Acta Psychologica North-Holland
61 (1986)
53-70
53
INFLUENCE OF STIMULUS-RESPONSE TRANSLATIONS RESPONSE PROGRAMMING: EXAMINING THE RELATIONSHIP OF ARM, DIRECTION, AND EXTENT OF MOVEMENT * Douglas
ON
D. Larish
University
of Iowa,
Accepted
April
USA
1985
Two experiments examined whether the programming relationships of direction and extent of movement (experiment 1) and arm, direction, and extent of movement (experiment 2) are affected by a cognitive recoding process called a stimulus-response translation. The programming of these task-defined parameters was studied via the movement precue method. The effect of a spatial translation was studied by manipulating stimulus-response compatibility. The results from both experiments showed that the patterns of reaction time for these parameters could be altered by indirect or noncompatible stimulus-response mappings. It was concluded that, when stimulus-response compatibility was deficient, effects that might be attributed to programming processes were due instead to the translation process. In experiment 2, the findings obtained from a spatially compatible stimulus-response ensemble demonstrated that the movement parameters of arm, direction, and extent could be selectively manipulated via the precue method. Therefore, it was concluded that this method may be a useful tool in understanding how motoric decisions are made prior to movement, but only when the spatial mapping among stimuli and responses is maximally compatible.
For over two decades, motor psychologists have been trying to understand the processes that underlie the programming and subsequent execution of voluntary, goal-directed motor acts. To this end, Rosenbaum (1980) introduced a modification of the partial advance information paradigm (Leonard 1958) - the movement precuing method - as * The initial experiment reported here is a portion of a doctoral dissertation submitted to the University of Wisconsin-Madison under the supervision of George E. Stelmach, to whom the author extends his gratitude. Support for the dissertation research was provided by Air Force Grant AFSOR-78-3691 awarded to G.E. Stelmach. Appreciation is extended to V. Diggles, H. McCracken, D. Rosenbaum, A.F. Sanders, H. Zelaznik, and an anonymous reviewer for their comments on an earlier version of this paper and G. Frekany for writing the computer programs to carry out the latter experiment. Requests for reprints should be sent to: D.D. Larish, Dept. of Health and Physical Education, Arizona State University, Tempe, AZ 85287, USA.
OOOl-6918/86/$3.50
0 1986, Elsevier
Science
Publishers
B.V. (North-Holland)
54
D.D.
Larish
/ Stimulus-response
translations
another tool for examining the motoric decisions underlying the programming process. In a typical precuing experiment, subjects execute movements that vary on a number of response parameters, with the individual parameters containing at least two values; e.g., arm (left-right), direction (forward-backward), extent (near-distant). Prior to the onset of a reaction stimulus, advance information, referred to as a precue, is provided about the characteristics of the forthcoming movement. More accurately, the precue is designed to convey either total, partial, or no knowledge about the intended movement. The fundamental assumptions are, of course, that one can preprogram the parameter(s) specified in the precue and that the motoric decisions associated with unspecified movement parameters are completed only after a reaction signal has been presented [l]. Hence, response latency primarily reflects the motoric programming time of any parameter(s) remaining unspecified prior to the reaction stimulus. Although the precuing technique was designed to examine characteristics of response programming after other cognitive (nonmotor) decisions have been made, there is reason to question whether Rosenbaum’s (1980) initial experiment satisfied this fundamental assumption. At the root of the problem was the use of colors as reaction stimuli: each of eight responses was associated with and cued by a different color. Since the stimulus-response (S-R) compatibility of such a mapping is far from optimal, it would be expected to introduce a cognitive recoding process called an S-R translation (Greenwald 1972; Fitts and Seeger 1953; Teichner and Krebs 1974). More specifically, when total or partial uncertainty about the response existed prior to stimulus onset, a non-natural color code-to-position code translation was required before the appropriate response could be planned and executed. It has been shown that the contribution of the translation process to reaction time (RT) can be substantial (Greenwald 1972; Fitts and Seeger 1953; Simon and Craft 1970; Simon and Wolf 1963; Teichner and Krebs 1974; Theios 1975). Further, S-R translations have been associated with a stage of processing not usually concerned with response programming and execution (Kerr 1978; Sanders 1980; Theios [l] In the present context, preprogramming refers to the ability to construct partially or fully a motor program prior to the onset of a reaction signal. Programming refers to those motoric decisions that must be made after the reaction signal has been presented. These operational definitions are consistent with those of Klapp (1978).
D.D.
Larish
/ Stimulus-response
translations
55
1975). Since the association between translation and programming processes has not been extensively stu.died under conditions that emphasize, as the movement precue method does, the response or motoric contribution to information processing, it is reasonable to ask whether the translation process should be considered a component of motoric programming or whether it should be viewed as a completely independent process. To define more clearly the relationship between these processes two issues were addressed here: (a) How does a spatial translation influence the pattern of reaction latencies between various parameters or combinations of parameters, and (b) are translation and programming processes localized in different stages of processing? By combining the precuing method with additive factor methodology (Sternberg 1969) the present investigation sought to examine these questions. Experiment
1
The precue task was performed in four spatial mapping conditions; these differed in the type and degree of spatial compatibility. The movement parameters manipulated were direction (right-left) and extent (near-far) of movement. The first S-R ensemble contained a direct, one-to-one spatial mapping among stimuli and responses. For the remaining S-R ensembles, the direct spatial mapping was altered to produce three conditions in which a spatial translation was required. The effect of the translation process on the programming of direction and extent of movement was determined by comparing the patterns of RT for these two parameters among the different S-R ensembles. If the translation process masks the actual programming relationship, the pattern of RT between direction and extent in the direct mapping condition (minimal translation) should be different than the pattern found in the indirect mapping conditions. It was also possible to examine the stage of processing issue because there was a second independent variable, set-size, that has been shown to interact with S-R translations (Brainard et al. 1962; Broadbent and Gregory 1965; Fitts et al. 1963; Sanders 1980). Set-size was defined by the number of responses possible on any given trial. It was manipulated via the number of movement parameters contained in the precue. An interaction of S-R compatibility with set-size would suggest that these factors are localized in a common processing stage, which Sanders (1980) and Theios (1975) have argued is independent from response programming. This processing stage has been labelled response choice by Sanders and response determination by Theios. Method Subjects Seven volunteers
(female = 4 and male = 3), ranging
in age from
21-30,
were
56
D.D. Larish / Stimulus-response translations
recruited from the University of Wisconsin-Madison and paid $20.00 for their participation. Each individual had participated in an earlier experiment that was similar to the one reported here. Apparatus The apparatus consisted of a response panel, a visual display, and a movement time (MT) clock. The response panel contained a central home key (1.3 cm diameter) and four target keys located 5 cm and 10 cm to the left and right of the home base. These latter targets had diameters of 1.3 cm and 2.6 cm, respectively. The S’s view of the response panel was shielded by a black cloth screen during experimental trials. The visual display was situated at eye level approximately 90 cm from the S. It contained a single, horizontal row of four letters (RLNF) that were easily distinguishable when illuminated. These letters served as direction (R = right; L = left) and extent (N = near; F = far) precues. The visual display also contained a horizontal row of five light-emitting diodes (LED’s), spaced 0.5, cm apart, that served as reaction stimuli. The center LED acted as a warning signal, and the remaining four LED’s were movement signals. These lights subtended the S’s view at an angle of approximately two degrees. An MT clock was also placed alongside the visual display. On-line control of the experiment was coordinated by a Digital Equipment Corporation PDP8e computer. Procedures and design Each trial was begun by depressing the home key, after which a precue was immediately presented. Ss were instructed to use this advance information to preprogram the movement parameter(s) contained in the precue. Once the precue was known, Ss were instructed to fixate the warning signal for the remainder of the trial. After a two-second precue period elapsed, a warning light was presented, and, following a variable foreperiod (600, 800, or 1000 msec), a reaction stimulus was illuminated. When the reaction stimulus appeared, the task was to release the home key and move to the specific response key as quickly and accurately as possible. Feedback about MT was then provided. Since two movement parameters were manipulated, four precue conditions were possible. These were: precue none (4-choice RT), precue extent (2-choice RT), precue direction (2-choice RT), and precue direction-extent (simple-RT). Four stimulus-response mapping conditions were used: spatial-motor compatible (SMC), spatial-motor incompatible (SMI), spatial-transformation (ST), and spatial-transformation incompatible (STI). In the SMC condition, movements were executed horizontally to the left or the right of the midline. The visual display was arranged in a horizontal fashion to provide a direct, one-to-one spatial mapping with the four possible responses (see fig. 1). In the SMI condition, Ss were instructed that movements would always be made in the opposite direction and extent of that signalled by the precues and movement stimuli. In the ST condition, the spatial position of the visual display was rotated to a vertical orientation. The two outside lights indicated the distant extent and the two inside lights indicated the near extent. The two top lights were paired with either the right or left direction and the bottom two lights were paired with the remaining direction. Finally, in the ST1 condition, the display was kept in a vertical orientation
D. D. Lnrish
Spatial
Motor
translations
57
Compatible
0 0 . 0 0 Spatial
/ Stimulus-response
Visual Display
Transformafion
Fig. 1. Stimulus-response (experiment 1).
ensembles for direction and extent in the SMC and ST conditions
while incompatibility was introduced in a manner similar to that described for the SMI condition. Consider the following circumstance: In the ST condition, if the t6p light had been paired with the left-distant response key, then in the ST1 condition, the correct movement was to the right-near response key. To create this latter S-R mapping condition, it had to be introduced after the ST condition was experienced. The independent variables were S-R mapping, Uncertainty, Direction, Extent, and Day (4 x 4 x 2 x 2 x 3); all were within-S factors. Four testing sessions were completed under each S-R mapping condition, with the first designated as practice. Ss were randomly presented the SMC and SMI conditions followed by the ST and ST1 conditions in that order. In each testing session, Ss received seven blocks of 52 trials. Within a block, the 48 precue/movement/foreperiod combinations appeared once. Four catch trials were also included. When an error was committed, the trial was repeated at the end of the block in which it occurred. On the average, one session lasted 50 min. The three principal dependent measures were RT, MT, and performance errors. Four types of errors were monitored on each trial: (a) release of the home key before the movement light was illuminated (anticipation error), (b) movement to an incorrect target key (target error), (c) MT greater than 180 msec (MT error), and (d) RT greater than that established a priori for each S (RT error). On Day 1, the maximum RTs for the SMC, SMI, ST, and ST1 conditions were 500 msec, 800 msec, 700 msec, and 1000 msec, respectively. The maximum RT on Day 2 was set at two standard deviations above the longest RT from Day 1, the maximum RT on Day 3 was set at two standard
D.D. Larish / Stimulus-response translations
58
deviations above the maximum RT from Day 2, and so on. If error rates exceeded lo%, the session was’ terminated and repeated the following day. Results Reaction time The experimental questions of interest were best answered by focusing the data analysis on the S-R mapping X uncertainty conditions interaction, F(9, 54) = 33.54, p < 0.05 (see table 1). To determine how the translation process affected the estimates of response programming time, simple main effect analyses within each S-R mapping condition were performed. These tests were significant (ps < 0.05) and subsequent analysis of pairwise differences was made using Tukey’s HSD procedure. For SMC and SMI, the no-uncertainty condition had the fastest RT, the extent uncertain and direction uncertain conditions had longer RTs, but were equivalent, and the extent-direction uncertain condition had the longest RT. In the ST and ST1 conditions the pattern of results was somewhat different: no uncertainty < extent uncertain < direction uncertain < direction and extent uncertain. To determine if S-R compatibility and set-size interacted, twelve interaction contrasts were computed using ScheffC’s procedure. These contrasts compared RTs of the two-choice conditions (direction uncertain and extent uncertain) to the RT of the four-choice condition (direction-extent uncertain). Interactions involving the direction uncertain condition were significant for the SMC-SMI, SMC-STI, and ST-SMI pairings (ps < 0.05). Interactions involving the extent uncertain condition were significant for the SMC-SMI, SMC-STI, SMI-ST, and ST-STI pairings (ps < 0.05). All the interactions were overadditive in nature and showed that the conditions requiring the most difficult translations (SMI and STI) were more adversely affected by increases in set-size than conditions requiring the easier translations (SMC and ST).
Table Mean S-R
SMC M
SD SMI M
SD
1 reaction mapping
time (msec)
for S-R
Parameter(s)
mapping
and response
uncertainty
conditions
uncertain
None
Extent
Direction
Direction-extent
224 22
253 26
259 26
214 26
233 15
315 28
317 25
352 23
220 11
254 17
268 19
284 23
237 19
313 35
327 38
360 33
ST A4
SD ST1 M
SD
in experiment
1.
D.D. Movement
Larish
/ Stimulus-response
translations
59
time
The main effects that reached significance were Uncertainty, F(3, 18) = 19.06, Direction, F(1, 6) = 8.88, and Extent, F(1, 6) = 111.60, ps < 0.05. Analysis of uncertainty conditions revealed that MT in the no-uncertainty condition was faster than in the extent uncertain, direction uncertain, and direction-extent uncertain conditions, while MTs in these latter conditions were not significantly different (see table 2). For direction, MT to the right response keys was faster than to the left response keys. For extent, MT to the near response keys was faster than to the far response keys. In addition, the direction-extent interaction was significant, F(l, 6) = 31.63, p < 0.05. Post hoc analysis showed that MT was fastest to slowest in the following order: right-near < left-near < right-far < left-far. Performance
errors
All Ss stayed within the bounds of 10%. In fact, error rates in this experiment were quite low. Anticipation errors were virtually nonexistant and were no greater than 0.3%. Reaction time errors were also low, less than 2%, and there was no evidence that these errors increased with translation difficulty. Both target and MT errors occurred more frequently, particularly in the indirect mapping conditions and the direction-extent uncertain condition. These error rates, however, did not exceed 5%. Discussion
The results of this experiment showed that the presence of a spatial translation altered the patterns of RT (hence estimates of programming time) when direction and extent of
Table 2 Mean movement time (msec) for S-R mapping and response uncertainty conditions in experiment 1. S-R mapping
Parameter(s) uncertain None
Extent
Direction
Direction-extent
13
75 10
71 9
19
9 71
16
75
II
10
11
10
11
69
12
14
76
9
10
10
10
69
73
13
14
10
10
10
SMC M SD
10
SMI M SD
ST M SD
ST1 M SD
8
60
D.D.
Larish
/ Stimulus-response
translations
movement were uncertain prior to stimulus onset. When the spatial mapping among stimuli and responses was direct, programming times were equivalent for direction and extent of movement. In contrast, when the direct spatial mapping was altered, RT in the direction uncertain condition was longer than in the extent uncertain condition (ST and STI). These results suggest that the differences in the observed RTs were caused by the translation process and not motoric processes associated with the programming of movement. The overadditive nature of the interactions between S-R compatibility and set-size indicate that the effects of these variables are localized in at least one common stage of processing, presumably response choice. Since response programming operations are thought to be distinct from those occurring in the response choice stage (Sanders 1980; Theios 1975) this finding further suggests that the translation process has the potential to mask one’s view of programming phenomena. Collectively, the results from this initial experiment indicate that it may not be possible to obtain reliable programming effects when the precuing task involves spatial recoding processes. The findings obtained here are also consistent with those reported by Goodman and Kelso (1980). In their initial experiment, the S-R ensemble was designed to require a verbal code-to-position code translation (words and digits served as reaction stimuli). In subsequent experiments, however, the S-R ensemble was spatially compatible. When the translation was a contributing component to RT, a pattern of results was found similar to that of Rosenbaum (1980). In contrast, when the spatially compatible S-R ensemble was used, RTs were equivalent in the one parameter uncertain conditions. A similar result was found in the two parameter uncertain conditions. If one accepts the position that response programming processes are best studied under conditions that maximize S-R compatibility, the results from the compatible condition may still be problematic for the movement precue method. Without being able to demonstrate differences in the times to program direction and extent, is it still plausible to advocate that these parameters can be selectively and independently manipulated via the precue method? This is an important question because others have argued that programming is based on a single unified decision about where to move rather than separate decisions for different task parameters (Goodman and Kelso 1980; Kerr 1976, 1978). Before accepting this latter position, it was deemed appropriate to reexamine the issue of compatibility when programming decisions involved a greater number of movement parameters.
Experiment
2
In this second experiment, the effect of a spatial translation was examined on the programming of arm, direction, and extent of movement. These parameters are identical to those manipulated by Goodman and Kelso (1980) and Rosenbaum (1980). Thus, this experiment not only reexamined the compatibility issue but it also provided a further look at the reproducibility of the results from these two studies.
D.D.
Larish
/ Stimulus-response
translations
61
Method Subjects Sixteen Ss (female = 4 and male the University of Iowa. To acquaint task. each individual participated in S-R ensemble prior to starting this
= 12) were recruited from motor learning classes at Ss with the precue method and basic experimental three practice sessions involving a direct mapping experiment.
Apparatus The apparatus consisted of a response panel, stimulus display, and MT clock. The response panel contained two vertical columns of five reaction switches (Cherry M62-0100 momentary contact switches). The two columns were 21 cm apart. Within each column the center key served as a home base and the remaining four served as target keys (see fig. 2). These targets were situated 3.5 cm and 7.0 cm directly above and below the home key and had diameters of 1.3 cm and 2.6 cm, respectively. Viewing of the response panel was occluded by a black cloth screen. The visual display contained two vertical columns of four LEDs. The two columns were 1.5 cm apart and, within each column, the LEDs were spaced 0.6 cm apart. A final LED was centered among this configuration and served as both a warning signal and central fixation point. The visual display was placed at eye level approximately 90 cm from the S. A MT clock (Hunter Model 120C Klockcounter) was also placed alongside the display. On-line control of the experiment was coordinated by a Digital Equipment Corporation PDP8a computer.
0 0 0 0
0 0 X 0 0
0 0 l
0
vDii$Ly
0 0 0 X 0 0
Response Panel
Fig. 2. Stimulus-response ensemble for compatible condition in experiment 2.
62
D. D. Larish / Stimulus-response translations
Procedures and design Two S-R compatibility conditions were used in this experiment. A compatible or direct S-R ensemble was achieved by positioning the stimulus display vertically. In this way the lights of the display maintained a one-to-one spatial relationship with target keys (see fig. 2). To achieve a noncompatible or indirect S-R ensemble, the stimulus display was rotated to a horizontal position. For this latter condition, an equal number of Ss were randomly assigned to each of two S-R mappings. In the first, top lights meant left arm and bottom lights meant right arm, left lights meant forward direction and right lights meant backward direction, and inside lights meant near extent and outside lights meant far extent. In the second mapping, top lights = right arm, bottom lights = left arm; left lights = backward direction, right lights = forward direction; inside lights = near extent, and outside lights = far extent. The trial sequence consisted of a precue, warning light, and reaction stimulus. The LEDs in the visual display were now used to give both precue and movement information. With three movement parameters, there were eight precue or uncertainty conditions. To provide advance information about all three movement parameters, only one light was presented in the precue. Advance information about two parameters was provided by illuminating two lights in the visual display. For example, to signal right arm-forward direction, the two uppermost lights on the right were turned on. Advance information about one parameter was given by illuminating four lights. For example, when the four top lights were presented, forward direction was precued. When all eight LEDs were illuminated, no advance information about the impending movement was provided. The reaction signal was always one of the lights shown in the precue pattern. Also, whenever the precue contained more than one light, all possible responses were equally likely. Finally, the precue interval lasted three seconds and variable foreperiods following the warning light lasted 700, 900, 1100, or 1300 msec. The presentation order of the compatible and noncompatible conditions was counterbalanced among Ss. For each of these two conditions, Ss participated in five testing sessions. The first two were designated as practice; only data from the latter three sessions were used in the data analysis. A testing session was completed when 384 error-free trials were performed, six replicates of the 64 precue-movement combinations. These trials were presented randomly and, when an error occurred, this trial was returned to the list of remaining trials yet to be completed. The dependent measures recorded were RT, MT, and performance errors. In addition to the four errors recorded in experiment 1, one other was added here: hand error, which was defined as initiating a movement with the incorrect hand. Reaction time errors occurred when RT was greater than 1000 msec and MT errors occurred when MT was longer than 200 msec. The maximum acceptable error rate during a testing session was 10%. If this limit was exceeded, then the session was terminated and repeated the following day. Results Reaction time The data were analyzed via a S-R Mapping (2) X Uncertainty (8) X Direction (2) x Arm (2) x Extent (2) x Days (3) ANOVA; all variables were within-S factors. As
D.D.
Lmish
/ Stimuius-response
translations
63
Table 3 Mean reaction time and movement time (msec) for S-R mapping and response uncertainty conditions of experiment 2. S-R mapping
Parameter(s) uncertain N
A
D
E
AD
AE
DE
ADE
281 56
325 60
345 66
325 60
361 68
348 66
366 66
319 64
110
110
111
111
111
114
40
49
109 24
111
49
27
24
25
32
292 55
386 69
391 12
361
11
473 82
431 13
433 14
486 76
109 34
110
109
111
109
110
110
113
32
25
42
24
21
39
46
Compatible
RT M SD
MT M SD Noncompatible
RT M SD
MT M SD Note:
N = none, A = arm, D = direction, E = extent.
in the earlier experiment, specific interest focused on the S-R Mapping-Uncertainty interaction, which was significant, F(7, 105) = 76.19, p < 0.05 (see table 3). For the compatible condition, post hoc simple main effect tests (followed by Tukey’s HSD procedure) showed that RT in the direction uncertain condition was greater than that for either arm uncertain or extent uncertain, but the RT in these latter two instances were equivalent (see table 3). Further, the RTs for the direction-extent uncertain and arm-direction uncertain conditions were equivalent and both were longer than the RT in the ,arm-extent uncertain condition. For the noncompatible condition, RTs in the direction uncertain and arm uncertain conditions were equivalent, and both were longer than in the extent uncertain condition. The RT for arm-direction uncertain was longer than direction-extent uncertain and arm-extent uncertain, whereas, RTs in these latter two conditions were equivalent. The above results demonstrate once again that deviations from a direct spatial mapping can alter the pattern of RTs for the various parameter uncertainty conditions. To determine if the effects of translation and set-size interacted, fifteen interaction contrasts were computed using Scheffe’s procedure. These contrasts examined the differences in RT between the two S-R mapping conditions for: (a) all combinations (three) of the two parameter uncertain and three parameter uncertain conditions (e.g., direction-extent vs arm-direction-extent), (b) all combinations (three) of one parameter uncertain and three parameter uncertain conditions (e.g., extent vs arm-direction-extent), and (c) all combinations (nine) of one parameter uncertain and two parameter uncertain conditions (e.g., extent vs direction-extent). Post hoc analysis showed that in all but two instances, arm-direction vs arm-direction-extent and arm
64
D.D.
Larish
/ Stimulus-response
translations
vs direction-extent, the interaction contrasts were significant ment 1, these interactions were overadditive in nature.
(ps -C 0.05). As in experi-
Movement time The analysis of MT showed that movements to near targets (M = 95.6 msec, SD = 28.6 msec) were completed more quickly than movements to far targets (M = 125.4 msec, SD = 34.1 msec), F(1, 15) = 192.03, p < 0.05 (see table 3). Movement times to the forward targets (M = 106.6 msec, SD = 26.4 msec) were also shorter than MTs to the backward targets (M = 114.4 msec, SD = 41.2 msec), F(1, 15) = 12.13, p < 0.05. All other main effects and interactions were not significant (p > 0.05). Performance errors The maximum error rate of 10% was exceeded in only two instances and both were associated with the noncompatible condition. One occurred on the second day of practice and the other occurred on the first day of nonpractice. Otherwise, Ss were extremely accurate in performing both the compatible and noncompatible tasks. For all five types of errors, the error rates were less than 1%. Discussion The results from this second experiment demonstrated that the spatial configuration of the S-R ensemble did have a direct effect on the estimates of programming time for arm, direction, and extent of movement. This was true for both the one-parameter uncertain and two-parameter uncertain conditions. Further, the interactions of S-R compatibility and set-size were overadditive, which supports the notion that the effects of these two variables are associated with a common stage of processing. The above findings, then, are in close agreement with two of the principal results obtained in experiment 1. The findings from the compatible mapping condition also indicate that the motoric decisions about direction can be made independently from decisions about arm or extent. Such a position is supported in two ways. First, when direction information was contained in the precue, a reduction was observed in RT relative to those conditions in which direction was not precued. This result suggests that direction can be preprogrammed independently from knowledge about arm or extent. The same cannot be said for these latter two parameters. Second, the time to program direction was longer than the times to program arm or program extent. Moreover, a transitive relationship existed between the programming of a single parameter and the programming of combinations of parameters. Note that the above findings are not consistent with those reported by Goodman and Kelso (1980). It is not clear why such a discrepancy was obtained because the compatible S-R mapping used in this second experiment was identical to that used by Goodman and Kelso. It should be mentioned, however, that the pattern of results described here have been replicated by Larish and Frekany (1985). The difference in programming time for direction relative to extent also conflicts with experiment 1. Although there are several variations in the methodology of these two experiments, perhaps the most significant is the amount of practice Ss received at
D.D.
Lmish
/ Stimulus-response
translations
65
the outset. Ss in the initial experiment were given considerably more practice than the Ss in the second experiment (7000 vs 2000 trials). It is possible that this discrepant result occurred because the less experienced or skilled individuals adopted a different strategy for programming than their more experienced counterparts. For example, an interesting hypothesis is that programming decisions are based on individual parameters when the levels of practice are low. Alternatively, programming may be based on a single decision about where to move when levels of practice are high. Although this strategy-dependent explanation is post hoc, it is consistent with the data reported here and it can be empirically tested. A second and somewhat related hypothesis is that one may be able to preprogram multiple responses when the number of S-R pairs is small (e.g., 2-choice RT) and the amount of practice is high, such as in experiment 1. Rosenbaum and Kornblum (1982) have recently shown that it is possible to prepare multiple responses. Therefore, the above hypotheses have some empirical support, but further experiments are certainly needed to explore these possibilities in greater detail.
General discussion How are the observed programming relationships among various taskdefined movement parameters influenced by a cognitive recoding process referred to as a S-R translation? The potential significance of this question can be put into perspective by first considering what effect the recoding process has on the overall magnitude of RT in precue experiments manipulating arm, direction, and extent of movement. The data from four such experiments are relevant in this regard. Two of these experiments (Rosenbaum 1980: experiment 1; Goodman and Kelso 1980: experiment 1) used S-R ensembles that required a complex S-R translation, whereas, the other two (Goodman and Kelso 1980: experiment 2; SMC condition, experiment 2 of present study) used ensembles that maintained a direct spatial mapping among stimuli and responses. Comparisons involving these data reveal that the presence of the recoding process increases mean RT by approximately 30-40s. This is a rather substantial increase considering that a basic assumption underlying information processing methodology is that experimental tasks should be designed to preclude (or least minimize) processing in stages not under study (Massaro 1975). Consequently, it can be argued that an accurate assessment of motoric planning processes is best achieved when the S-R ensemble is designed to minimize the recoding process, a view also recently expressed by Bauer and Miller (1982). This assertion was further substantiated on two accounts. First, the patterns of RT for the movement parameters studied here were
66
D.D.
Larish
/ Stimulus-response
translations
sensitive to variations in the computational load of response choice processes. Second, the effects of the spatial translation were shown to be localized in a stage of processing not usually associated with response programming (Kerr 1978; Sanders 1980; Theios 1975). This latter finding, however, does not mean that the effect of spatial translations are confined to the response choice stage. Experimental variables can affect more than one stage of processing (see Sternberg 1969). Thus, it is possible that translation processes could also interact with response programming processes. To be confident that translation processes are localized only in response choice, it will be necessary to demonstrate that other programming variables (e.g., complexity) interact with arm, direction, and extent but have an additive affect with S-R compatibility. And even if translation and programming processes are localized within a common stage, it seems necessary to determine their individual as well as collective contribution to RT. Irrespective of the stage of processing issue, the changes in RT attributable to spatial recoding still suggest that reliable programming effects may be masked when the S-R ensemble fails to maximize spatial compatibility. Also recall, Goodman and Kelso (1980) found that RT was sensitive to an S-R translation (verbal-to-position) different from the one studied here. This converging evidence further suggests that the confounding influence of the translation process is generalizable to a variety of noncompatible S-R mappings. Other recent evidence from Larish and Frekany (1985) is also relevant in this regard. They reexamined the programming relationships among arm, direction, and extent via a spatially compatible S-R ensemble, and further modified the precue task to preclude perceptual and decision processing attributed to differences in set-size, a change suggested by Zelaznik et al. (1982). Although the data from Rosenbaum (1980) indicated that these parameters were programmed serially and without regard to a specific order, the results from Larish and Frekany showed that they were programmed in a parallel fashion and in a specified order. In addition, a hierarchical arrangement best characterized the relationsohip among the three parameters. The latter finding also conflicts with one of the principal conclusions made by Rosenbaum (1980). Thus, it appears that the precue method may still be a useful tool for understanding how motoric decisions are made prior to movement initiation, but only when a compatible S-R ensemble is used.
D.D.
Larish
/ Stimulus-response
translations
67
The results obtained here also have implications for the study of parameters other than arm, direction, and extent. Perhaps most critical is that the experimental manipulation of any nonspatial movement parameter may confound translation and programming processes. For example, consider the parameter of movement duration. The programming effects of this parameter have been studied quite extensively within recent years (Klapp 1977; Klapp and Erwin 1976; Klapp et al. -1974; McCracken 1979; Spijkers and Steyvers 1984; Zelaznik and Hahn 1985; Zelaznik et al, 1982) and the available evidence suggests that it is an important programming variable. However, the manipulation of duration has been achieved via noncompatible or indirect S-R mappings. Consequently, there is the potential that translation and programming effects are confounded and that inferences about response programming are in error. If spatially compatible S-R mappings cannot be achieved with nonspatial parameters of movement, how might the confounding issue be resolved? At least three solutions to this problem seem possible. First, use simple-RT methodology. If the complete response to-be-produced is known in advance, then translation processes can be completed prior to stimulus onset. Hence, they would not contribute to RT. It must be acknowledged, however, that some (e.g., Klapp 1978) have argued simple-RT methods are not suitable for studying response programming phenomena (but also see Henry (1980) and Sternberg et al. (1978) for counterarguments). A second approach might be to study individuals who are highly trained and skilled in the chosen experimental task (see Zelaznik and Hahn 1985). Extreme amounts of practice can substantially reduce the contribution of the translation to RT (Larish 1980) and perhaps it is possible to turn a noncompatible mapping into a compatible one. This tack also has its liabilities though. It is difficult to know a priori how much practice will be necessary and sufficient. Further, this suggestion may be feasible for relatively simple types of translations but not for translations that are complex, and this option may be necessarily limited to small numbers of S-R pairs. Third, rather than attempting to reduce or eliminate the prevalence of S-R translations, perhaps it would be more worthwhile to examine the interaction between translation and programming processes. The results from McCracken (1979) and Spijkers and Steyvers (1984) illustrate this point. Both studies used a ‘mixed’ version of the precue method to examine the programming of duration and direction. That is,
68
D.D.
Larish
/ Stimulus-response
translations
the programming of duration involved an S-R translation, whereas, the programming of direction did not. Each of these studies reported findings which suggest that there was some temporal overlap in the processing of duration and direction. Since these experiments were not intended to examine this specific issue, it is difficult to ascertain the exact nature of the temporal overlap (parallel processing). But, two explanations seem plausible: (a) the programming of the compatible parameter was completed during the translation of the noncompatible parameter, or (b) the translation for the noncompatible parameter was completed first and then the programming of both parameters was done in parallel. Issues such as these have not received much attention within the programming literature. However, the mixed version of the precue method may offer one way to identify more precisely the exact relationship between translation and programming ~processes. Miller (1983) has developed an experimental protocol, based in part on the precue method, that may also be useful in this regard. In summary, the movement precue method is a viable technique for examining the motoric decisions associated with response programming. It must be kept in mind, however, that the most accurate estimates of programming time are likely to be obtained when the mapping among stimuli and responses is spatially compatible. At the same time, it would be a mistake to ignore the potential interaction between translation and programming processes. In certain situations these processes may have special relationships that, if discovered, would not only advance our understanding of response programming but would also contribute to our understanding of information processing models in general. References Bauer, D.W. and J. Miller, 1982. Stimulus-response compatibility and the motor system. Quarterly Journal of Experimental Psychology 34A, 367-380. Brainard, R.W., T.S. Irby, P.M. Fitts and E.A. Alluisi, 1962. Some variables influencing the rate of gain of information. Journal of Experimental Psychology 56, 105-110. Broadbent, D.E. and M. Gregory, 1965. On the interaction of S-R compatibility with other variables affecting reaction time. British Journal of Psychology 56, 61-67. Fitts, P.M. and C.M. Seeger, 1953. S-R compatibility: spatial characteristics of stimulus and response codes. Journal of Experimental Psychology 46, 199-210. Fitts, P.M., J.R. Peterson and G. Wolpe, 1963. Cognitive aspects of information processing: II. Adjustments to stimulus redundancy. Journal of Experimental Psychology 65,423-432.
D. D. Larish
/ Stimulus-response
translations
69
Goodman, D. and J.A.S. Kelso, 1980. Are movements prepared in parts? Not under compatible (naturalized) conditions. Journal of Experimental Psychology: G.eneral 109, 475-495. Greenwald, A.G., 1972. On doing two things at once: time sharing as a function of ideomotor compatibility. Journal of Experimental Psychology 94, 52-57. Henry, F.M., 1980. Use of simple reaction time in motor programming studies: a reply to Klapp, Wyatt, and Lingo. Journal of Motor Behavior 12, 163-168. Kerr, B.T., 1976. Decisions about movement direction and extent. Journal of Human Movement Studies 3, 199-213. Kerr, B.T., 1978. ‘Evaluating task factors that influence selection and preparation for voluntary movements’. In: G.E. Stelmach (ed.), Information processing in motor control and learning. New York: Academic Press. pp. 55-69. Klapp, S.T., 1977. Response programming, as assessed by reaction time, does not establish commands for particular muscles. Journal of Motor Behavior 9, 301-312. Klapp, S.T., 1978. ‘Reaction time analysis of programmed control’. In: R. Hutton (ed.), Exercise and sports sciences reviews. Santa Barbara, CA: Journal Affiliates. pp. 231-253. Klapp, S.T. and C.I. Erwin, 1976. Relation between programming time and duration of the response being programmed. Journal of Experimental Psychology: Human Perception and Performance 2, 591-598. Klapp, S.T., E.P. Wyatt, and W.M. Lingo, 1974. Response programming in simple and choice reactions. Journal of Motor Behavior 6, 263-271. Larish, D.D., 1980. On the relationship between response organization processes and response programming. Unpublished doctoral dissertation, University of Wisconsin at Madison. Larish, D.D. and G.A. Frekany, 1985. Planning and preparing expected and unexpected movements: reexamining the relationship of arm, direction, and extent of movement. Journal of Motor Behavior 17, 168-189. Leonard, J.A., 1958. Partial advance information in a choice reaction time task. British Journal of Psychology 49, 89-96. McCracken, H.D., 1979. Programming direction, extent, and duration in aimed hand movements. Unpublished doctoral dissertation, University of Wisconsin at Madison. Massaro, D.W., 1975. Experimental psychology and information processing. Chicago, IL: Rand McNally College Publishing Company. Miller, J., 1983. Can response preparation begin before stimulus recognition? Journal of Experimental Psychology: Human Perception and Performance 9, 161-182. Rosenbaum, D.A., 1980. Human movement initiation: specification of arm, direction and extent. Journal of Experimental Psychology: General 109, 444-474. Rosenbaum, D.A. and S. Kornblum, 1982. A priming method for investigating the selection of motor responses. Acta Psychologica 51, 223-243. Sanders, A.F., 1980. ‘Stage analysis of reaction time’. In: G.E. Stelmach and J. Requin (eds.), Tutorials in motor behavior. New York: North Holland Publishing Co. pp. 331-354. Simon, J.R. and J.L. Craft, 1970. Effects of altered display-control relationships on information processing from a visual display. Journal of Applied Psychology 54, 253-257. Simon, J.R. and J.D. Wolf, 1963. Choice reaction time as a function of angular stimulus-response correspondence and age. Ergonomics 6, 99-105. Spijkers, W.A.C. and F.J.J.M. Steyvers, 1984. Specification of direction and duration during programming of discrete sliding movements. Psychological Research 46, 59-71. Sternberg, S., 1969. The discovery of processing stages: extension of Donders method. Acta Psychologica 30, 276-315. Sternberg, S., S. Monsell, R.L. Knoll and C.E. Wright, 1978. The latency and duration of rapid movement sequences: comparisons of speech and typewriting’. In: G.E. Stelmach (ed.), Information processing in motor control and learning. New York: Academic Press. pp. 118-152.
70
D.D.
L.arish
/ Stimulus-response
translations
Teichner, W.H. and M.J. Krebs, 1974. Laws of visual choice reaction time. Psychological Review 81, 75-98. Theios, J., 1975. ‘The components in response latency in simple human information processing tasks’. In: P.M.A. Rabbitt and S. Dornic (eds.), Attention and performance V. New York: Academic Press. pp. 418-440. Zelaznik, H.N. and R. Hahn, 1985. Reaction time methods in the studying of motor programming: the precuing of hand, digit, and duration. Journal of Motor Behavior 17, 190-218. Zelaznik, H.N., D. Shapiro and M.C. Carter, 1982. The specification of digit and duration during motor programming: a new method of precuing. Journal of Motor Behavior 14, 57-68.
