A comparison of stimulus synchronous activity in the primary motor cortices of athletes and non-athletes

Abstract

In this study, we measured primary motor cortex (MI) activity during a reaction time task to examine the appearance of MI activity that synchronized with the stimulus presentation (stimulus synchronous MI activity, SSMA). Because brain activity was expected to be enhanced by the repetitive/extensive activation, we hypothesized that the SSMA would be more clearly observable in athletes who were trained to perform reactive movements than in non-athletes. MI activity was measured in ten athletes and ten non-athletes by magnetoencephalography. The tasks were a simple reaction task and a Go/Nogo reaction task in which the subjects were asked to abduct their right index fingers in response to a visual stimulus. The Go/Nogo reaction time task was adopted to confirm the presence of the SSMA, because the MI activity in response to a Nogo stimulus did not overlap with the MI activity that was synchronous with the execution of the movement. The results show that the SSMA was clearly apparent in the athlete group (9/10). In the non-athlete group, however, only three subjects showed the SSMA (3/10). Moreover, the MI activity of the athletes tended to be larger than that of the non-athletes, even though the athletes did not specifically practice these index finger movements during their daily training. We concluded that long-term physical training promotes MI activity and the effects of reactive task repetition were more clearly apparent in the MI activity of the athletes.

Appendix

The SSP method explains the measured magnetic fields by the linear combination of specific magnetic field patterns (the time-independent spatial distribution). The time course of activity amplitude of N sources was indicated by the time-varying amplitude ${\mathbf{a}}(t) = \{ a_{1} (t),a_{2} (t), \ldots ,a_{N} (t)\} ^{{\text{T}}}$ corresponding to spatial pattern S = (s ₁, s ₂,...,s _N ), where the vectors s _i are unit vectors that characterize time-independent spatial distributions:

$$ {\mathbf{b}}(t) = {\sum\limits_{i = 1}^N {a_{i} (t){\mathbf{s}}_{i} } } + {\mathbf{n}}(t), $$

(1)

where b(t) is the magnetic field and n(t) is the noise. The estimate for the amplitude a(t) is

$$ {\mathbf{a}}(t) = {\mathbf{S}}^{ + } {\mathbf{b}}(t), $$

(2)

where S ⁺ is the pseudoinverse of S.

The procedure to estimate the time course of the activity is described below:

• 1. The time-independent spatial distribution of the motor activity in MI was calculated from the SVM task. First, the dipole position of the motor activity was estimated in the SVM task using multi-dipole estimation software that was attached to the MEG system. Then, the spatial distribution of the magnetic field was calculated at this position. One motor dipole was assumed to be in each of the two hemispheres, and two orthogonal components were defined for each dipole so as not to fix the orientation of the motor dipole. Finally, four spatial patterns were defined for the motor activity: ${\mathbf{S}}_{m} = ({\mathbf{s}}_{{m1}} ,{\mathbf{s}}_{{m2}} ,{\mathbf{s}}_{{m3}} ,{\mathbf{s}}_{{m4}} )$.

• 2. The visual activity was used not to precisely explain the brain activity in the visual cortex, but instead to achieve the explanation of the visually evoked magnetic fields. The spatial distribution of the visual activity was obtained from the VEF experiment. Since it was difficult to estimate a reliable visual dipole position from all of the subjects, the magnetic field pattern at one or two peak latencies up until 200 ms after the stimulus onset was used as the spatial pattern: S _v  = {s _v1, (s _v2)}. The value of the peak latency was determined from the measured waveforms of each subject.

• 3. Then, two pairs of two motor activities and one or two visual activities were defined by S = (S _m , S _v ), and the time-varying amplitude vectors a(t) were calculated simultaneously using Eq. 2. The goodness of fit (GOF) and the root mean square (RMS) of the residual error were determined: cumulative GOF between 100 and 200 ms from the stimulus onset should be over 70%, or the error RMS between 100 and 200 ms should be less than twice the RMS of the measured data before the stimulus onset. Because the GOF is worse if the activity is weak, the RMS residual error was also determined. The equations for the GOF and the RMS are:

  $$ {\text{cumulative}}\,{\text{GOF}} = {\left( {1 - \frac{{{\sum\nolimits_{t = ts}^{te} {{\sum\nolimits_{i = 1}^{64} {(m_{i} (t) - e_{i} (t))^{2} } }} }}} {{{\sum\nolimits_{t = ts}^{te} {{\sum\nolimits_{i = 1}^{64} {m_{i} (t)^{2} } }} }}}} \right)} \times 100\% , $$

  (3)
  $$ {\text{residual}}\,{\text{error}}\,{\text{RMS}}(t) = {\sqrt {\frac{{{\sum\nolimits_{i = 1}^{64} {(m_{i} (t) - e_{i} (t))^{2} } }}} {{64}}} }, $$

  (4)

  where m _i (t) defines the magnetic field measured by the ith MEG sensor with a latency t in the time window between ts (100 ms) and te (200 ms), while e _i (t) denotes the estimated magnetic field, which is calculated as follows:

  $$ {\mathbf{e}}(t) = {\mathbf{Sa}}(t). $$

  (5)

• 4. Because the estimated amplitude a(t) is that for the unit vector of the spatial distribution, the dipole moment of the motor activity at its location was recalculated from the amplitude of the two orthogonal components. Hence, the dipole moment of the motor activity was always a positive value.