JOURNALOFNEUROPHYSIOLOGY Vol. 71, No. 3, March 1994. Printed

RAPID

in U.S.A.

PUBLICATION

Modulation of Preparatory Neuronal Activity in Dorsal Premotor Cortex due to Stimulus-Response Compatibility D. J. CRAMMOND AND Dkpartement de Physiologie, SUMMARY

AND

J. F. KALASKA Universite’ de Montrial,

Montreal,

CONCLUSIONS

1. Neuronal activity was recorded in the dorsal premotor cortex (PMd) of two monkeys performing a multidirectional, instructed-delay (ID) reaching task in which visuospatial cues signaled the direction of movement either congruent with the instruction cue (“direct-delay” trials, DD) or redirected 180° opposite to the cue (“redirected-delay” trials, RD). Therefore, this task had two degrees of stimulus-response (S-R) compatibility because in one-half of the trials the spatial attributes of the visual cue were incongruent with those of the intended movement. 2. The majority of PMd cells discharged both at short latency to the RD or DD cues and subsequently with sustained activity during the remaining ID period (IDP). The earliest responses (~250 ms) in both DD and RD trials covaried with cue location and so could be either a “visuospatial” response or a neuronal correlate of the selection of action with highest S-R compatibility, namely move to the stimulus. In contrast, later IDP activity usually covaried with the direction of movement signaled by the cues, independent of their spatial location, supporting the hypothesis that IDP discharge in PMd ultimately encodes attributes of intended reaching movements. INTRODUCTION

A role for the dorsal premotor cortex (PMd) in the preparation for movement is supported by many studies reporting a modulation of neuronal discharge after presentation of visuospatial instructions signaling motor action (Godschalk et al. 1985; Kurata and Wise 1988; Mushiake et al. 199 1; Riehle and Requin 1989). However, task designs in which the location of the instruction signal corresponds to the movement end point introduce ambiguity as to whether preparatory discharge reflects visuospatial sensory or attentional processes, or, is related to movement preparation, per se. Recent studies have dissociated the location of the instruction signal from that of the impending movement target. For example, Alexander and Crutcher ( 1990) found that the instructed-delay period (IDP) activity of many neurons in primary motor cortex (M 1, 40%) and the supplementary motor area (SMA, 36%) was primarily related to the location ( or direction) of the visual instruction and independent of subsequent action. Furthermore, using tasks in which identical visuospatial stimuli cued different motor responses, the IDP activity of 17% of PMd cells was attributed to stimulus attributes alone (Boussaoud and Wise 1993 ), whereas the motor instructional significance of visual cues altered the IDP activity of 63% (di Pellegrino and Wise 1993) and 7 1% (Boussaoud and Wise 1993) of PMd cells. In contrast, Lurito et al. ( 199 1) reported that the reaction time activity of M 1 cells was seldom only movement-

Quebec H3C 3J7, Canada

or stimulus-related in a task requiring movements to be made at an angle from the stimulus in one-half of the trials and argued that this strict dichotomy of function is not evident at the single-cell level of analysis. We have been studying activity in PMd and Ml using a multidirectional reaching task (Crammond and Kalaska 1989, 1990). In the present study the hypothesis that IDP activity in PMd is related to movement preparation was further tested with the use of visuospatial cues to signal movements either toward the cue (direct-delay trials, DD) or in the diametrically opposite direction (redirected-delay trials, RD). In contrast to some previous findings, the IDP activity of the majority of PMd neurons reflected spatial features of both the instruction and the motor response. METHODS

Subjects Two adult rhesus monkeys (Macaca mulatta), 4-5 kg, were seated in a primate chair with their heads fixed. They were trained to make visually guided reaching movements from a central position to any of eight equally spaced LED triplets (red, green, and yellow) at a radius of 8 cm, using an apparatus that is described elsewhere (Kalaska et al. 1989).

Instructed-Delay

(ID) Tasks, Recording, and Analysis

Each cell was initially tested in an eight-direction ID task (Crammond and Kalaska 1989) to identify its preferred direction (PD) of movement and then tested in this modified task. Each trial began by positioning the handle over the central red LED for a variable period of time (Fig. 1, Pre-CUE range 1.O-3.0 s). Subsequently, a CUE stimulus appeared in one of two locations corresponding to the PD of the cell or in the opposite direction (oPD). A green CUE signaled movement towards the stimulus in DD trials; a yellow CUE signaled movement away from the stimulus in the opposite direction in RD trials. The CUE remained illuminated for 1.0-3.0 s, after which it and the central red LED were extinguished and the peripheral red target LED was illuminated (GO signal). Trials of both types and both directions were presented in a randomized block design. Conventional single-unit recording techniques were employed. Mean discharge rate, including partial spike intervals, was calculated for each trial for the first 250 ms of the IDP after CUE onset (early IDP) and the last 500 ms before the appearance of the GO signal (late IDP). A t test was then used to compare cell responses in RD and DD trials as a function of either CUE location or movement direction. RESULTS

Ninety-seven cells that were directionally tuned in the IDP before reaching in eight directions were tested during

0022-3077/94 $3.00 Copyright 0 1994 The American Physiological Society

1281

1282

D. J. CRAMMOND

AND

CUE

. . .

I

0

DIRECT:

a

.

p-y--p~gy--q

REDIRECTED:

I

// // Pre-Cue

I I I

// // Cue

I I I

// // Movement

I

FIG. 1. The sequence of visual stimulus events during direct-delay trials (DD) (top row) and redirected delay trials (RD) (bottom row), signaling movement (direction 45 O) to the top right target. In DD trials, a green CUE (large crossed circle) appears at target location and remains on until GO signal, i.e., extinction of central red (large solid circle) LED and illumination of red LED at target. In RD trials, a yellow (large circle) CUE appears at the target opposite to the intended movement, i.e., at 2 15”.

performance of the DD and RD tasks in two directions. Localization of these cells to PMd was determined using the following two criteria: 1) the absence of any measurable evoked motor activity during intracortical microstimulation ( 11 pulses, 0.2 ms duration, 330 Hz) using currents of up to 50 PA and 2) histological verification that electrode penetrations were in the immediate vicinity of, or anterior to, the superior precentral sulcus (in preparation). In DD trials, IDP activity typically comprised two temporal components of neuronal modulation: a brief, short-latency phasic burst (or pause) of early IDP activity after appearance of a CUE stimulus (Figs. 2 and 3), which was immediately followed by either a sustained increase (or decrease) in tonic discharge or a gradually incrementing (or decrementing) ramp-like pattern of late IDP activity. Early IDP activity alone was recorded in 11 cells, whereas late IDP activity alone was recorded in 23 cells. Sixty-one PMd cells had both early and late IDP responses both of which had similar directional tuning properties. The remaining two cells had directionally tuned activity in other IDP epochs. IDP responses recorded at the PD and oPD in DD trials did not differ between the two- and eight-direction tasks. Comparison ofIDP and RD tasks

activity ofPMd

cells in DD

Eighty-four ( 86.6% ) PMd cells showed a modulation of IDP discharge in RD compared with DD trials. For example, the cells in Figs. 2 and 3 responded at short latency with a brief phasic burst after onset of either the DD (A) or RD (D) CUE at the cells’ PD. As typified by the initial phasic bursts of these two cases, early IDP activity often appeared to be very similar in DD and RD trials with cues in the same spatial location. However, by the final 500 ms the late IDP activity had diverged so that the directionality of neuronal discharge was appropriate for the direction of the intended movement whether at the PD (cf. A and C) or oPD (cf. B and 0). Because in most cells the late IDP activity was reciprocal between the PD and oPD in the DD task, the

J. F. KALASKA

effect of the RD CUE was an apparent “reversal” of the sign of late IDP activity. This effect was verified by statistical analysis. For instance, in DD trials, 58 of 97 PMd cells were directionally tuned in the first 250 ms after CUE appearance at the PD and oPD locations. Of these, 40 of 58 (69.0%) showed the same response when cues appeared at the PD in DD and RD trials (Fig. 2, A and D), and 4 1 of 58 (70.7%) when cues were at the oPD. Therefore, the earliest IDP response of most cells covaried with CUE location. Nevertheless, the remaining 18 and 17 cells, respectively, began to show modulations in cell activity in RD trials compared with DD trials within the first 250 ms after CUE presentation (Fig. 2, B and C, and Fig. 3, A and D; and B and C). In contrast, during the final 500 ms of the IDP, 84 of 97 ( 86.6%) PMd cells were directionally tuned in DD trials. Of these, 77 of 84 (91.7%) showed statistically different responses for cues at the PD in RD versus DD trials, i.e., the discharge of most cells no longer covaried with CUE location at the end of the IDP (Fig. 2, A and C; and B and D, and Fig. 3, B and D). Instead, 52 of 84 cells (6 1.9%) now showed statistically similar responses for DD and RD cues that signaled movement to be at the PD, indicating that the late IDP activity of those cells covaried with intended movement direction independent of the spatial location of the instructional cues (Fig. 2, B and D and Fig. 3, B and 0). The remaining 25 of 77 cells with modulated late IDP activity likewise showed a “reversal” of the directional sign of activity in RD versus DD trials, but their late IDP activity was not identical to that in DD trials for the same movement direction. In some cases, the late IDP activity was stronger in RD trials than in DD trials (Fig. 2, A and C and Fig. 3, A and C), and in others it was weaker. These cells, therefore, showed an interaction between cue location and movement direction in their late IDP activity in RD trials. It is also possible that this interaction reflected a change in the preferred direction of the cells between DD and RD trials (di Pellegrino and Wise 1993 ), but the use of only two directions in this task precludes us from showing such an effect. Nevertheless, although the late IDP discharge for DD and RD trials in the same intended movement direction were not always quantitatively identical, they were qualitatively similar in showing the same sign of activity change (excitation or suppression of discharge; Figs. 2 and 3). Comparison

of IDP activity of Ml cells in DD and RD tasks

Of the small sample of 26 motor cortex cells studied, 7 (26.9%) had significant and directionally tuned modulation of IDP activity in the DD task. Significant modulation of early IDP activity was observed in one of six neurons and in six of seven cells with late IDP activity after presentation of RD CUES. In four of these latter cells late activity reversed fully to resemble late IDP activity in DD trials. DISCUSSION

Very few PMd cells responded only in relation to the spatial attributes of either the instructional stimuli or the eventual movement direction in each trial. These results do not support the existence of two segregated populations of cells in PMd that are predominately sensory or motor, vi-

MODULATED

PREPARATORY

NEURONAL

ACTIVITY

IN

PMD

1283

FIG. 2. Response of a dorsal premotor cortex ( PMd) cell with both early and late instructed delay period (IDP) activity in DD (A and B) and RD ( C and D) trials for movements at cell’s preferred direction (O”, A and C) and opposite PD ( 180”, B and 0). The two vertical dashed lines indicate the times of onset of the CUE (Zefl) and GO (rig&) signals superimposed on single trial rasters (top) and peristimulus time histograms (bottom). Horizontal scale bar: 500 ms, vertical scale bar: 20 impulses/s. In DD trials, the green CUE at 0” (A ) evoked both early (strong phasic burst) and late ( sustained tonic) IDP activity and at 180” (B) caused a suppression of activity in both IDP epochs. In RD trials, a yellow CUE at 180” ( C) evoked a transient suppression of early IDP activity that was followed by a sustained level of late IDP activity and at 0” (0) evoked an early IDP response as in A and late IDP response as in B.

suospatial or visuomotor, during the IDP. Instead, nearly all of the cells tested in this task showed IDP activity whose sign covaried with both cue spatial location and with the eventual direction of movement at different times in the RD trials, but whose discharge did not always quantitatively fit a simple sensory versus motor dichotomy. This task-specific modulation of IDP activity in RD trials can be interpreted in several different ways. One explanation implicates PMd in the process of sensorimotor transformation (Kalaska and Crammond 1992). It suggests that the earliest response component of many cells is predominantly visuospatial, signaling the location of a behaviorally relevant sensory cue. Later IDP activity is visuomotor, signaling the intended movement direction instructed by the cue. According to this explanation, the sensory versus motor dichotomy is still valid qualitatively, but is expressed at different times in the discharge of individual cells, rather than predominantly by different populations of cells. A second possibility involves the effects of attention and the internal representation of target locations. di Pellegrino and Wise ( 1993) reported a significant number of PMd neurons with sustained IDP activity when the subject had to direct and maintain attention at the remembered location of instructional cues. It is possible that the modulation of IDP activity in RD trials reflects an internal redefinition

of intended target location or redirection of the monkeys’ attention to the learned target location opposite to the RD cue, immediately after its appearance. The absence of control over the direction of attention in this task and the likelihood that its spatial location, or that of any internally redefined target, covaried with the spatial attributes of the intended movements does not allow us to distinguish among these possibilities. Yet another explanation, based on the concept of S-R compatibility (Fitts and Seeger 1953), suggests that IDP activity reflects the selection of the appropriate response to a stimulus (di Pellegrino and Wise 1993; Mitz et al. 199 1). According to this proposal, even the earliest responses after RD cues may be more “motor” than “sensory” because they reflect the selection and initial preparation of the response of highest S-R compatibility, namely reaching toward the spatial location of the cue. In RD trials, however, this is the incorrect response. The rapidly occurring and internally derived modulation of cell IDP activity in RD trials presumably reflects several congruent processes including recognition of the information content of the instruction cue and the extra neuronal processing required to recode the selection of the motor response of lower S-R compatibility, to satisfy the correct associative rule. Our results suggest that nearly all cells with directionally tuned IDP activity in PMd would contribute to this selection pro-

1284

D. J. CRAMMOND

AND J. F. KALASKA

FIG. 3. Response of a PMd cell with more complex behavior in the RD task. Same format as Fig. 2. In DD trials, a green CUE at 0” (A ) evoked an early phasic IDP response and at 180° (B) caused a sustained suppression of both early and late IDP activity. In RD trials, a yellow CUE at 180’ ( C) initially evoked a transient suppression of the early IDP response as in (B) that was immediately followed by a strong phasic burst and weak late IDP activity. Note this delayed phasic burst did not overlap in time with the early IDP response recorded in A or D and that there was no visual stimulus in the right hemifield or observed eye movement to account for this activity. A yellow CUE at O” (D) evoked early IDP activity as in A and a suppression of late IDP activity as in B.

cess. The finding that late IDP activity for the same intended movement direction in DD versus RD tasks is not always quantitatively identical may be a neuronal correlate of the selection of similar motor responses under conditions of different associative rules with differing degrees of spatial compatibility ( di Pellegrino and Wise 1993) . Notwithstanding the smaller proportion of M 1 neurons encoding a directional signal in the IDP of this task, a similar transformation process appears to occur in both PMd and M 1. This work was supported by Medical Research Council Grant MT-7693 to the MRC Research Group in Neurological Sciences and a Fonds de la Recherche en Sante du Quebec Fellowship. Present address of D. J. Crammond: Laboratory of Neurophysiology, National Institute of Mental Health, Poolesville, MD 20837. Address reprint requests to J. F. Kalaska. Received 26 October 1993; accepted in final form 8 December 1993. REFERENCES ALEXANDER, G. E. AND CRUTCHER, M. D. Neural representations of the target (goal) of visually guided arm movements in three motor areas of the monkey. J. Neurophysiol. 64: 164-178, 1990. BOUSSAOUD, D. AND WISE, S. P. Primate frontal cortex-effects of stimulus and movement. Exp. Brain Res. 95: 28-40, 1993. CRAMMOND, D. J. AND KALASKA, J. F. Comparison of cell activity in cortical areas 6,4, and 5 recorded in an instructed-delay task. Sot. Neurosci. Abstr. 15: 383.5, 1989. CRAMMOND, D. J. AND KALASKA, J. F. Cortical neuronal activity recorded in a delay task that dissociates location of cue stimulus and movement endpoint. Sot. Neurosci. Abstr. 16: 178.12, 1990.

G. AND WISE, S. P. Visuospatial versus visuomotor activity in the premotor and prefrontal cortex of a primate. J. Neurosci. 13: 1227-1243, 1993. FITTS, P. M. AND SEEGER, C. M. S-R compatibility: spatial characteristics of stimulus and response codes. J. Exp. Psychol. 46: 199-210, 1953. GODSCHALK, M., LEMON, R. N., KUYPERS, H. G. J. M., AND VAN DER STEEN, J. The involvement of monkey premotor cortex neurons in preparation of visually cued arm movements. Behav. Brain Res. 18: 143157, 1985. KALASKA, J. F. AND CRAMMOND, D. J. Cerebral cortical mechanisms of reaching movements. Science Wash. DC 255: 15 17-l 523, 1992. KALASKA, J. F., COHEN, D. A. D., HYDE, M. L., AND PRUD’HOMME, M. A comparison of movement direction-related versus load direction-related activity in primate motor cortex, using a two-dimensional reaching task. J. Neurosci. 9: 2080-2 102, 1989. KURATA, K. AND WISE, S. P. Premotor cortex of rhesus monkeys: set-related activity during two conditional motor tasks. Exp. Brain Res. 69:

DI PELLEGRINO,

327-343, 1988. LURITO, J. T., GEORGAKOPOULOS,

T., AND GEORGOPOULOS, A. P. Cognitive spatial-motor processes. 7. The making of movements at an angle from a stimulus direction: studies of motor cortical activity at the single cell and population levels. Exp. Brain Res. 87: 562-580, 199 1. MITZ, A. R., GODSCHALK, M., AND WISE, S. P. Learning-dependent neuronal activity in the premotor cortex: activity during the acquisition of conditional motor associations. J. Neurosci. 11: 1855- 1872, 199 1. MUSHIAKE, H., INASE, M., AND TANJI, J. Neuronal activity in the primate premotor, supplementary, and precentral motor cortex during visually guided and internally determined sequential movements. J. Neurophysiol. 66: 705-718, 1991. RIEHLE, A. AND REQUIN, J. Monkey primary motor and premotor cortex: single-cell activity related to prior information about direction and extent of an intended movement. J. Neurophysiol. 6 1: 534-549, 1989.