INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

For purposeful and efficient performance of motor action, preparatory processes often follow the process of motor selection, before the actual execution of the required movements (Evarts et al. 1984). When preparing a reaching movement, subjects must develop readiness to reach for a particular motor target using either the right or left arm. Neuronal activity in the cerebral cortex of primates related to motor preparation has been referred to as motor-set-related or, simply, as set-related activity (Johnson et al. 1996; Scott et al. 1997; Tanji and Evarts 1976; Weinrich and Wise 1982), although the processes of motor selection and preparation have not been separated in previous reports. Set-related activity is prominent in both the ventral (PMv) and dorsal (PMd) parts of the premotor cortex (di Pellegrino and Wise 1993; Godschalk et al. 1985; Kurata 1989; Kurata and Wise 1988; Weinrich et al. 1984; Wise 1985). Since the PMv and PMd have different anatomical connectivities (Barbas and Pandya 1987; Luppino et al. 1999; Matelli et al. 1998) and are viewed as two separate areas, with neuronal activity emphasizing different aspects of behavioral factors (Boussaoud and Wise 1993a,b; Kurata 1993; Kurata and Hoffman 1994; Kurata and Hoshi 1999), it is of interest to study the participation of each area in different aspects of motor preparation. To what extent are the readiness to acquire a target in space and the readiness to use a selected arm represented in each part of the premotor cortex? To answer this question, we analyzed monkeys' neuronal activity while performing a motor task in which the processes of motor selection and the development of preparation were largely separated. Here, we show that both target-location and arm-use were reflected in the preparatory activity in PMd neurons, whereas the readiness to reach for a particular target was a dominant factor for PMv neurons.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals and apparatus

We used two male monkeys (Macaca fuscata, 8 kg), which were cared for according to the National Institutes of Health guidelines and the Guidelines for Animal Care and Use published by our institute. During experimental sessions, each monkey sat in a chair with its head restrained and placed its hands on two touch pads, 17 cm apart, in front of the chair. We used the same experimental methods as described previously (Hoshi et al. 2000) to monitor and record single-unit activity, electromyographic (EMG) activity, and eye positions.

Behavioral task

The monkeys were trained to perform a target-reaching task by following two sets of instructions indicating which target to reach and which arm to use (Fig. 1A). The task started when the animal placed a hand on each touch pad, after an intertrial interval of 3 s, and gazed at a fixation point (FP) that appeared in the center of the monitor screen. If fixation was maintained for 1.2 s, the monkey was given the first instruction (400 ms) about the arm or target. The instruction cue was the appearance of a central color cue at the FP and a white square (8 × 8°) to the left or right of the FP. The small color cue, which covered the central FP and appeared at the same time as the white square, showed whether the instruction was for the arm or target. For monkey 1, a green circle or red square indicated the arm instruction, and a blue circle or red cross indicated the target instruction. For monkey 2, a green square and blue cross indicated the arm and target instructions, respectively. A square to the left indicated the left arm (arm instruction) or left target (target instruction), whereas a square to the right indicated the right arm or right target. If fixation was maintained for 1.2 s during the subsequent delay period (the 1st delay), the second instruction (400 ms) was given. Then, if fixation was maintained for 1.2 s during a second delay, squares appeared on either side of the FP (set cue, 1,000 ms), telling the animal to get ready to reach for the target in response to the disappearance of the FP (GO signal); subsequent reaching for the target with a reaction time <1 s was rewarded with fruit juice. The intertarget distance was 55 mm (10.5°). During the set-cue period, the monkeys prepared four movements, as shown in Fig. 1B. Before the appearance of the GO signal, the first monkey was required to fixate on the FP; the second monkey was not required to do so. The order of appearance of the arm and target instructions was alternated in a block of 20 trials. A series of five 250-Hz tones after a reward signaled reversal of the order.

View larger version (31K): [in this window] [in a new window]   Fig. 1. Behavioral task and recording sites. A: temporal sequence of the behavioral events. The top row shows a trial in which the 2 instructions were given in the order "arm" then "target." The bottom row shows a trial in which the 2 instructions were given in the order "target" then "arm." B: the 4 movements performed by the monkeys. During the set-cue period, they prepared to perform the 4 movements indicated by dotted lines. RA, right arm; LA, left arm; RT, right target; LT, left target. C: 3 examples of electromyographic (EMG) activity. Averaged EMG activity for 10 trials of each movement, indicated in the left column, was aligned to the onset of touch pad release (). Scale bar, 1 s. D: cortical map of the recording sites. This report refers to the activity of neurons found in the 2 cortical motor areas: the dorsal (red) and ventral (blue) premotor areas (PMd and PMv, respectively). AS, arcuate sulcus; CS, central sulcus; SPS, superior precentral sulcus; Sp, spur of AS. Scale bar, 5 mm.

While performing the task, we monitored the following muscles bilaterally: biceps and triceps brachii, deltoid, trapezius, flexor and extensor carpi radialis, supraspinatus, infraspinatus, pectoralis major, rhomboid, and the neck and paravertebral muscles. Three examples of EMG activity recorded from the right limb of monkey 2 are shown in Fig. 1C. Although they show movement-related activity, they do not show consistent changes in activity before actual execution of movements.

Recording sites

We first identified cortical sulcal patterns and measured the three-dimensional structure around the recording sites using an ultrasound imaging technique (LOGIQ  System, GE Medical Systems) (see Tokuno et al. 2000). Because both the probe for the ultrasound measurement and electrodes for cell recording were mounted on manipulators on the same stereotactic frame, the precision for the localization of recording sites with reference to cortical landmarks was within a millimeter. Subsequently, we applied intracortical microstimulation through the tips of inserted electrodes (ICMS; 11 to 44 pulses, 200-µs width at 333 Hz, current 5 to 50 µA). In this study, we tentatively defined the primary motor area (MI) as the area where ICMS evoked limb movements or muscle twitches (with more than 50% probability) with currents <40 µA with 11 pulses. The site that we refer to as the PMd was caudal to the genu of the arcuate sulcus (AS), medial to the spur of AS, lateral to the superior precentral sulcus, and rostral to the MI (Fig. 1D). For the PMv, the recording sites were caudal to the AS including its caudal bank and convexity, lateral to the spur, and rostral to the border with MI. Set-related cells were sampled mostly within a rostromedial sector of the PMv.

Data analysis

We classified a neuron as "task-related" if its distribution of discharge rates (spikes per second) in six task periods (prefixation, pre-1st cue, 1st cue and delay, 2nd cue and delay, set cue, and movement) was significantly different in at least one of eight trial types (Friedman test, P < 0.01, corrected). For the set-cue period, we used a time window from set-cue appearance until 200 ms before the GO signal, and determined whether the activity was selective for the combination of the two instruction cues, rather than for the second cue only. For each neuron, we applied a general linear model (GLM) analysis for two categories: one is the four Second Cues (right arm, RA; left arm, LA; right target, RT; and left target, LT), the other is the four Combinations of the first and the second cues (RA-RT, RA-LT, LA-RT, and LA-LT), regardless of the order of the instructions. The model equation is described as

(1)

where ₁ and ₂ are the fitted coefficient and ₀ is the intercept. The two null hypotheses are that ₁ = 0 (i.e., the 2nd cue does not affect the neurons' response) and that ₂ = 0 (i.e., the combination of 2 cues does not affect the neurons' response). Thus if the probability of (₂ = 0) was <0.01, we judged that the neuron had activity reflecting specific combinations of the first and the second cues (i.e., arm and target instructions), and classified the neuron as having "set-related" activity reflecting future movements. For this analysis, we first calculated the discharge rate (spikes per second) of each trial during the set-cue period across all recorded trials, then analyzed the activity using Eq. 1 looking at the second cue itself and the combination of the first and the second cues.

To reveal the relationship to the arm and target for each set-related neuron, we used a three-way ANOVA looking at three factors (arm, target, and order of 2 instructions). Based on this analysis, we classified set-related neurons into three classes: 1) selective only for the arm (P of arm factor <0.01, P of target factor 0.01, P of interaction between arm and target 0.01); 2) selective only for the target (P of arm factor 0.01, P of target factor <0.01, P of interaction between arm and target 0.01); and 3) selective for both arm and target (P of arm and target factor <0.01, or P of interaction between arm and target <0.01).

To assess how much the arm and target were represented in neuronal activity, we used two indexes defined with the following equations

(2)

(3)

In the equations, D is the average discharge rate in the set-cue period, categorized with an appropriate subscript. Thus, D_RA-RT means the discharge rate during the set-cue period before reaching to the right target with the right arm. The indexes, ranging from 1 to +1, include information on laterality. If "right" is greater than "left," the index is positive, and vice versa. Since we recorded neuronal activity from the left hemisphere, "right" means "contralateral," and "left" means "ipsilateral."

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We found task-related activity in 401 PMd neurons (monkey 1, 179; monkey 2, 222) and 271 PMv neurons (monkey 1, 122; monkey 2, 149). In this report, we focus on neuronal activity in the set-cue period (i.e., from the appearance of the set-cue to the onset of the GO signal). Neuronal activity in other task periods will be the subject of separate papers. Of the 401 PMd neurons, 211 exhibited set-cue period activity that was selective for the combination of the two instruction cues (set-related activity, see METHODS). Among the remaining 190 neurons, 55 showed nonselective increase of activity during the set-cue period compared with the pre-first cue period (paired t-test, P < 0.01). Based on the three-way ANOVA, a majority of the 211 set-related neurons (n = 112, 53%) was classified as selective for both the arm and target (Table 1). A prominent example of a PMd neuron in this category is shown in Fig. 2A, where the activity increased markedly if the combination of the two instructions was "left arm" and "right target" (LA-RT, the 4th row in Fig. 2A). The neuron's activity was selective for both the arm and the target (3-way ANOVA; arm factor, P < 0.001; target factor, P < 0.001; interaction of arm and target, P < 0.001). Its arm index was 0.972, and its target index was 0.837 (Eqs. 2 and 3 in METHODS).

View larger version (28K): [in this window] [in a new window]   Fig. 2. Two examples of neuronal activity. A: activity of a neuron in the PMd. In the raster displays, each row represents a trial and each dot shows when the cell discharged. The 8 marks in each trial represent when the task events appeared, as denoted in the 1st panel of the left column. The 1st and 2nd instructions in each panel are shown on the top (RA, right arm; LA, left arm; RT, right target; LT, left target; 1, the 1st cue; 2, the 2nd cue). Below each raster display, spike density functions (SDFs) are drawn (sigma = 20 ms). The ordinate represents the instantaneous firing rate (spikes per second). Both raster and SDFs are aligned to the onset of the set cue. This neuron was active in the set-cue period if the monkey prepared to reach the right target with the left-arm (the 4th row), but was less active during preparation for the other 3 movements. B: activity of a neuron in the PMv. This neuron was active in the set-cue period if the monkey prepared to reach the left target (the 2nd and 3rd rows), regardless of the arm to be used.

Of the 271 task-related PMv neurons, 109 exhibited set-related activity (see METHODS). Among the remaining 162 neurons, 73 showed nonselective increase of activity during the set-cue period compared with the prefirst cue period (paired t-test, P < 0.01). We found that the most pronounced activity properties among PMv neurons differed from those among PMd neurons. The majority of the set-related PMv neurons (n = 65, 60%) were selective only for the target (3-way ANOVA, see Table 1). An example of a PMv neuron exhibiting a prominent target selectivity is shown in Fig. 2B, where the set-related activity was intense if the target was on the left, regardless of arm use (3-way ANOVA; arm factor, P = 0.385; target factor, P < 0.001; interaction of arm and target, P = 0.788: arm index, 0.006; target index, 0.954). On the other hand, the number of neurons showing both arm and target selectivity was smaller in the PMv than in the PMd (² test, P < 0.001; Table 1). Furthermore, activity selective only for the arm was rare in the PMv.

Subsequently, we performed a series of quantitative analyses. We first examined the distribution of the arm and target indexes between PMd (n = 211) and PMv (n = 109) neurons. As plotted in the scatter diagram in Fig. 3A, both the target and arm indexes for PMd neurons were broadly distributed. In contrast, for PMv neurons, the arm index was distributed around the central vertical line indicating a value of zero, while the target index was widely distributed (Fig. 3B). The histograms summing the number of neurons for each index, shown at the top and to the right of the scattergram, indicate that a positive arm index was more frequent for PMd neurons. The distribution of positive and negative indexes was significantly different (P = 0.003 by t-test, with a mean index of 0.082), in favor of contralateral-arm selectivity. As for the target index, the distribution of positive and negative values did not differ (P = 0.378 by t-test, mean = 0.01), indicating nonpreferential selectivity for the right/left target. On the other hand, for PMv neurons, the distributions of the arm and target indexes did not suggest a preference in laterality (mean of arm index = 0.003, P = 0.884 by t-test; mean of target index = 0.01, P = 0.829). Next, we attempted to compare the distribution of the arm and target index directly, by examining the cumulative frequency of each index among neurons in each area. The cumulative distributions of the absolute values of the arm and target indexes for PMd neurons are plotted in Fig. 3C. The distributions did not differ significantly (Kolmogorov-Smirnov test, P = 0.105). In contrast, for PMv neurons, the value of the target index was larger than that of the arm index (significant difference of distribution at the level of P < 0.001). Finally, we compared the distribution of the arm (Fig. 3E) and target (Fig. 3F) indexes between PMd and PMv neurons. The absolute value of the arm index was larger for PMd neurons, with a significant distribution difference (P < 0.001, Kolmogorov-Smirnov test). In contrast, the value for the target index was larger for PMv neurons (P = 0.002).

View larger version (44K): [in this window] [in a new window]   Fig. 3. Quantitative analysis. A and B: scatter plot displays of arm index (horizontal axis) vs. target index (vertical axis) for set-related activity. A: data for the PMd. An asterisk indicates the neuron shown in Fig. 2A. B: data for the PMv. Double asterisks indicate the neuron shown in Fig. 2B. Univariate density histograms are drawn on top (for the arm index) and to the right (for the target index) of the scattergram, with arrows indicating the mean of each index. Shaded histograms indicate neurons with significant selectivity for the arm (horizontal axis) or target (vertical axis). A black arrow indicates that the mean was different from zero, and a gray one indicates that the mean was not different from zero. C and D: cumulative frequency histograms of the absolute arm index (dotted line) and absolute target index (solid line). C: data for the PMd. D: data for the PMv. E and F: cumulative frequency histograms of the absolute arm index (E) and absolute target index (F) of the set-related activity for PMd (black line) and PMv (gray line) neurons.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this study, we compared preparatory neuronal activity in the PMd and PMv while monkeys were performing a behavioral task that prompted readiness to start a reaching action, which was selected using instruction cues given before the preparatory period. We found that the majority of PMd neurons exhibited activity tuned to both target location and arm use in the preparatory period for a subsequent arm-reaching action. In contrast, PMv neurons predominantly showed selectivity for target location, and much less selectivity for arm use.

Is it possible that the neuronal activity during the preparatory period reflected oculomotor or limb motor activity? For the first monkey, the requirement to gaze at the fixation point prevented eye movements during the set-cue period. For the second monkey, we analyzed eye movements and positions during the set-cue period, and did not detect any trends for any particular movement patterns or specific eye positions. Furthermore, the time courses and patterns of set-related neuronal activity recorded from the two monkeys were not different. As for the limb-motor activity, we did not detect any consistent changes in muscle activity monitored by EMG recordings (24 limb and trunk muscles). For these reasons, there seems to be little, if any, possibility that limb or oculomotor activity accounted for the set-related neuronal activity. Further, it is unlikely that the activity was related to head movements or orienting responses, because we found no task-related activation in muscles in the neck or in upper-cervical paravertebral muscles.

The PMd and PMv are viewed as separate areas with different cytoarchitecture (Barbas and Pandya 1987) and anatomical connectivity (Caminiti et al. 1998; Luppino et al. 1999; Matelli et al. 1998; Rizzolatti et al. 1998; Wise et al. 1997). Accumulating evidence indicates that these two areas are involved in different aspects of motor behavior. Lesion studies have suggested that the PMd plays a role in associating visual information with movements (Halsband and Passingham 1985; Kurata and Hoffman 1994; Petrides 1985). Neurons in the PMd respond to the appearance of cues instructing future movements (Crammond and Kalaska 1994; Johnson et al. 1996; Weinrich and Wise 1982) and exhibit sustained activity during the subsequent motor-set period, reflecting such motor parameters of forthcoming movements as direction and amplitude (Johnson et al. 1999; Kurata 1993; Kurata and Wise 1988; Messier and Kalaska 2000; Wise and Mauritz 1985), movement trajectory (Hocherman and Wise 1990), motor target (Shen and Alexander 1997), or force (Hepp-Reymond et al. 1999). Caminiti and co-workers reported that PMd neurons combine visual and somatic information for visual reaching (Burnod et al. 1992; Caminiti et al. 1991). In the process of motor selection, information about motor target and arm use appear to be integrated in the PMd (Hoshi and Tanji 2000). On the other hand, lesions in the PMv induce a tendency to select an object ipsilateral to the lesioned area (Schieber 2000), attentional deficits in the peri-personal space (Rizzolatti et al. 1983), or deficits in shift-prism adaptation (Kurata and Hoshi 1999) or in hand-preshaping for grasping objects (Fogassi et al. 2001). Neuronal activity in the PMv is reported to reflect target location (Gentilucci et al. 1988; Godschalk et al. 1985; Mushiake et al. 1997), the three-dimensional shape of motor targets (Murata et al. 1997), or peripersonal space (Fogassi et al. 1996; Graziano et al. 1997).

Direct comparison of neuronal activity in the PMd and PMv of the same individuals revealed differences during the instruction cue, motor-set, and movement periods (Boussaoud and Wise 1993a,b). PMv neurons reflected spatial attention to the appearance of target objects and sensorial processing of visual information concerning motor targets, whereas PMd neurons reflected motor instruction or the execution of movements reaching to different targets. Our study extended these findings, pointing to preferences in the use of neuronal activity in the PMd and PMv during the preparatory period before the initiation of intended actions. A majority of PMd neurons reflect preparation for action, whereas PMv neurons are involved more in preparation to acquire a target, regardless of the arm used to achieve that action. We need to add, however, that the differences we found here are likely to be reflecting a part of a broad range of functional aspects involving selective use of the PMd and PMv.