Another dissociation between the PD and cerebellar patients was observed in the asynchrony results. As is typically observed for synchronized tapping at intervals below 1 sec (Aschersleben and Prinz, 1995; Fraisse and Voillaume, 1971), the taps for all of the participants occurred prior to the tones. However, this asynchrony was greater for the PD patients (overall mean of 136 msec) than for both the controls and cerebellar patients (62 and 70 msec, respectively).

As outlined in the introduction, three different factors could contribute to the increase in the asynchrony (see Equation (19.3)). First, the increased asynchrony may be due to the fact that PD patients perceive their taps to occur later than do healthy individuals. The difference in perceptual delays for the perception of the tone and the tap influences the average asynchrony directly. For example, it has been shown that when participants tap with their foot, they precede the pacing tones by 50 msec more than when they tap with their finger (Aschersleben and Prinz, 1995). This effect may also be related to how PD patients integrate different sources of information to estimate the time at which the tap has occurred. A number of researchers have proposed a role for the basal ganglia in sensory integration, even though PD patients do not show obvious sensory impairments. To fully account for the observed difference between groups, one would have to posit that the PD patients perceived their taps to have occurred 66 msec later than the age-matched controls.

Second, negative asynchronies could result from an internal clock that is operating at a faster rate than the external pacing signal. A phase correction process would then prevent the error from accumulating across successive taps, but the faster rate of the internal clock would result in taps occurring prior to the tones. In contrast to the perceptual delay hypothesis, this explanation offers a parsimonious account of why the asynchronies were larger in the 900-msec condition for all of the groups. The mean error of the clock is likely to be proportional to the length of the timed interval, causing larger asynchronies for longer intervals.

Pharmacological studies in humans (Rammsayer, 1993) and rats (Meck, 1996) have suggested that the rate of an internal pacemaker may be altered by dopamine levels. However, in this work, the clock has been hypothesized to slow down when dopamine levels are low, not to speed up. Of course, we did not monitor dopamine levels, and we did not attempt a within-subject comparison in which the patients were tested both on and off their medication. It would be interesting to see if the mean asynchrony lead varied with medication level.

While the relationship of the asynchrony to dopamine is unclear, this behavioral change is reminiscent of the speeding up that is observed in PD patients when engaged in an extended action. For example, PD patients tend to speed up during unpaced, repetitive tapping (Ivry and Keele, 1989; O'Boyle et al., 1996). Similarly, although they have difficulty initiating locomotion, once started, their steps become smaller and marked by a faster cycle time. Such changes could be interpreted as reflecting a bias for an internal clock to operate faster when engaged repetitively.

Ivry and Richardson (2002) offer an alternative model that could account for the reduced cycle time. In their view, the basal ganglia operate as a threshold device, gating when centrally generated responses are initiated. They conceptualize the loss of dopamine as an increase in the threshold required to initiate a response. If we assume that this threshold drifts toward more normal levels with repetitive use, then the same input pattern will trigger a response at shorter latencies over successive cycles. Thus, the increased negative lag could reflect a change in a thresholding process, rather than a disturbance of sensory integration times, or a change in the operation of an internal timing process.

Third, as shown in Equation (19.3), the observed asynchronies caused by a difference between clock speed and the pacing rate will be modulated by a. Lower gains of the error correction process would allow the error to accumulate to a larger degree, resulting in larger asynchronies. Assuming that the internal timing mechanism runs too fast in all groups, the differences in a alone potentially could account for a substantial part of the observed differences between groups. In the current study, we can estimate the differences between internal clock speed and pacing signal for each group, given the values of the error correction parameter. It turns out that the differences in error correction can only partly explain the group differences in asynchronies. To fully account for this effect, we would have to posit an additional difference in clock speed, on average 20 msec faster for the PD patients.

At present, the results do not allow us to discriminate between accounts based on changes in perceptual delays or inaccuracies in the internal timing signal (caused by different clock speed or changes in threshold). Moreover, the relative contribution of reduced error correction to the increased asynchrony depends on the assumption that error correction behaves linearly. When the asynchronies deviate substantially from zero, as is the case for the PD patients, this assumption is likely to be violated (Pressing, 1999). Thus, converging evidence from independent methods is needed to distinguish between these factors.

0 0

Post a comment