Acta Psychologica North-Holland
54 (1983) 51-67
RESPONSE
CHARACTERISTICS
RESTRUCTURED
GEORGE
HANDWRITING
51
OF PREPARED
AND
*
E. STELMACH
Uniuersrty of Wisconsrn, USA
Hans-Leo
TEULINGS
Uniuersity of Nijmegen,
The Netherlands
Handwriting responses were investigated under advance planning and movement restructuring conditions to examine if individual stroke characteristics were affected when a handwriting response had to be restructured prior to initiation. A movement priming paradigm was employed that created a strong bias for the execution of one handwriting pattern compared to that of another through expectancy manipulations. Two experiments are reported which contrasted the production of a well-prepared handwriting pattern to that of an unprepared version of the same pattern. In Experiment 1, subjects prepared and executed writing patterns that were spatially and temporally different, while in Experiment 2, the preparation was spatially identical, but required a change in the force-time relationship. Reaction time and the movement execution characteristics (i.e., duration, maximum velocity, and length per stroke) were determined and compared. The data revealed that most of the advance planning benefits were localized in initiation processes with some evidence for “on line” programming. As the ratios between writing strokes were well preserved, comparisons of writing size as a function of preparation generally supported space-time invariance.
Handwriting ‘is a complicated but well-trained psychomotor skill that involves the coordination of orthogonal muscle systems of the forearm, hand and fingers with intricate timing relationships. Therefore, handwriting is an interesting skill to study for gaining insights into how complex movements are structured prior to their execution and to what extent a well-prepared movement pattern still can be restructured. Over
* This research was conducted at the University of Nijmegen and was partially sponsored by the Fulbright Senior Scholar program and by the Netherlands Organization for the Advancement of Pure Research (Z.W.O.). Mailing address: G.E. Stelmach, Motor Behavior Lab., University of Wisconsin, 2000 Observatory Drive, Madison, WI 53706, USA.
OOOl-6918/83/$3.00
0 1983, Elsevier Science Publishers
B.V. (North-Holland)
the past decade there has been considerable interest in how control can pass from some abstract, nonmotoric level to a muscle useable code (see Stelmach and Diggles 1982, for a review). The most often invoked term used to explain this aspect of action specification is that of a motor program. The original conception of an abstract motor program (Henry and Rogers 1960) was in large part the result of motor skill researchers extracting data and conceptions almost simultaneously from neurophysiology and information processing psychology, as evidenced by Lashley’s (1951) finding that sensory processes play a minor part in regulating the intensity and duration of nervous discharge, and that the initial conditions prior to movement are not essential for accurate movement reproduction. Such data led directly to the idea of central representation of action and to the problems of serial order, i.e., how many preplanned units of action could be generalized in an appropriate sequence, how, in error, the sequences are not executed in correct serial order, and how the constituent components of action are put together. From the early, restrictive definition associated with Henry and Rogers (1960) and Keele (1968) the idea of a motor program has changed considerably over the years so as today a motor program is thought of as an abstract non-muscle specific representation of a motor act (Van Galen 1980). One of the primary reasons for a change in definition comes from the widely observed phenomena from handwriting that when one signs their name on a check, blackboard or with a constrained joint, even though the musculature involved is different, as are the movement directions and force levels generated by the muscles, the writing patterns are extremely similar (Merton 1972; Raibert 1977). One of the most fundamental concerns in the motor control area has been to understand how a motor program is actually constructed. To this end, one of the most widely adopted experimental approaches has been to vary the nature of the response to be executed by manipulating the movement parameters (e.g., arm, direction, extent, velocity) that are likely to be contained in a motor program (Glencross 1973: Goodman and Kelso 1980; Kerr 1976; Klapp 1975; Klapp and Erwin 1976; Megaw 1972; Rosenbaum 1980). Findings pertaining to the structure of a ‘prepared’ motor program have indicated that programming is not a unitary process (Klapp 1979), that some combinations of parameters can be specified in a serial or parallel manner (Rosenbaum 1980; Klapp 1979; Megaw 1972) that certain parameters have a fixed order or
G. E. Stelmach,
H. L. Teulings/Response
characteristics
of handwrrting
53
organization (Rosenbaum 1980) and as the complexity of a response increases, the apparent programming time also increases (Henry and Rogers 1960; Sternberg et al. 1978). Recently, Larish and Stelmach (1983) utilized a modification of the popular movement priming technique (developed by LaBerge et al. 1970), to study restructuring aspects of motor programming processes. This approach consisted of creating a bias for executing one response by creating a high expectancy level and then occasionally requiring the subjects to perform an alternative response. The underlying rationale of this experiment was that, through expectancy manipulations (80, 50, 20%), differential preparation strategies could be induced that would allow inspection of initiation and execution processes. Using this methodology and employing a discrete arm response varying in complexity, the results from the Larish and Stelmach study found considerable differences between advance planning and the restructuring aspects of motor programming. The experiments reported in this article utilized the methodology described by Larish and Stelmach (1983) but only created two differential preparation states prior to the production of a writing pattern to examine if well-prepared writing patterns were performed exactly like restructured patterns. One of the interesting aspects of handwriting, and a reason that this type of task was chosen for these experiments, is that the movement characteristics during execution are free to vary. The research from Nijmegen has clearly shown that handwriting movements often considered to be continuous for letter and words can be decomposed into individual strokes that allow dynamic description of size, velocity and duration parameters of the writing trace to help understand the control of motor output (Hulstijn and van Galen 1983; Teulings and Thomassen 1979; Teulings et al. 1983). As previous programming studies either controlled movement directions and/or constrained execution dimensions (Glencross 1973; Kerr 1976; Klapp 1979; Larish and Stelmach 1983; Rosenbaum 1980; etc.) little is known on whether prepared response sequences exhibit differential movement characteristics compared to unprepared ones. In other words, are programming processes as indexed by reaction time (RT) localized entirely in initiation processes or is there some overlap into execution processes? While RT was used to index the programming operations, the execution characteristics were segmented via the vertical velocity component and categorized with respect to duration, maximum
G. E. Stelmcrch, H.-L.
54
Teulrngs/Re.~p~n.re
churcrctrmtics
of humhvritmg
velocity, and length parameters per stroke. Thus, comparisons in RT and stroke characteristics were made between well prepared writing patterns (80% trials) and those that had to be restructured due to an in-appropriate preparation (20% trials). Experiment 1 contrasted the production of a prepared writing pattern to one which required complete movement restructuring while experiment 2 compared different programming states and only varied the size prescription of the writing pattern. Since only force-time parameters are required for changes in writing size, experiment 2 will also inspect the space-time phenomena under advance planning and parameter restructuring conditions. One of the theoretical issues explored in this experiment is whether changes in writing size are limited to initial strokes of the writing pattern. If learned handwriting movements are represented in some abstract code and executed as a whole, it would be expected that in the restructuring situation the subject may have some difficulty to specify the correct movement parameters.
Experiment
1
Method
Subjects Ss
were
University
13 right-handed male and female psychology students from the Katholic of Nijmegen. The) were either paid or given class credit for participation.
Appuratus
The writing movements were recorded by a computer controlled digitizer (Vector General Data Tablet DTI). The position of the tip of the electronic pen, expressed in their horizontal and vertical coordinates with a combined RMS error better than 0.2 mm was sampled at a rate of 200 Hz. The pen tip was an ordinary ballpoint refill and the S wrote on a sheet of paper fixed on the digitizer surface. The digitizer was positioned such that the S’s individual writing slope was parallel to the horizontal axis of the digitizer. Direct vision of the writing hand was eliminated by the placement of a shield above the writing surface. A display (Vector General Graphics Display Series 3 Model 2D3) was positioned at a distance of 125 cm directly in front of the S at eye level and it allowed the tachistoscopic presentation of stimuli (the stimuli were built up within 1 msec). Procedure
Each trial began with a buzzer and after a 1 set delay, a writing stimulus was displayed for 100 msec. The two stimuli used for this experiment are shown in fig. 1.
G. E. Stelmach,
H. - L. Teulings/Response
churacteristrcs of handwrrting
55
The S’s task was to initiate a response to the writing stimulus as fast as possible. All pen movements were recorded and stored and one second later the recorded writing trace, the stimulus, and the reaction time were displayed to the S for 2 sec. Immediately after the feedback disappeared, the buzzer for the next trial sounded. The fed-back writing trace served also as a means to keep the writing size constant. For that purpose, the bodies of the letters were fitted between two horizontal lines representing a writing size of 3 mm. These lines were also useful to inform the Ss that they were still writing horizontally and correctly. The Ss were familiarized with the writing task and then given sufficient practice (100 trials of each allograph) until they had no difficulty with them. Then each S participated in four blocks of 55 trials where the writing pattern displayed and to be written altered between bye and ynl. Each block was separately randomized and there were two restrictions in the randomization procedure. The first five trials were exclusively the most probable stimulus and no two successive catch trials occurred. Only the last two blocks were used in the analysis. The whole experimental session took one hour. As an aid to preparation, the highly probable pattern was continually visible to the S during a block of trials and when the S heard the buzzer (1 set prior to the imperative stimulus) they were to prepare the displayed writing response and write it as fast as possible when it was flashed on the display. The distinguishing feature between blocks was the designation of which pattern appeared most frequently (80%). Within a block of trials, 40 trials consisted of the flashed pattern matching the continually visible pattern, 10 where it did not and 5 that were catch trials (no imperative signal). Depending on the block, the Ss participated in sessions where they either executed ynl 80 percent of the time but on 20 percent of the trials they had to switch to bye or they executed bye 80 percent of the time but had to switch to ynl on 20 percent of the trials. Thus, for each pattern there were well prepared executions (80% trials) and ones that were executed only after they were restructured (20% trials).
Fig. 1. Stimuli hye and ynl as they were presented performed.
on the display
screen and as they had to be
56
G. E. Stelmrrch, H.-L.
Teulings/Response
charmteristics
Fig. 2. Two responses hye. ynl together with their vertical the time marks. selected by the computer algorithm.
velocity
OJ hundwruing
patterns.
The dots represented
It was stressed that all up and down movements had to be made clearly, that at the start of a trial the pen had to be in contact with the paper, and that the pen should not be lifted during the writing movement. The only constraint on the writing task was that the first stroke of hye had to be in the upward direction and the first stroke of ynl had to be in the downward direction. Response
analysts
The vertical coordinate as a function of time was differentiated and filtered at 16 Hz yielding the vertical velocity and time marks were determined where the vertical velocity changed sign (i.e., a downward movement is passed into an upward movement or v.v.) (Teulings and Thomassen 1979). The writing patterns bye and .rnl were chosen such that the movement from one time mark to the following could be regarded as one stroke (Hollerbach 1980). An example of how the two patterns in this study were segmented by use of the vertical velocity time function is shown in fig. 2. In addition to the intervals between these time marks the maximum velocity and the net length of the strokes were also determined. During the analysis of each trial the recorded writing trace and the vertical velocity curve were displayed. The time marks, as calculated by a computer algorithm were made visible so that the analysis of each trial could be verified. If the algorithm did not work properly, the time marks could be readjusted by hand using a moving cursor controlled by a turning dial. This was done in about 16 percent of the non-expected and 6 percent of the expected responses. If the response obviously contained an execution error the trial was excluded from analysis. Results Errors
The error rate over all trials was 15 percent, with the most noticeable errors occurring in the first stroke (non-expected responses, 19%: expected responses, 5%) strokes omitted or added (non-expected responses, 2%; expected responses, 6%).
57
Table 1 Mean reaction time and standard deviations (in parentheses)
in msec for each of the allograph
patterns. Expectancy
level
0.80
0.20
be
318
433
Ynl
(50) 328
(60) 470
(57)
(113)
Reaction
time
Average RT over Ss for each pattern and probability level was determined and the differences between probability levels compared. The mean RTs for the four conditions are reported in table 1. As expected, the table reveals that for both allograph patterns, the RTs are considerably faster in the high expectancy condition (80%) compared to those of low expectancy. Both patterns were found to be initiated significantly faster in the two high expectancy conditions (318 versus 433 and 328 versus 470 msec, for bye and ynl respectively; sign test, N = 13, x = 0, p -C 0.01). By a subjects X pattern (bye versus yd) X expectancy (20% versus SOW) analysis of variance on the means over
I
I
I
I
EXPERIf-lENT
I
I
1
I
I
HYE
P=80%
I
I
I
I
I
I
I
I
1
2
3
4
5
6
7
8
STROKE Fig. 3. Movement for hye.
length
NUnBER
per stroke plotted
as a function
of stroke number
and probability
level
G. E. Stelmach,
58
H: L. Teulrn~s/Response
EXPERII’IENT
0
HYE
1
I’
I
I
I
I
I
I
I
1
2
3
4
5
6
7
8
STROKE Fig. 4. Maximum for hp.
characterrstrcs of handwttrng
velocity
NUIIBER
per stroke plotted
as a function
of stroke number
and probability
level
replications per block, we also found a significant interaction between the factors writing pattern and expectancy condition (F(1, 12) = 58.6. p c 0.01). The expectancy effect is apparently stronger for ynl. Response characteristics The response characteristics (duration per stroke, maximum velocity per stroke, length per stroke) of the writing patterns for bye and ynl were compared between advance planning and movement structuring conditions (20% and 80%. respectively). Generally, no significant differences were found between both conditions (sign test, two-tailed, N = 13, x > 3, p > 0.1). An exception to the foregoing was found in the production of the length and the maximum velocity of the first stroke of bye. In the unexpected condition, the observed values are less than in the expected conditon (1.14 versus 1.23 cm and 9.9 versus 11.0 cm/set; sign test, two-tailed, N = 13, x = 2, p -C 0.05) (see figs. 3 and 4). This suggests that the preparation for the first stroke of ynl caused the first stroke of bye to be produced smaller but within the same interval of time.
Experiment
2
This experiment used identical methodology as that employed in experiment 1 with the exception that the stimulus and the response set changed. One writing pattern was
G.E. Srelmach, H. - L. Teulmgs/Response characteristrcsof handwriirng
59
presented (bye) and the Ss were required to write it in one of the two sizes (large or small). As only size was varied between probability conditions, the main difference between 80 percent and 20 percent conditions was the restructuring or readjustment of force-time parameters required for proper execution. Method Subjects Ss were 14 right-handed male and female psychology population as the Ss in experiment 1. Apparatus and response analysis The stimulus and the movement recording apparatus 1 as well as the analysis of the recorded trials.
students
drawn from the same
were the same as in experiment
Procedure The bye pattern in experiment 1 was used (see fig. 2), however, the displayed stimuli appeared in one of two sizes. As an aid to preparation for the correct size, the fed back trace presented after execution contained two horizontal lines the distance of which approximated the actual writing size required. If the stimulus was of a large size the distance between the feedback lines was 3 mm; if the stimulus had a small size the distance was 1.5 mm. All other experimental procedures described in experiment 1 were identical. Analogously to experiment 1, the distinguishing feature between blocks was the designation of which size (large/small) pattern appeared most frequently (80%) within a block of trials. There were 40 trials where the displayed size matched the most frequent size, 10 where it did not and 5 that were catch trials. Depending on the block, the Ss participated in blocks where they prepared and executed hye as a large size (3 mm) but on 20 percent of the trials they had to write the same pattern in half the size (1.5 mm) or prepared and executed the small size (80%) and had to occasionally switch to a large version at the time of the imperative signal. Results Errors The error rate over all trials was 11 percent with four classes of errors accounting for most of them. The first stroke was incorrect (non-expected 4% expected 3%); strokes were omitted or added (non-expected 68, expected 4%); there was a pause during the second or following strokes (non-expected 4%, expected 1%) or the size was incorrect (non-expected 2%, expected 2%). Further, as a check on experimental procedure it was necessary to determine writing sizes in both conditions. The average height of the small letters for the large condition was 5.0 mm and for the small condition it was 2.7 mm. Reaction time The average RTs per S within one size and expectancy condition were determined and compared, and the mean reaction times for the four conditions are reported in
60
Table 2 Mean reaction
G. E. Stelmach,
H.-L.
time and standard
Teulings/Response
deviations
(in parentheses)
Expectancy 0.80 hye -large
301 (87)
hye -small
317 (90)
chaructenstics
of handwriting
in msec for bye written large or small.
level 0.20 362 (132) 383 (170)
table 2. By a subjects X size (large versus small) x expectancy (20% versus 80%) analysis of variance on the means over replications per block we found that RTs were faster in the high expectancy conditions (80%) compared to those in the low expectancy conditions (F(1, 13) = 6.16, p < 0.05). Also the interaction between size and expectancy condition appeared to be significant (F(1, 13) = 5.28, p c 0.05). However, it still holds that both handwriting sizes were initiated faster in the advance planning conditions (301 versus 362 and 317 versus 383 msec, for large and small sizes respectively; sign test, one-tailed, N = 14, x = 4, p < 0.1). The main effect due to size was non-significant (Fc 1). In another analysis of variance, combining the RT data of both experiment 1 and 2 (thus, experiment X subjects X allograph X expectancy), we found a significant interaction between experiment and expectancy condition (F(1, 25) = 4.32, p -C 0.05), which suggests that the expectancy effect is in fact much smaller when only a different size has to be specified than when a different pattern has to be specified. Response characteristics The average length per stroke for the production of the two allograph sizes can be seen in table 3. This table reveals that the execution of the two sizes produces substantially different lengths per stroke, as required. However, producing an expected versus a non-expected size of an allograph had no effect on stroke length (sign test. two-tailed, N = 14, x = 3, p > 0.05). Maximum velocities are also reported in table 3 and it is apparent that with an increase in length per stroke there is a proportional increase in maximum velocity across the eight strokes. Again there is, in general, no difference in maximum velocity per stroke between expected and non-expected allographs (sign test, two-tailed, N = 14, x = 3, p > 0.05). Exceptions are stroke 4 and, more importantly, stroke 1 of the large responses. When the Ss executed the large size, they performed the first stroke faster when they prepared the correct, large size than when they prepared the small size (sign test, two-tailed, N = 14, x = 1, p < 0.01). While the observed stroke parameters for length and velocity are along the lines suggested by the space-time invariance (Viviani and Terzuolo 1980) the data on durations per stroke reveal a more complicated picture. Although the duration of only a few strokes appears to be affected by the level of expectancy the most striking expectancy effect is demonstrated in a different way. In fig. 5 we plotted the durations per stroke for small versus large size in the 80% condition, in which the S executed the
hye-small80%
hye-large 80% hye-small80%
hye-small20%
14.52 7.58
9.05
12.16
0.97 1.62
0.84
hye-small2051; hye-large 80%
Maximum velocity (cm/xc) hye-large 20%
1.47
1
_
Stroke number
9.17 4.47.
20.21 9.97
_-
4.45
0.34
8.14
0.34 0.66
0.81 1.58 0.74
10.41
0.61
1.53
3
18.71
2
velocities per stroke for the hye allograph
Length (cm) hye-large 20%
Mean lengths and maximum
Table 3
7.76 4.08
3.97
6.88
0.23
0.23 0.45
0.42
4
written
7.28 3.64
3.78
6.55
0.31
0.31 0.58
0.56
5
5.83 3.01
6.10
5.57
0.16
0.16 0.31
0.31
6
small and large for each probability
2 d!$ 5 4 3” 2 a% 2a 3. 2, a 9
0.68 1.21 0.70
12.97 8.07 13.92 8.08
0.32 0.58 0.30
6.12 3.58 6.61 3.33
:& 1.21
8
!G % 3 f: .a %
0.55
7
level.
G. E. Stelmach,
62
H.. L. Teulings/Response
charactermtics
of hundwritmg
200 EXPERIRENT
2
HYE
(P=80%>
c 2 11
150
W Y i? L
100
: 5 Iz
50
Q
0
I
\
I
I
I
I
I
I
‘1
2
3
4
5
6
7
8
STROKE
NUIIBER
Fig. 5. Duration per stroke for hye-large and bye-small in the 80% condition plotted as a function of stroke number.
I
I
I
I
EXPERlllENT
1
2
3
I
I
2
4
STROKE
WE
5
I
I
1
6
7
8
NWIBER
Fig. 6. Duration per stroke for hye-large and hye-small in the 20% condition plotted as a function of stroke number.
G. E. Stelmoch,
H.-L.
Teulings/Response
charactenstics
of handwriting
63
prepared size. Comparisons by means of sign tests (two-tailed, N = 14) on the durations per stroke revealed that there were no differences between large and small sizes (x = 2, p < 0.05, except for stroke 3) supporting the well-known time constancy invariance (e.g. Merton 1972). In fig. 6 we plotted the durations per stroke for the 20% condition as a function of large and small sizes. In contrast to fig. 5, there are substantial differences due to writing size across all eight strokes plotted (sign test, two-tailed. N = 14, x = 2, p < 0.05). The interaction between size and expectancy condition is supported by an analysis of variance on the duration data of each stroke separately (F(1, 13) = 6.99, p < 0.05). This suggests that in the 20% condition the force-time phasing is not normally specified. Yet, it can be seen by the shape of the curve that the ratios between durations of the individual strokes are invariant. This implies that it is more difficult to increase the force when an increase in size is required when a small size has been prepared, than to decrease the force level when a decrease in size is required. Remarkably, this duration effect remains for all eight strokes suggesting that there is an overall force-time specification that modulates the entire allograph.
Discussion The principle intent of the present experiments was to determine whether differences in the state of preparation are associated with changes in movement production characteristics. Two experiments were performed where subjects were preparing continually a writing pattern and occasionally they were required to switch to another pattern. This experimental procedure created opportunities to inspect handwriting characteristics (duration, maximum velocity, and length per stroke) under advance planning and movement restructuring conditions. In both experiments, RT was used to assess differential preparation states and, as expected, in experiment 1 the average difference between preparation states was a substantial 128 msec. For this experiment, the 20% condition required a complete restructuring of the prepared response prior to initiation. In contrast, the average difference for experiment 2 was only 63 msec which is considerably smaller than observed in experiment 1. The smaller difference was probably observed because movement restructuring was restricted to changes in only the size of the handwriting response produced. These findings are similar to other programming data that demonstrated that force (distance) specification in a motor program does not take as much time as direction or limb specification (Bonnet et al. 1982; Megaw 1972; Rosenbaum 1980). Further, the RT data which demonstrated the benefits of motor program preparation extend the usefulness of the priming paradigm as a
64
G. E. Stelmoch,
H.-L.
Teulrngs/Response
choracterrstrcs
of handwrirrng
method to create different programming states (Larish and Stelmach 1982) while avoiding most of the criticisms leveled against the precuing technique (Rosenbaum 1980; Zelazink 1978). Despite the substantial reductions in RT for prepared handwriting responses in both experiments, examination of the execution characteristics, such as movement length, duration, or maximum velocity per stroke, revealed that they are highly independent of preprogramming. One interpretation of this finding is that this is exactly what one would expect to find as altering preparation states does not effect execution processes. That is, the benefits of advance planning (preprogramming) are limited to the initiation period between the stimulus and response. As mentioned earlier, there are limited data on this issue as most programming studies have constrained the movements executed (Glencross 1973; Kerr 1977b; Klapp 1979; etc.). At issue in this interpretation is how much of a movement sequence is programmed in advance. There are many who would argue that it is only the first 150 msec of movement that is preprogrammed (Keele and Posner 1968), as this is the minimal time required to process sensory feedback. But, neurophysiological data suggest, however, that the minimal feedback processing time may be even considerably less (Evarts 1971; Allen and Tsukahara 1974). Therefore, one could argue that advance planning of a response implies only preprogramming of the first one or two strokes, taking about 150 msec, since the subsequent strokes could be programmed during the execution of the response. Partial support for this position was found in experiment 1 where it was observed that for the 20 percent condition a significantly slower and shorter first stroke was found in the production of bye. Inspection of this difference can be observed in figs. 3 and 4 where it can be seen that in the unexpected condition the first stroke was significantly shorter than in the expected condition. Similarly, in the production of the first stroke of ynl a comparable difference was observed although non-significant. As seen in the significant differences observed in velocity and length parameters in table 3, the type of preparation within a given size affects the execution of the first stroke of an allograph pattern. This indicates that not all effects of programming are indexed by the reaction time. The approximate cost of reparameterizing the size for both writing sizes was 60 msec. Apparently, this additional time was not sufficient to get the force-time relationship organized so that the sequence could be
G. E. Stelmach,
H. - L. Teulings/Kesponse
characterrstics of handwiting
65
normally executed. An examination of advance planning effects as seen by comparing length and velocity parameters within a handwriting size reveals that the observed preparation effects are limited to the first stroke. One interpretation of this part of the data is that the motor programming in these handwriting tasks is limited to the first stroke with the remainder of the programming done “on line”. This observation is similar to that proposed by Hulstijn and van Galen (1983) who found that inter digit intervals appear not to be influenced by the sequence length, thereby reflecting a local programming process for the execution of each individual digit. Interpreting our data as such is somewhat at odds with Sternberg et al. (1978) who proposed that an entire sequence of words (speech) or characters (typewriting) are being programmed prior to the initiation of the first response and that a global search process is necessary for the execution of each individual response in the sequence. A slightly different picture of motor programming is obtained when one compares large versus small writing sizes at each of the expectancy conditions (80% vs 20%). It was found that under the reparameterization condition, that is, when a movement sequence is prepared under a specific force-time relationship and this preparation has to be restructured, the original preparation interferes with the new one. The surprising aspect of this finding is that the lack of adequately reparameterizing the durations in the handwriting sequence for the 20% condition remains throughout the execution of the allograph and the timing ratios of successive elements are preserved. As for the debate over the nature of motor programming and the type of units involved, these data suggest that motor programming for handwriting may be characterized as a series of horizontal and vertical movements which are modulated by some underlying timing component involving the entire sequence. A few years ago, Wing (1980) came to essentially the same conclusion from his work on durations between stroke segments and suggested that the basic program unit being timed was not a single stroke but an underlying metronomic process. The question of whether programming states affect movement execution characteristics is an important issue to be resolved. The experiments reported here used novel methodology in an attempt to shed some light on this issue by varying advance planning (preprogramming) and movement restructuring requirements in a handwriting task. The data obtained did reveal several differences in response characteristics
between preparation states as seen in the observed context effects found in the initial stroke. These context effects suggest that there is an overlap between initiation and execution processes that warrants further study.
References Allen, G.I. and W. Tsukahara, 1974. Cerebrocerebellar communication systems. Physiological Review 54. 957-1006. Bonnet, M.. J. Requin and G.E. Stelmach, 1982. Specificatmn of duration and extent in motor programming. Bulletin of the Psychonomic Society 19. 31-34. Evarts, E.V.. 1971. Activity of thalamic and cortical neurons in relation to learning movement in the monkey. International Journal of Neurology 8. 321-326. Glencross, D.J.. 1973. Response complexity and the latency of different movement patterns. Journal of Motor Behavior 5. 95-104. Goodman, D. and J.A.S. Kelso. 1980. Are movements prepared in parta? Not under compatible (naturalized) conditions. Journal of Experimental Psychology: General 109. 475-495. Henry. F.M. and D.E. Rogers, 1960. Increased response latency For complicated movements and a “memory drum” theory of neuromotor reactwn. Research Quarterly 31. 44X-458. Hollerbach, J.M.. 1981. An oscillatory theory of handwriting. Biological Cybernetics 39. 139-156. Hulstijn, W. and G.P. van Galen. 1983. Programming in handwriting: reaction time and movement time as a function of sequence length. Acta Psychologica 54, 23-49. Keele, SW.. 1968. Movement control in skilled motor performances. Psychological Bulletin 70. 387-403. Keele, SW. and M.I. Posner. 1968. Processing of visual feedback in rapid movements. Journal of Experimental Psychology 70, 155- 158. Kerr. B.T., 1976. Decisions about movement direction and extent. Journal of Human Movement Studies 3. 199-213. Klapp, S.T., 1975. Feedback versus motor programming in the control of aimed movements. Journal of Experimental Psychology: Human Perception and Performance 1. 147-153. Klapp, S.T.. 1979. Response programming as assessed by reaction time. does not establish commands for particular muscles. Journal of Motor Behavior 9. 301-312. Klapp, ST. and G.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. LaBerge, D.H., P. van Gelder and J. Yellott, 1970. A cuing technique in choice reaction time. Perception and Psychophysics 8. 51-62. Larish, D.D. and G.E. Stelmach. 1983. Preprogramming, programmmg and reprogramming of aimed hand as a Function of age. Journal of Motor Behavior 14. 322-340. Lashley. K.S., 1951. ‘The problem of serial order in behavior’. In: L.A. Jeffress (ed.), Cerebral mechanisms in behavior: the Hixon symposium. New York: Wiley. Merton. P.A., 1972. How we control the contraction of our muscles. Scientific American 226 (5). 30-37. Megaw, E.D., 1972. Direction and extent uncertainty in step-input tracking. Journal of Motor Behavior 4. 171-186. Raibert, M.H., 1977. Motor control and learning by the state-space model. Techmcal report, Artificial Intelligence Laboratory Al-TR-439. Cambridge. MA: MIT.
G. E, Selmach,
H. L. Teulings/Response
characterisrrcs of handwriting
61
Rosenbaum, D.A., 1980. Human movement initiation: specification of arm, direction and extent. Journal of Experimental Psychology: General 109, 444-474. Stelmach, G.E. and V. Diggles, 1982. Control theories in motor behavior. Acta Psychologica 50, 83-105. Sternberg, S., S. Monsell, R. Knoll and C. 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. Teulings, H.L. and A.J.W.M. Thomassen, 1979. Computer aided analysis of handwriting. Visible Language 13, 2188231. Teulings, H.L., A.J.W.M. Thomassen and G.P. Van Galen, 1983. Preparation of partly precued handwriting movements: the size of movement units in writing. Acta Psychologica 54, 165-177. Van Galen, G.P., 1980. ‘Handwriting and drawing: a two stage model of complex motor behavior’. In: G.E. Stelmach and J. Requin (eds.), Tutorials in motor behavior. Amsterdam: North-Holland. pp. 567-578. Viviani, P. and V. Terzuolo, 1980. ‘Space-time invariance in learned motor skills’. In: G.E. Stelmach and J. Requin (eds.), Tutorials in motor behavior. Amsterdam: North-Holland. pp. 525-539. Wing, A.M., 1980. The height of handwriting. Acta Psychologica 46, 141-151. Zelazink, H.N., 1978. Precuing response factors in choice reaction time: a word of caution. Journal of Motor Behavior 10, 77-79.
