and movement (right) period combining both arm movement to the left and to the right.. Sequential effect index with significant change are shown in blue. A: Distribution of sequential effect index for all recordings in SMA of Monkey B. B: Distribution of sequential effect index for all recordings in SMA of Monkey E.