Psychological Research (2003) 67: 291–302 DOI 10.1007/s00426-002-0129-y O R I GI N A L A R T IC L E Lucette Gullaud-Toussaint Æ Annie Vinter The effect of discordant sensory information in graphic production: two distinct subject groups Received: 23 November 2000 / Accepted: 13 November 2002 / Published online: 19 February 2003 Springer-Verlag 2003 Abstract The present study investigated the underlying processes used to cope with discordant sensory information induced in a mirror-drawing task. Two experiments were carried out in which adults copied simple geometrical figures made up of either horizontal and vertical segments or oblique segments meeting at a right angle in both a normal and a mirror condition. Experiment 1 identified individual differences in relation to preferred graphic movement directions; some subjects preserved the visual directions that occurred in normal drawing by reversing the direction of drawing movements (perceived-direction group), while others preserved normal drawing directions that produced reversed visual directions (performed-direction group). Experiment 2 was performed to elucidate whether these two distinct behaviors resulted from different strategies used to cope with visuo-proprioceptive discordances. The main results showed that preference for the perceived directions led to longer pauses, slower movement velocity, greater movement dysfluency, and greater spatial orientation accuracy. By contrast, longer reaction time and greater angular accuracy characterized performance in the performed-direction group. These results were interpreted as indicating that two distinct information-processing strategies can be used when resolving sensory discordance in graphic production. Keywords Mirror drawing Æ Feedback control Æ Feedforward control Æ Visuo-proprioceptive conflicts L. Gullaud-Toussaint (&) Laboratoire d’Analyse de la Performance Motrice Humaine, MSHS, 99 Avenue du Recteur Pineau, 86000 Poitiers, France E-mail: Lucette.Toussaint@mshs.univ-poitiers.fr A. Vinter Laboratoire d’Etude des Apprentissages et du Développement, CNRS/UMR 5022, Dijon, France Introduction Are visual information and proprioceptive information both used in graphic production? Van Doorn and Keuss (Van Doorn, 1992; Van Doorn & Keuss, 1992, 1993) suggested that subjects are able to compensate for the absence of visual information in handwriting by an increased contribution of proprioceptive processing, as revealed by the increase in axial pen force, movement time, and writing size. Similarly, other experiments have shown that subjects compensate for the absence of proprioceptive cues through an increased contribution of visual processing. Teasdale, Forget, Bard, Paillard, Fleury, and Lamarre (1993) revealed that when vision was available in a graphic task, a deafferented patient and control subjects performed similarly, while without vision, the spatial control of movements was more strongly impaired in the patient than in the controls. Thus, the movements of the hand within the limits of the graphic space depend on afferent information provided by the visual as well as by proprioceptive systems. This raises the question of the respective role of each system in drawing movement production. Some authors have reported that visual and proprioceptive information may control different aspects of movement execution. For example, Meulenbroek and Thomassen (1991) suggested that two independent spatial reference systems were involved in straight-line production. One system, designated as the anatomical reference system, corresponded to the anatomical structure of the effectors (the hand or the fingers), and resulted in the preferential production of oblique lines. The second system, called the geometrical reference system, resulted in a bias towards horizontal and vertical orientations. Meulenbroek and Thomassen (1991) have suggested that oblique orientations are strong candidates for predominantly motoric control with a reduced sensitivity to the availability of visual information, while horizontals and verticals are more likely to be under perceptual control, with a strong dependency on visual 292 information. The authors demonstrated that the prevalence of visual control depended on the type of segments to be drawn in a situation where visual and proprioceptive modalities were available. Other studies revealed that the role played by visual feedback on graphic accuracy concerns the topokinetic component of movements (i.e., the spatial orientation), while the morphokinetic component (i.e., the shape production) seems to be preserved whatever the visual exposure conditions (Broderick & Laszlo, 1987; Paillard, 1991; Van Doorn & Keuss, 1993). In a previous experiment devoted to the study of sensory information in simple drawing production (geometrical figures composed of two segments meeting at a right angle), Gullaud (1998) confirmed that the withdrawal of visual information did not modify the angular accuracy of drawings (morphokinetic component), although vision was necessary for calibration of the displacement of the pen (line orientations or topokinetic component) within the limits of the graphic space in the case of models composed of horizontals and verticals. With regard to oblique models, the author noted that, in the absence of vision, compensatory strategies were sufficient to ensure the spatial orientation accuracy of the drawings. These strategies were probably based on an increase in proprioceptive control, as suggested by the increased movement time and writing size (see also Van Doorn, 1992). These overall results suggest that visual feedback controls specific aspects of graphic movement execution. Vision seems to play a dominant role in calibrating the topokinetic component of drawing, especially when the figures are composed of horizontal and vertical segments (geometrical reference system, Meulenbroek & Thomassen, 1991; Gullaud, 1998). The role of proprioception in graphic movement execution is less known owing to methodological difficulties in isolating proprioceptive information. In order to elucidate the role of proprioception, the use of a discordant sensory information paradigm appears appropriate, because the resolution of visuo-proprioceptive conflicts implies the dominance of either the visual or the proprioceptive modality (Uhlarik & Canon, 1971; Redding & Wallace, 1988a, 1988b). Lajoie, Paillard, Teasdale, Bard, Fleury, Forget and Lamarre (1992) provided an illustration of visual information dominance when drawing under a mirror condition. Using a six-pointed star, these authors observed that a deafferented patient had no problem in copying the pattern, while controls encountered certain difficulties in producing fast and accurate movements. The authors suggested that the greater difficulty encountered by controls in the production of obliques as opposed to vertical segments revealed the dominance of vision, because resolving the conflict implied a multiple directional choice for obliques (combination of up, down, left, and right directions), whereas it implied a binary choice for verticals (reversal of the top-down and down-top directions). Before being able to reach a level of performance similar to the deafferented patient, the control participants had to learn how to use reversed visual feedback by recalibrating their proprioceptive map. Gullaud and Vinter (1996, 1998) examined the mirror-drawing production of simple open geometrical figures, which were visually less constraining than the closed figures used in the Lajoie et al. experiment (1992). They were interested by the syntactical level, which describes graphic behavior in terms of a system of rules (Van Sommers, 1984), i.e., starting rules (start at the topmost and/or leftmost point), progression rules (draw from top to bottom and/or from left to right), and a threading rule (draw without pen-lifts). These rules govern drawings and result from an interaction between biomechanical and visual constraints (Thomassen, Tibosch & Maarse, 1989; Vinter & Meulenbroek, 1993). For instance, going from left to right induces less biomechanical constraints (flexion movements) and optimal visual control. In some situations, subjects have to choose between biomechanical facilities and visually guided movement, as was the case in Gullaud and Vinter’s experiments (1996, 1998), where the mirror-drawing condition induced left to right and right to left reversals. The directional analysis of graphic movement showed that two subject groups could be distinguished. Some subjects seemed to favor vision (the perceived-direction group): they visually preserved the perceived movement directions as they occurred in the normal condition, which thus induced a reversal of the directions drawn on the sheet of paper. In contrast, other subjects seemed to favor biomechanical constraints (the performed-direction group): they tended to preserve the directions drawn on the sheet of paper as they did in the normal condition, this strategy provoking a reversal of the directions perceived in the mirror. Thus, the current literature suggests that visual dominance may vary as a function of the geometrical figure to be drawn (horizontal-vertical versus oblique figures) as well as of the type of subjects (perceivedversus performed-direction group). Using a directional analysis, Experiment 1 aimed to show that two groups of subjects effectively emerge in a mirror-drawing task. The Gullaud and Vinter experiment (1996) included too few subjects in each group to provide substantial information. A more systematic study performed with a larger sample of participants was needed. As previously observed, we expected that most of the subjects would visually preserve the movement directions and movement sequences of mirror-drawing production as they do under the normal condition (perceived-direction group), while directions and movement sequences should be preserved and not be strongly dependent on the presence of the mirror for the other subjects (performed-direction group). The analysis of errors in mirror production seemed appropriate in the present experiment. The great difficulties encountered by the perceived-direction group when drawing oblique segments in the mirror condition could reveal the dominance of vision (Lajoie, Paillard, Teasdale, Bard, Fleury, Forget & Lamarre , 1992), while the performed-direction group should encounter greater 293 difficulties in the production of horizontal and vertical segments, visual control being dominant for these types of segment (Meulenbroek & Thomassen, 1991; Gullaud & Vinter, 1998). Experiment 1 Method Subjects Thirty-two right-handed adults (mean age 23.9 years, range 18–28 years) who were students at the University of Bourgogne participated in the experiment, for which they received course credits. They were unaware of the aims of the study. All of them spontaneously used their right hand to write their name on a sheet of paper, were familiar with a language using only left-to-right handwriting and did not have visual or motor problems. Stimuli Fig. 1 Illustration of the experimental set-up used in the mirrordrawing task. An opaque box was positioned over the paper sheet, which prevented the subject from having a direct view of the model, his writing hand, and the pen. An opening at the left of the opaque box allowed the subject to visually control his/her movement from a mirror located at the left (8 cm) of the sheet. The small rectangle at the far end of the mirror is the place where the models appeared. A videotape (experiment 1) and an xy digitizer connected to a computer (experiment 2) were used to record the drawing productions Eight models composed of two segments meeting at a right angle were used in this experiment. Half of the models, called ‘‘hv structure,’’ were composed of the four possible combinations of horizontal and vertical line segments (e.g., the four corners of a square shape), whereas the other half, called ‘‘ob structure,’’ were composed of the four possible combinations of oblique line segments (e.g., the four corners of a diamond shape). The line segment length was 1 cm for all models. Each model was printed in black ink on a 6·6 cm white card, and models were presented one at a time. Subjects were instructed to copy each model one at a time and as accurately as possible. The starting points, movement directions, and movement sequences were free and pen-lifts were allowed between two segments. Procedure The subjects participated in two successive conditions within a single experimental session. In the normal condition, they were able to look normally at their drawing performance. In the mirror condition, they performed the experimental task looking at the model, their hand, and the trace through a mirror. The mirror was located at the left of the sheet of paper and was placed perpendicular to the graphic workplane. As illustrated in Fig. 1, a mask prevented the subjects from having a direct view of the model and of their drawing movements. The subjects began the task by copying four random series of the eight models in the normal condition. In a second phase and in order to familiarize them with the difficulties involved in the mirror task, they copied 12 one-segment models (four horizontals, four verticals, and four obliques) in the mirror condition. After familiarization, they drew the eight models presented in a random order four times under the mirror-drawing condition. Because one of the 32 subjects was unable to draw in the mirror condition despite numerous trials, 1,984 drawings (31 subjects · 2 conditions · 4 trials · 8 models) were retained for the analysis. Data analysis Each drawing was recorded by means of a videotape. For each model, we coded the direction and the sequence of the drawing movements. The mean frequencies of occurrence of verticals [to the top (90) and to the bottom (270)], horizontals [to the right (0) and to the left (180)] and obliques [from bottom left to top right (45), from top left to bottom right (315), from top right Fig. 2 Illustration of discordances between ‘‘performed’’ and ‘‘perceived’’ movement directions to bottom left (225) and from bottom right to top left (135)] were thus computed. In the coding of movement directions, minor angular deviations were not considered. We were mainly interested in coding whether a vertical was drawn, for instance, in the 90 or in the 270 direction. For convenience, we will term performed movements the movements realized on the sheet of paper, and perceived movements those visible in the mirror. As illustrated in Fig. 2, the performed and perceived drawing movements are conflicting in the mirror-drawing condition. Exceptions are for the vertical directions (90, 270). Because the same information is given by an analysis of either the performed or the perceived movement directions, we restricted the analysis to the frequencies of occurrence of performed movement directions only. The tape recording allowed us to pick up and compute the number of errors in movement direction. Errors were characterized by an entire movement or by a ‘‘micro–movement’’ (minimum length of 1 mm) performed in the wrong direction or orientation when compared with the model. In the latter case, a deviation of 22.5 on both sides of each standard segment (i.e., 0, 45, 90, 135, 180, 225, 270, and 315) was defined, and an error was coded when the segment orientation exceeded this deviation. 294 Fig. 3 Simultaneous representation of subjects and graphic movement directions as a function of the first and second factorial axis. Numbers 1–31 indicate the individual subjects. D0, D45, D90, D135, D180, D225, D270, and D315 correspond to movement directions performed to 0, 45, 90, 135, 180, 225, 270, and 315, respectively Results Movement directions The main goal of the first experiment was to check whether two groups of subjects could be identified in the mirror condition on the basis of the reversal or nonreversal of performed movement directions, as observed in a previous study (Gullaud & Vinter, 1996). Consequently, for each subject, the frequency of occurrences of each direction (0, 45, 90, 135, 180, 225, 270, and 315) produced on the sheet of paper (performed directions) when drawing in the mirror condition was computed. The data were collected in a 31 (number of Fig. 4 Polar representation of the frequency of performed movement directions as a function of Condition (normal vs mirror) and Group (perceived-direction vs performed-direction) subjects) by 8 (number of directions) frequency table. This table was composed of 248 cells, including the frequency of occurrences of each direction for each subject. It was analyzed using a correspondence analysis (a type of principal component analysis using frequency tables as data; for details on the procedure and analysis, see Greenacre, 1984; Benzécri, 1973). This procedure illustrates the presence of different groups of subjects as a function of the movement directions they mainly used in the mirror condition. The results are presented in Fig. 3. The first factorial axis explained 47.99% of the variance. As expected, this axis divided subjects into two groups. The first group included 16 subjects who predominantly produced left-oriented segment directions (135, 180, and 225). In contrast, the second group included the 15 remaining subjects who predominantly produced right-oriented segment directions (0, 45, and 315). The first and second groups were termed the perceived-direction and performed-direction groups respectively. Subjects in the perceived-direction group might have favored left-oriented performed movement directions because the resulting perceived directions agreed with those realized in the normal condition. By contrast, subjects in the performed-direction group might have favored left-oriented perceived movement directions because the resulting performed directions agreed with those realized in the normal condition. This interpretation was tested in the following part by an examination of the frequency of performed directions. Figure 4 illustrates the mean frequencies of occurrence for performed directions produced by each group. ANOVAs were carried out with Group (perceived direction versus performed direction) as the betweensubjects factor, and with Condition (normal versus mirror) and Block of trials (four blocks) as the withinsubjects factors. Significant interactions were investigated by means of Newman-Keuls post hoc tests. All subjects performed left-oriented movements more frequently in the mirror than in the normal condition [horizontal, F(1, 29)=86.86, p<.001; oblique 45–225, F(1, 29)=33.62, p<.001; oblique 135–315, F(1, 29)=198.42, p<.001]. A significant interaction between 295 Fig. 5 Illustration of the dominant movement sequences realized by perceived-direction vs. performed-direction groups for each model in normal (standard sequence) and mirror conditions (performed and perceived sequences). For a given model, dotted drawing showed that performed or perceived sequences never occurred in the normal condition, while bold continuous drawing showed that the standard sequences were preserved in mirror production Group and Condition for horizontals [F(1, 29)=10.08, p<.005], top left to bottom right obliques [F(1, 29)=41.84, p<.001] and top right to bottom left obliques [F(1, 29)=10.87, p<.005] showed that the left-oriented movement directions were more frequent in the perceived-direction group than in the performed-direction group when drawing in the mirror. Whatever the segment orientation, no other significant effect was observed, and in particular there was no effect of Block of trials (p>.09). Movement sequencing For each model and condition, the dominant movement sequences observed in the perceived-direction and performed-direction groups were established. The results are presented in Fig. 5. The standard sequence corresponds to the usual movement sequence produced in the normal condition. Because no differences were observed between the groups for each model (p>.22), the frequencies of occurrence of the dominant sequences in the normal condition were averaged for all subjects (second column), thereby simplifying graphic presentation. The perceived and performed sequences correspond to movements as they were seen in the mirror or realized on the sheet of paper respectively. Note that whatever the models, the dominant movement sequences were produced without pen-lifts. Figure 5 shows that the models composed of horizontals and verticals only (models a, b, 296 c, d) were reversed in the mirror, because of the mirror transformations induced on the left-right and right-left horizontal directions. Figure 5 reveals that, for both groups, the standard movement sequences were preserved in the mirror condition for the models c, d, g, and h [F(1, 29)=1.77, p>.19]. By contrast, performance varied as a function of Condition and Group for models a, b, e, and f [F(1, 29)>16.02, p<.001]. For each of these four models, subjects in the perceived-direction group visually preserved the sequences used in the normal condition for the same model (indicated by bold continuous drawing on Fig. 5). But because of the mirror transformation, the resulting performed sequences were new in comparison to the normal condition (indicated by dotted drawing on Fig. 5). Subjects in the performed-direction group preserved the performed sequences as they did in the normal condition for the same model (models a, b, e, and f). Their movement sequence choices did not seem to depend on what they saw in the mirror, the perceived sequences never being preserved (as illustrated by dotted drawings). Number of errors No error was recorded in the normal condition. The mean number of errors (i.e., entire movement and micro-movement performed in the wrong direction or orientation) made in the mirror-drawing task was computed. An ANOVA was carried out with Group as the between-subjects factor, and Structure and Block of trials as the within-subjects factors. Whatever the Group, the mean number of errors was greater for oblique models (1.44) than for horizontal-vertical ones (0.87) [F(1, 29)=24.45, p<.001], and decreased with practice (block 1=1.8, block 2=1.1, block 3=1.0, block 4=0.7) [F(3, 87)=16.69, p<.001]. Contrary to our expectations, no effect of Group was observed (p>.12). However, differences between groups appeared with regard to on-line movement corrections: the micromovements were more frequently observed in the perceived-direction group (57%) than in the performeddirection group [39.7%; v2(2)=9.23, p<.005]. Discussion Experiment 1 showed that two distinct subject groups were observed when participants had to cope with visuoproprioceptive discordance in a mirror-drawing task. Some subjects preserved the visual directions and sequences that occurred in normal drawing by reversing the direction of drawing movements (perceiveddirection group), while others preserved normal drawing directions and sequences, which produced reversed visual directions (performed-direction group). The analysis of movement directions and sequences suggests that vision could be the principal determinant for the perceived-direction group when resolving the visuoproprioceptive conflicts, to the detriment of biomechanical constraints. An additional argument supporting the idea of the visual guidance of graphic movements in the mirror-drawing condition for this group was provided by the occurrence of micro-movements for segments produced in the wrong direction when compared with the model. By contrast, subjects in the performeddirection group produced significantly less on-line movement corrections, the drawing of the requested model being reversed in consequence. These latter subjects noted that their drawing did not match the model only when it was entirely completed. Contrary to our expectations, a higher number of errors (in the mirror condition) for oblique models when compared with horizontal-vertical models was observed in both groups. Recall that, in accordance with Lajoie and co-workers’ hypothesis (1992), the greater difficulties observed for obliques could suggest the dominance of visual information when resolving discordances. From a visual point of view, obliques are effectively the most complex segments to draw because the conflict entails a multiple choice, whereas for horizontals resolving the conflict entails a binary choice. The aforementioned result of the present study suggests the dominance of visual information when resolving the conflict, whatever the movement direction favored by the subjects. Moreover, the distinction between the geometrical and the anatomical systems of reference reported by Meulenbroek and Thomassen (1991) in straight line production was not confirmed when simple geometrical figures were drawn through a mirror. It may be that the complexity induced by the conflicting task leads subjects to favor vision, the ‘‘Queen of the senses,’’ whatever the line-segment orientation (horizontal-vertical or oblique). Taken together, these results suggest that although visual information was important regardless of the movement directions favored by the subjects in the mirror condition, it may have played a different role for each group. As suggested by movement directions and on-line corrections, performance of subjects in the perceived-direction group could have been under visual guidance in the mirror drawing task. By contrast, the important number of directional errors of entire movements produced in the mirror condition for the performed-direction group shows that these subjects used visual information before and later during movement production, probably to check whether their graphic production and the models were congruent. It is possible that biomechanical constraints are stronger in the latter group, which could explain the preservation of performed movement directions and the necessity for the subjects to plan the resolution of visuoproprioceptive discordance before drawing production. This hypothesis was investigated in the second experiment, by analyzing spatial accuracy and temporal features of drawing movements. 297 Experiment 2 The second experiment investigated the underlying processes used by the perceived-direction and performed-direction groups when solving visuoproprioceptive conflicts. As already mentioned, the availability of visual information during movement execution is determinant with regard to the topokinetic component of graphic behavior, whereas the morphokinetic component seems to be preserved whatever the visual exposure conditions (Paillard, 1991; Gullaud, 1998). We thus expected that the importance of on-line visual control in the perceived-direction group would be manifested through a greater accuracy of spatial orientation than in the performed-direction group. In the present experiment, the analysis of temporal features of drawing production, including movement programming and movement execution, was considered of prime importance in understanding how both groups proceeded to solve visuoproprioceptive conflict. Did the subjects solve the sensory discordance before or during graphic production, illustrating a programming-based strategy or a feedback-based strategy respectively? Not just the latency phase (reaction time) but also later parts of the response (movement time and duration of pauses between two segments) needed to be analyzed to answer this question. We expected that the greater importance of on-line visual corrections for the perceived-direction group would be reflected in longer movement time, numerous velocity peaks, and longer duration of pauses, the last-mentioned being necessary in order to recalibrate the movement in another direction. By contrast, if the preservation of performed movement directions has some basic determinants grounded in the biomechanical system (performed-direction group), this should imply the elaboration of a feedforward plan taking into account the mirror transformations before movement initiation, in order to respect the syntactical rules. In a previous experiment, Plamondon and Clément (1991) showed that movement directions affect movement time without disturbing performance accuracy. They reported that more time was needed to produce right to left than left to right graphic progressions. Consequently, the location of the starting point and the movement direction were imposed on participants in Experiment 2, in order to explore how changes in movement directions affected graphic performance in the mirror-drawing task for both groups. The reader is reminded that each group behaves differently with regard to the directions most often produced on the paper sheet when mirror drawing (Gullaud & Vinter, 1996), with right to left directions in the perceived-direction group and left to right directions in the performed-direction group. For each model, participants were thus forced to use their preferred movement sequence on the paper sheet (i.e., from the left in the perceived-direction group, and from the right in the performed-direction group) and their nonpreferred sequence (i.e., from the right in the perceived-direction group, and from the left in the performed-direction group). Method Subjects Eighteen of the 31 subjects who participated in the first experiment were selected on the basis of the frequency of movement directions favored in the mirror-drawing condition. This frequency was computed using models a, b, e, and f (see Experiment 1, Fig. 5), for which subjects behaved differently and preserved either a performed or a perceived movement sequence. Ten subjects in the perceived-direction group of Experiment 1 were kept in Experiment 2 because in more than 62.5% of cases (mean 84.8%) they adopted the same perceived movement sequence as in the normal condition. The eight remaining subjects were from the performed-direction group of Experiment 1. They preserved a performed movement sequence in more than 62.5% of the cases (mean 75%). Stimuli and apparatus As mentioned previously, four models of Experiment 1 (models a, b, e, and f) were retained in the second experiment. Two models were composed of horizontal and vertical line segments (‘‘hv’’ structure), and the two remaining models were made up of oblique line segments (‘‘ob’’ structure). The task was performed on an xy digitizer (Calcomp Drawingboard) connected to an IBM-PC computer. The xy coordinates were sampled at a frequency of 100 Hz, with a spatial accuracy of 0.2 mm (for technical details, see Teulings & Maarse, 1984). Procedure The experimental conditions were similar to those used in the first experiment: a normal and a mirror condition. Each drawing trial started with a high-frequency acoustic signal, which indicated to the subjects that they had to begin their drawing production. When they had finished, an interactive recording procedure allowed the experimenter to stop the process of data acquisition. As illustrated in Fig. 6, two opposite movement sequences were used in this experiment and the starting point as well as the movement directions were imposed for each model by arrows. The term preferred sequence indicates that subjects copied the model using a sequence that they spontaneously produced in the mirror task (Experiment 1), the reverse being true for the nonpreferred sequence. Because two distinct behaviors (reversal vs nonreversal of performed sequences) were observed in the mirror condition as a function of the Group, the preferred sequence was the visually perceived one in the perceived-direction group, while it corresponded to the performed sequence in the performed-direction group. It should be remembered that, as shown in Experiment 1, all the subjects behaved similarly in the normal condition. In this experiment, 64 models were drawn by subjects (4 models · 2 sequences · 4 trials · 2 conditions). The total number of recorded movement sequences was 1,152. All models were presented in a random order in each condition. Data analysis Each drawing was identified and smoothed with a second-order filter with a transition band between 8 and 20 Hz. The filtered data were displayed alongside tangential velocity profiles. An interactive analysis program1 was used to extract different temporal variables for each segment. The temporal variables were 1 The recording apparatus (interface and pen), acquisition and data analysis programs were made by the Nijmegen Institute for Cognition and Information (NICI). 298 Fig. 6 Illustration of the models used in experiment 2 and of the preferred versus nonpreferred sequence for each model. These two sequences were determined as a function of subjects’ performance spontaneously realized in the first experiment. Arrows indicated the starting point and the movement directions for each model the reaction time (ms) and the duration of pauses (ms) between the two segments of each model, the tangential velocity (cm/s), and the number of velocity peaks for each segment (each peak being due to successive acceleration and deceleration phases in drawing movements). Two measures of drawing accuracy were computed: the spatial orientation of the drawing in the workspace and the size of the angles (in degrees). For both of these dependent variables, the root mean square error (RMSE) was analyzed, because it is considered the best overall measure of performance accuracy (Henry 1975; Schmidt 1988). The other variables were the mean number of errors, as computed in the first experiment (the total number of entire movements and ‘‘micro-movements’’ performed in the wrong direction or orientation when compared with the model) and the frequency of micro-movements only (on-line movement corrections). The dependent variables were analyzed by means of separate ANOVAs with Group (perceived direction vs performed direction) as the between-subjects factor, and Condition (normal vs mirror), Structure of models (hv vs ob), and Sequence (preferred vs nonpreferred) as within-subjects factors. NewmanKeuls post hoc tests were used to investigate the significant interactions. Fig. 8A, B RMSE of the spatial orientation of models (A) and of angular accuracy (B) as a function of Group (perceived-direction vs performed-direction), Condition (normal vs mirror), and Structure (hv vs ob) Fig. 7 Mean number of errors produced in the mirror-drawing condition as a function of Group (perceived-direction vs performed-direction), Structure (hv vs ob) and Sequence (preferred vs nonpreferred) Results Spatial accuracy of drawing productions As presented in Fig. 7, most of the errors were made when subjects used a nonpreferred sequence and when they drew oblique models [F(1, 16)=6.83 and 19.72, p<.001, respectively]. However, the occurrence of errors tended to be greater for oblique models than for horizontals and verticals when subjects in the performeddirection group drew in a nonpreferred direction [Group · Sequence · Structure, F(1, 16)=3.81, p=.066]. A qualitative analysis of errors confirmed the importance of visual guidance of mirror-drawing movements for the perceived-direction group. Micromovements were more frequent in the perceived-direction group (42.37%) than in the performed-direction group (7.5%) [v2(2)=17.12, p<.001]. These results reproduced those of Experiment 1. The subsequent analyses included the drawings produced without errors, that is 1,053 drawings out of the 1,152 collected. The RMSE relative to the spatial orientation of models in the workspace is displayed in Fig. 8A. On average, the RMSE was higher in the mirror 299 Fig. 9A–D Reaction times (A), duration of pauses (B), tangential velocity (C), and mean number of velocity peaks (D) as a function of Group (perceived-direction vs performed-direction), Condition (normal vs mirror), and Sequence (preferred vs nonpreferred) condition [F(1, 16)=28.45, p<.001], for oblique models [F(1, 16)=31.68, p<.001], and in the performed-direction group [F(1, 16)=5.48, p<.05]. The difference between groups mainly resulted from mirror-drawing productions [Group · Condition, F(1, 16)=4.17, p=.058]. Both groups performed differently as a function of Structure [Group · Structure, F(1, 16)=6.54, p<.025]. The difference between hv and ob structure in the performed-direction group was only tendential (Newman-Keuls, p=.066) and it disappeared in the mirror condition, while oblique models were significantly less accurate than horizontal-vertical models in both normal and mirror conditions in the perceived-direction group (p<.001). No other significant effect was observed, and in particular there was no effect of Sequence (F<1). Figure 8B depicts the RMSE of angular accuracy. No main effect of Group was revealed by ANOVAs (F<1). For all subjects, the angular error was higher in the mirror condition [F(1, 16)=16.59, p<.001] and for oblique models [F(1, 16)=16.94, p<.001], but it was not modified by Sequence (F<1). The RMSE varied as a function of Condition and Structure [F(1, 16)=4.38, p=.053], and between Group, Condition, and Structure [F(1, 16)=4.29, p<.05]. The effect of Structure appeared to be similar in the normal and mirror conditions in the perceived-direction group (p>.26). By contrast, subjects in the performed-direction group had difficulty in drawing horizontal-vertical models in the mirror condition when compared to performances in the normal condition [F(1, 16)=15.26, p<.001]. No other significant effect appeared. Analysis of temporal variables Mean reaction times are displayed in Fig. 9A. ANOVA revealed that they were longer in the mirror than in the normal condition [F(1, 16)=37.83, p<.001], and for a nonpreferred sequence than for a preferred one [F(1, 16)=9.81, p<.005]. However, the marginally significant interaction between Condition and Sequence [F(1, 16)=3.98, p=.056] showed that the increase in reaction times when using a nonpreferred sequence mainly occurred in the mirror-drawing condition. Differences appeared between the perceived-direction and the performed-direction group in the mirror task only [F(1, 16)=4.48, p<.05], the increase in reaction times being significantly more important in the performeddirection group. No effect of Structure [F(1, 16)=1.06, p=.32] and no other interaction were revealed by the ANOVAs ( p>.33). The mean duration of pauses, illustrated in Fig. 9B, also increased in the mirror condition for all subjects [F(1, 16)=16.83, p<.001], but it increased significantly more in the perceived-direction group than in the performed-direction group [F(1, 16)=4.58, p<.05]. When a nonpreferred sequence was imposed, an increase in the duration of pauses was observed [F(1, 16)=4.46, 300 p<.05], especially in the mirror condition and for the perceived-direction group [Group · Condition · Sequence, F(1, 16)=3.98, p=.06]. No other significant effect appeared. As shown in Fig. 9C, slower movements were observed in the mirror-drawing condition [F(1, 16)=15.60, p<.001], for a nonpreferred sequence [F(1, 16)=4.60, p<.05], and in the perceived-direction group [F(1, 16)=5.90, p<.05]. In the mirror task, velocities decreased significantly more in the perceived-direction group than in the performed-direction group [F(1, 16)=4.22, p<.05], especially for a preferred sequence [Group · Condition · Sequence, F(1, 16)=8.33, p<.01]. For all subjects, the mean tangential velocity was higher for oblique models (3.9 cm/s) than for horizontal-vertical models (3.6 cm/s) [F(1, 16)=14.88, p<.001]. Finally, the analysis of the mean number of velocity peaks (Fig. 9D) revealed that movements were less fluent in the mirror condition [F(1, 16)=27.86, p<.001] and for oblique models [hv structure =2.4, ob structure =3.0; F(1, 16)=11.85, p<.005]. Movements tended to be more dysfluent in the mirror condition for the perceived-direction group than for the performeddirection one [F(1, 16)=3.93, p=.064]. Moreover, the significant interaction between Group, Condition, and Sequence [F(1, 16)=6.27, p<.025] showed that the increase in movement dysfluency in the mirror condition for the perceived-direction group was greater for the preferred sequence, while this was true for the nonpreferred sequence in the performed-direction group. No other significant effect was revealed by the ANOVAs (p>.23). Discussion Experiment 2 confirmed that subjects’ drawing behavior differed as a function of their group (perceived-direction versus performed-direction) with regard to temporal parameters and performance accuracy. While no difference between groups appeared in the normal condition, the subjects from each group reacted differently when resolving the visuoproprioceptive discordance. The occurrence of micro-movements, the longer duration of pauses, the slower movement velocity, and the greater movement dysfluency show that the perceived-direction group encountered more difficulties during movement execution in the mirror-drawing task. By contrast, the major changes in reaction times observed in the performed-direction group show that much of the conflict was resolved during movement planning. In addition to these temporal differences, the groups differed with regard to spatial accuracy. Subjects in the performed-direction group were less accurate in reproducing model orientation than those in the perceived-direction group, and they did not exhibit any modifications of the morphokinetic component (angular accuracy), especially when drawing the oblique models. In accordance with the literature showing that on-line visual feedback plays a major role in spatial orientation accuracy (Broderick & Laszlo, 1987; Gullaud, 1998; Paillard, 1991; Van Doorn & Keuss, 1993), the present experiments confirm that the perceived-direction group makes greater use of visual information than the performed-direction group in graphic movement production. Differences between the production of oblique models and horizontal-vertical models were expected from the literature. In straight line production (Meulenbroek & Thomassen, 1991) and in simple geometrical drawing tasks performed in nonconflicting conditions (Gullaud, 1998; Gullaud & Vinter, 1998), it was shown that the visual system was dominant for horizontal and vertical line segments, while motoric control was prevalent for obliques. Consequently, we expected that horizontalvertical drawing production would be easier for the perceived-direction group, and the drawing of obliques easier for the performed-direction group. However, in the present experiment no significant Group · Structure of models interaction was revealed by the ANOVAs (on errors and temporal variables). For both groups, errors were more frequent for obliques than for horizontals and verticals, and movements were less fluent for the former than for the latter. These results suggest that visual control predominates over motoric control for all subjects, whatever the directions they spontaneously use when drawing through a mirror. Movement sequences (preferred vs nonpreferred) were imposed on participants in order to explore how changes in movement direction affected the graphic performance of perceived-direction and performeddirection groups. As expected (Plamondon & Clement, 1991), graphic accuracy (i.e., spatial orientation and size of angle) did not change as a function of left to right or right to left movement progressions in either group, while changes in movement execution parameters (i.e., tangential velocity and number of velocity peaks) were mainly observed in the perceived-direction group for preferred and nonpreferred sequences (Newman-Keuls, p>.05). Thus, difficulties encountered in terms of movement execution in the mirror task by the perceiveddirection group cannot be explained by the production of movements from right to left. Increases in reaction times in both groups and in the duration of pauses in the perceived-direction group were observed when a nonpreferred sequence was used in the mirror condition. These results suggest that when subjects were forced to use a sequence that they did not spontaneously produce in the mirror, they encountered more difficulties in the planning of movement direction. General discussion Both experiments confirm that two different information-processing strategies appear when adults are required to draw in a conflicting task. With regard to movement sequence, some of them favor the visually perceived sequence, while others favor the performed 301 sequence. The importance of visual guidance in the perceived-direction group is revealed by the analysis of the spatial accuracy of drawing production, especially with regard to spatial orientation of movements in the graphic workspace. Moreover, the temporal data show that the perceived-direction group encounters more difficulties during movement execution in the mirrordrawing task. These findings suggest that subjects in the perceived-direction group rely mainly on a visual feedback control of movement when they have to resolve visuoproprioceptive discordances. By contrast, visual guidance does not seem to play a major role in the performed-direction group. The changes in reaction times observed in this latter group show that much of the conflict is resolved during motor programming. Subjects in this group would rely mainly on a feedforward control of movements when drawing in the mirror. Such results seem in accordance with prism adaptation studies in which some authors have reported that two motor control strategies can occur (Redding & Wallace, 1996). In this case, when both starting and target locations were simultaneously visible, the relative difference between locations were used to initiate a feedforward movement plan (i.e., feedforward control strategy). When starting locations were not visible, the differences between target location and effector location, when available, were directly used for feedback corrections (i.e., feedback control strategy). However, whereas the nature of strategic control appeared to be task-specific in prism exposure, the present experiments on mirror drawing suggest a more subject-dependent approach to account for the type of strategies used to solve visuoproprioceptive conflicts. These results call for further investigations, and especially for developmental experiments. Research in motor development has revealed age-related differences in the use of feedforward and feedback processes (Bard, Hay & Fleury, 1985; Hay, 1979, 1984; Olivier, Audiffren & Ripoll, 1998), with a critical period in development of movement control at around 7–8 years. The changes in spatial accuracy and timing in movements were interpreted as a changing dominance of pro- (5-year-olds) and retroactive control (7-year-olds), followed by a mixed control in older children (11-year-olds). Investigation of mirror-drawing production in children seems interesting to observe how changes in feedforward and feedback processes can affect performance in the case of visuoproprioceptive conflicts. Before the age of 6 years, most children should belong to the performed-direction group because of the dominant role of feedforward processes, while at 7–8 years most children should behave like the perceived-direction group because of the dominant role of feedback processes. Moreover, we might expect to observe a differential dichotomy in older children, as was the case in adults. Wondering why adults react differently when resolving visuoproprioceptive discordance also opens up a fruitful avenue of research. References Bard, C., Hay, L., & Fleury, M. (1985). Contribution of vision to the performance and learning of a directional aiming task in children aged 6, 9 and 11. In Y.E. Clark & J. H. Humphrey (Eds.), Motor development: current selected research, vol I. (pp. 19–33). Benzécri, J. P. (1973). L’analyse de données, vol 2. Paris: Dunod. Broderick, P., & Laszlo, J. I. (1987). The drawing of squares and diamonds: a perceptual-motor task analysis. Journal of Experimental Child Psychology, 43, 44–61. Greenacre, M. J. (1984). Theory and applications of correspondence analysis. London: Academic Press. Gullaud, L. (1998). Syste`me de re´fe´rences spatiales et production graphique se´quentielle: Une approche au moyen des paradigmes de suppression visuelle et de conflits visuo-proprioceptifs. Unpublished Ph.D. Thesis, University of Dijon, France. Gullaud, L., & Vinter, A. (1996). The role of visual and proprioceptive information in mirror-drawing behavior. In M. L. Simner, C. G. Leedham, & A. J. M. W. Thomassen (Eds.), Handwriting and drawing research: basic and applied issues (pp. 99–113). Amsterdam: IOS Press. Gullaud, L., & Vinter, A. (1998). Rôle de la vision dans la direction de mouvements graphiques simples. L’Anne´e Psychologique, 98, 401–428. Hay, L. (1979). Spatial-temporal analysis of movements in children: motor programs versus feedback in the development of reaching. Journal of Motor Behavior, 11, 189–200. Hay, L. (1984). Discontinuity in the development of motor control in children. In W. Prinz & A. F. Sanders (Eds.), Cognition and motor processes (pp. 351–360). Berlin, Heidelberg, New York: Springer. Henry, F. M. (1975). Absolute error vs. ‘‘E’’ in target accuracy. Journal of Motor Behavior, 7, 227–228. Lajoie, Y., Paillard, J., Teasdale, N., Bard, C., Fleury, M., Forget, R., & Lamarre, Y. (1992). Mirror drawing in a deafferented patient and normal subjects: visuoproprioceptive conflict. Neurologia, 42, 1104–1106. Meulenbroek, R. G. J., & Thomassen, A. J. W. M. (1991). Strokedirection preferences in drawing and handwriting. Human Movement Science, 10, 247–270. Olivier, I., Audiffren, M., & Ripoll, H. (1998). Age-related differences in the preparatory processes of motor programming. Journal of Experimental Child Psychology, 69, 49–65. Paillard, J. (1991). Les bases nerveuses du contrôle visuo-manuel de l’écriture. In C. Sirat, J. Irigoin, & E. Poulle (Eds.), L’e´criture, le cerveau, l’oeil et la main (pp. 23–52). Paris: Brepols. Plamondon, R., & Clément, B. (1991). Dependence of peripheral and central parameters describing handwriting generation on movement direction. Human Movement Science, 10, 193–221. Redding, G. M., & Wallace, B. (1988a). Components of prism adaptation in terminal and concurrent exposure: organization of eye-hand coordination loop. Perceptual Psychophysics, 44, 59–68. Redding, G. M., & Wallace, B. (1988b). Adaptive mechanisms in perceptual-motor coordination: components of prism adaptation. Journal of Motor Behavior, 20, 242–254. Redding, G. M., & Wallace, B. (1996). Adaptive spatial alignment and strategic perceptual-motor control. Journal of Experimental Psychology: Human Perception and Performance, 22, 379– 394. Schmidt, R. A. (1988). Motor control and learning: a behavioral emphasis, 2nd edn. Champaign, Ill.: Human Kinetics. Teasdale, N., Forget, R., Bard, C., Paillard, J., Fleury, M., & Lamarre, Y. (1993). The role of proprioceptive information for the production of isometric forces and for handwriting tasks. Acta Psychologica, 82, 179–191. Teulings, H. L., & Maarse, F. J. (1984). Digital recording and processing of handwriting movements. Human Movement Science, 3, 193–219. 302 Thomassen, A. J. W. M., Tibosch, H. J. C. M., & Maarse, F. J. (1989). The effect of context on stroke direction and stroke order in handwriting. In R. Plamondon, C. Y. Suen, & M. L. Simner (Eds.), Computer recognition and human production of handwriting (pp. 213–230). Singapore: World Scientific. Uhlarik, J. J., & Canon, L. K. (1971). Influence of concurrent and terminal exposure conditions on the nature of perceptual adaptation. Journal of Experimental Psychology, 91, 233–239. Van Doorn, R. R. A. (1992). Axial pen pressure during handwriting movements in conditions with and without vision. Journal of Human Movement Studies, 23, 41–52. Van Doorn, R. R. A., & Keuss, P. J. G. (1992). The role of vision in the temporal and spatial control of handwriting. Acta Psychologica, 81, 269–286. Van Doorn, R. R. A., & Keuss, P. J. G. (1993). Spatial invariance of handwriting: a matter of definition. Reading and Writing, 5, 281–296. Van Sommers, P. (1984). Drawing and cognition: Descriptive and experimental studies of graphic production processes. Cambridge: Cambridge University Press. Vinter, A., & Meulenbroek, R. G. J. (1993). The role of manual dominance and visual feedback in circular drawing movements. Journal of Human Movement Studies, 25, 11–37.