Article Figures & Data
Figures
- Figure 1.
A , Memory-guided anti-reach task. The monkey had to keep ocular fixation on a small central red dot throughout the trial. The color of the central hand stimulus (green/blue) instructed randomly interleaved pro- or anti-reaches with the movement goal (dotted circles; not visible to the monkey) either at the location (left, right, up, or down) of the previously flashed cue (CUE) or opposite of it. Disappearance of the central hand stimulus after a variable memory period (MEM) instructed the movement (MOV). Visual feedback (FDB) appeared only after the monkey touched the correct goal on the screen (success) or after a wait time had expired short enough not to allow posture corrections (failure). B , Localization of recording sites (cross indicates estimated center) in the PRR in the medial intraparietal sulcus (IPS) of monkey TI. (Shaded areas in the left one-half of the axial view are imaging artifacts.) C , Typical delay period tuning in PRR. Outer panels, Response profiles (PSTHs; aligned with movement instruction at t = 0; smoothed with 50 ms Gaussian kernel) of a single example neuron strongly selective for the four different locations of the cue/reach goal during the standard memory-guided pro-reach. The inner panel shows the corresponding tuning curve with the resulting directional tuning vector (see Materials and Methods) for the late memory period (shaded area in PSTHs). Smoothed PSTHs or interpolated curves were only used for illustration; all quantitative analyses were performed based on direct spike counts.
- Figure 2.
A , Example neuron with sustained motor goal tuning. A comparison of the PSTHs for up (dark, continuous) and down (light, dotted) cue positions during pro- (green) and anti-reaches (blue) shows that this neuron is only activated by a motor goal “up,” independent of the cue position. Note that, for each neuron, all eight possible conditions were recorded; responses for the remaining four conditions were intermediate and are not shown for convenience. B , Example neuron with visuomotor tuning. Conventions are as in A only with left/right instead of up/down positions shown. This neuron is first tuned for left-side cues, and later for left-side motor goals. Abbreviations are as defined in the legend to Figure 1.
- Figure 3.
A , Time course of total decode performance and learning curves. The performance for predicting all eight trial conditions (4 directions by 2 task rules; 12.5% chance level) becomes significant ∼100 ms after cue onset and reaches its maximum during the early memory period (e-MEM). The decline in the later memory period (l-MEM) can be explained by the prevalence of pure motor goal tuning of most neurons at this time, which allows decoding the motor goal well (Fig. 3 B), but the task rule poorly (Fig. 6). Inset, Prediction performance improves monotonically with the number of neurons used for classification. Note that the neurons were randomly selected among all neurons recorded in the region of interest, regardless of their tuning properties or task relatedness (see Materials and Methods). B , Dynamics of sensory versus motor representations. The classifier was trained with pro-reach trials only (4-way classification of direction; 25% chance level), and then used to predict a direction in anti-reach trials. The performance value denotes the probability that the predicted direction coincides with the cue or motor goal position, respectively. This decode revealed a moderate transient representation of the cue position during part of the cue period (light curve). This sensory representation is then quickly replaced by a strong motor goal representation in the population of PRR neurons (dark curve). Note that correct prediction of the motor goal and the cue are mutually exclusive in this kind of decode, which explains the below-chance performances (see Results). Because of this statistical dependence, performance values for the cue versus motor goal prediction were each tested for being larger than chance level (t test, p < 0.05/0.01), not for being different from each other. Other abbreviations are as defined in the legend to Figure 1. Error bars indicate SEM.
- Figure 4.
Spatiotemporal tuning classes. Comparing spatial tuning of all 141 neurons in cue and late memory period reveals few major classes (classes with population densities <5% are in italics): (1) neurons with pure motor goal tuning [not tuned (37) or tuned during pro-reaches only (19) in the cue period, and tuned for the motor goal during the memory period], (2) neurons with visuomotor tuning [visual tuning during cue period, motor goal tuning during memory period (10)], (3) very few neurons specifically tuned during pro- or anti-reaches only [mostly during the memory period (6 of 8)], and (4) neurons not tuned in both cue and memory period (21). Neurons that are tuned neither during the cue nor the memory period might be tuned during later periods of the trial. We here focused on the transition from the cue to the memory period and discuss the motor goal and sensorimotor neurons, because we were mainly interested in the neural dynamics of the cue-to-goal representation in PRR.
- Figure 5.
A , Motor goal representation in pro- versus anti-reaches. Prediction performance for the motor goal was delayed in anti- (dark) compared with pro-reaches (light). Curves show the average performance in the eight-way total decode (12.5% chance level) when analyzed separately for pro- and anti-reach trials. The delay (56 ± 8 ms) was calculated by cross-correlating the two performance curves (see Materials and Methods). B , Reaction times in pro- versus anti-reaches. Arm movement onset of correct trials in a speeded reaction time task was delayed in anti- compared with pro-reaches. [Data were collected at 6 (monkey TA) and 4 (TI) separate days in the highly trained animals.] Abbreviations are as defined in the legend to Figure 1. Error bars indicate SEM.
- Figure 6.
Representation of the task rule in PRR. The prediction of the task rule (mapping conditions, pro/anti) was significantly above chance (50%) already shortly before the cue period (note that the fixation period had constant duration), indicating explicit task rule representations in PRR. Prediction was best at the end of the cue period and during the early memory period. This is probably attributable to additional implicit task rule information conveyed by the different relative spatial tuning of the neurons in pro- and anti-trials (see Discussion). Abbreviations are as defined in the legend to Figure 1. Error bars indicate SEM.
