Performance for single-step saccades in the control experiments. Lower horizontal dispersion of saccade endpoints for IPSp and S1, plotted in the same format as in Fig. Ipsilateral saccade data not shown. Performance during IPSp stimulation did not differ significantly from performance during stimulation of S1. Data are presented in the same format as in a. Saccades during stimulation of IPSp and S1 were not significantly different.
For both sites, the horizontal error in this single-step task was greater than that observed in the double-step task of Experiment 1 compare Figs. By contrast, the required saccade to T2 in Experiment 1 was approximately vertical and was equally often directed leftward and rightward with respect to T1.
A saccade to T1 in the double-step task introduces the need to update the memory trace of T2, but it also requires the second-saccade to originate from a contralateral eye position.
Eye-in-orbit position modulates the activity of neurons in area LIP and many other attentional and oculomotor areas of the monkey 35 — If IPSp stimulation disturbed neurons with similar properties, this could have effects on saccade planning or memory that depend on eye position and might thus explain the effects on second-saccades observed in the double-step saccade task. This possibility was tested in a further control experiment Experiment 3.
The task required a single memory-guided saccade to T2, but, rather than fixating centrally at the start of a trial, as in the previous two experiments, T1 was used as the initial fixation point. The saccade required in this task was therefore identical to the second-saccade in Experiment 1, but because no saccade intervened between the appearance of T2 and the saccade toward it, there was no need to update the memory trace of T2.
If the bias and dispersion effects observed in the double-step task were caused by effects of IPS stimulation on planning saccades from contralateral eye positions, then such effects should also be evident in this control task. Thus, the bias and precision effects observed in Experiment 1 cannot be explained by effects of IPSp stimulation on executing memory-guided saccades from a contralateral eye position.
Discussion This study investigated the neural bases of spatial updating across saccades by using noninvasive cortical stimulation TMS and a double-step saccade task. Stimulation over the posterior termination of the right IPS IPSp during contralateral leftward saccades impaired spatial updating of a remembered target location.
Specifically, participants overcompensated for the leftward horizontal displacement of the eye caused by the first-saccade, as shown by a rightward shift in the endpoints of the second-saccades. Stimulation of IPSp also increased the dispersion of second-saccade endpoints, but only when TMS was applied immediately after the offset of the first-saccade. Moreover, we showed that the effects on second-saccades were not merely a consequence of changes in the execution of first-saccades; nor were they attributable to a disruption of the memory trace of T2 in retinal coordinates; nor were saccades from T1 to the remembered location of T2 affected when there was no preceding saccade.
Taken together, these experiments indicate that stimulation of IPSp introduces bias and random noise into the coordinate transformation that underlies spatial updating across saccades. The data from this former study were interpreted as evidence for a time-locked involvement of the PPC in spatial updating; however, this conclusion is subject to two important criticisms.
First, the authors applied TMS to a single parietal site only with a relatively nonfocal circular coil; hence, it cannot be determined whether the findings arose directly from PPC stimulation or were merely a nonspecific consequence of right hemisphere disruption. More critically, however, this former study did not assess the influence of TMS on second-saccades in the absence of any requirement for spatial updating.
It is therefore unclear whether stimulation of the PPC influenced spatial updating, or simply affected the planning or execution of second-saccades in the double-step task.
In contrast to the transient effects of IPSp stimulation on saccadic dispersion, the rightward bias in second-saccade endpoints occurred independently of the timing and presence of stimulation on any given trial. This observation implies a modulation of neural processing that outlasts the direct period of cortical stimulation and endures sufficiently to propagate across trials i.
Animal studies suggest that such changes in excitability may be traced to long lasting modulatory effects of TMS on synaptic transmission, gene expression, and neurotransmitter functioning 33 , The finding that second-saccades overcompensated for leftward first-saccades suggests that an exaggerated representation of contralateral eye position or displacement is used in the transsaccadic coordinate transformation.
In humans and monkeys, it remains controversial as to which of these potential sources of extraretinal information is used to update the internal representation of space.
Indeed, the primary characteristic that divides current models of spatial updating is whether the transformation is considered to be driven by an eye-in-orbit position signal or a position-free eye displacement signal 7 , 11 , 12 , 38 , 40 — Our results cannot distinguish between these two possibilities because the direction of eye displacement and the final eye-in-orbit position covaried in the saccade sequences used here.
Future studies could identify the crucial extraretinal signal by exploring which of these variables determines the behavioral effects of TMS as reported here. This could be achieved by adding a manipulation of eye position to the double-step task. A similar approach has been used to study spatial updating in monkeys during inactivation of the PPC 30 and in patients with damage to the parietal cortex More broadly, our findings have implications for understanding disorders of spatial representation associated with parietal damage, such as unilateral neglect 44 and optic ataxia Several investigators have suggested that deficient transsaccadic spatial updating mechanisms contribute to the symptoms observed in these patients 43 , 46 , Because unilateral neglect is particularly common after damage to the parietal lobe of the right hemisphere 48 , our findings provide a plausible neuroanatomical substrate for these deficits.
Whether similar hemispheric asymmetries exist for the stimulation effects reported here remains to be determined. In sum, we have demonstrated a crucial role for the posterior intraparietal area in updating representations of space across saccades, a characteristic that, in our view, asserts this cortical area as the most probable of several recently identified potential homologs of monkey LIP 19 — 21 , 26 , Sixteen right-handed volunteers participated in each of the three experiments.
Thirteen of the 16 participants from Experiment 1 also completed both control experiments. The mean age was All participants had normal or corrected-to-normal vision and gave informed written consent before participation.
All fixation and target stimuli were red spots diameter, 0. All saccade tasks were therefore performed without any visual references other than fixation and target stimuli. The double-step task used in Experiment 1 is shown in Fig. The same sequence of stimuli was used for Experiment 2 except that the central fixation was extinguished at the same time as T1, which cued the participants to execute a saccade to the remembered location of T2.
In Experiment 3, each trial began with a fixation stimulus located in one of the four positions used for T1 in Experiment 1. The central stimulus used as the initial fixation point in the first two experiments remained in Experiment 3, as shown in Fig.
The eye tracker was calibrated before each block of 24 trials if necessary. A chin rest was used to prevent head movement. Stimulation sites were identified on an individual basis before participation in the behavioral sessions by using T1-weighted MR brain scans.
Sites of stimulation were defined by their position with respect to sulcal landmarks Fig. Standardized coordinates were obtained for each stimulation site by spatially normalizing each participant's anatomical image to the Montreal Neurological Institute template by using SPM2 software Wellcome Department of Imaging Neuroscience, London, U. Note that normalization was performed only after completion of the experiment for the purpose of relating the stimulated sites to other TMS and fMRI studies; it was not used to identify cortical loci or position the TMS coil.
TMS was delivered by using a Magstim Rapid system 2. The intensity of stimulation was set to the maximum comfortable level, expressed as a percentage of motor threshold, and adjusted for each site to control for differences in the distance between the scalp and cortex Consecutive testing sessions were separated by at least 24 h.
For the double-step saccade task in Experiment 1, TMS was timed according to the predicted onset or offset of the first-saccade, rather than at fixed times relative to a display event. Predictions for leftward and rightward first-saccades were generated independently by using an exponentially weighted average of previous saccade latencies.
This procedure ensured that the distribution of TMS onset times would be comparable within and between different testing sessions. For the control tasks in Experiments 2 and 3, in which there was a single saccade to T2 only, TMS was timed on each trial according to a randomly selected sample from the participant's first-saccade latency distribution from Experiment 1. Data from a representative participant were used for the three participants that did not participate in Experiment 1. Eye position data were filtered offline by using a nonlinear exponential smoothing algorithm The output of the algorithm for each trial was inspected visually for accuracy.
Trials that contained blinks, incorrect behavioral responses e. Early and late TMS conditions were obtained by parsing the bimodal distribution of TMS onset times into two separate clusters. The tails of each cluster were truncated to remove sections of the time courses where parameter estimates were unreliable. Only the central ms period of each cluster was analyzed. One participant was removed from the analyses for Experiment 2 because his mean endpoint error was almost 3 SD beyond the group mean.
Simple main effects were examined by using pairwise comparisons with Bonferroni correction where appropriate. The dispersion of saccade endpoints was analyzed as a function of TMS onset time and stimulation site.
Additional details of methodology and data analysis are provided in SI Text. The authors declare no conflict of interest. This article contains supporting information online at www.