Visual and haptic discrimination of symmetry in

Perception, 2004, volume 33, pages 315 ^ 327

DOI:10.1068/p5017

Visual and haptic discrimination of symmetry in unfamiliar displays extended in the z-axis Soledad Ballesteros, Jose¨ M Reales

Facultad de Psicolog|¨ a, Universidad Nacional de Educacio¨n a Distancia, Juan del Rosal 10, 28040 Madrid, Spain; e-mail: [email protected] Received 25 November 2002, in revised form 3 October 2003; published online 10 March 2004

Abstract. We investigated, in two experiments, the discrimination of bilateral symmetry in vision and touch using four sets of unfamiliar displays. They varied in complexity from 3 to 30 turns. Two sets were 2-D flat forms (raised-line shapes and raised surfaces) while the other two were 3-D objects constructed by extending the 2-D shapes in height (short and tall objects). Experiment 1 showed that visual accuracy was excellent but latencies increased for raised-line shapes compared with 3-D objects. Experiment 2 showed that unimanual exploration was more accurate for asymmetric than for symmetric judgments, but only for 2-D shapes and short objects. Bimanual exploration at the body midline facilitated the discrimination of symmetric shapes without changing performance with asymmetric ones. Accuracy for haptically explored symmetric stimuli improved as the stimuli were extended in the third dimension, while no such a trend appeared for asymmetric stimuli. Unlike vision, haptic response latency decreased for 2-D shapes compared with 3-D objects. The present results are relevant to the understanding of symmetry discrimination in vision and touch.

1 Introduction The top and the bottom parts of most objects are often quite different but the left and right sides are frequently similar. In nature, objects exhibiting bilateral symmetry about a vertical axis are very common. Perceptual psychologists have always been interested in studying the role of symmetry in visual perception. In contrast, very few studies have been designed to study the discrimination of this shape property in the haptic modality. The discrimination of bilateral symmetry is important, not only for human perception, but also for animal perception. For example, spontaneous preferences for visual patterns showed that bees can detect the vertical axis of patterns with bilateral symmetry without any training (Lehrer et al 1995). This is biologically significant, because those hives that are more symmetric have more nectar. Honeybees detect an axis of bilateral symmetry and learn to discriminate a bilaterally symmetric pattern from one that is nonsymmetric (Horridge 1996). In human vision, bilateral symmetry has been intensively studied over the years (eg Julesz 1971; Locher and Nodine 1973; Locher and Wagemans 1993; Palmer 1989; Palmer and Hemenway 1978; Rock 1983; Royer 1981; Wagemans 1995; Wagemans et al 1992) because bilateral symmetry about a vertical axis has been considered the most salient organisational principle of a given stimulus (eg Barlow and Reeves 1979; Corballis and Roldan 1975; Palmer and Hemenway 1978; Royer 1981; Wagemans et al 1992). Even young babies appreciate it (Bornstein et al 1981). Compared with the number of visual studies, research conducted to disclose the role of bilateral symmetry in touch has been scarce (see Appelle 1991). Millar (1978) conducted a study using small unfamiliar symmetric and asymmetric raised-dot displays and a matching task that did not demand symmetry discrimination. The main result was that differences in dot numerosity (or texture) were detected better than differences in symmetry, leading to the conclusion that haptic explorers relied on texture for spatially coding shape. Millar's interpretation was further supported by a series of

316

S Ballesteros, J M Reales

experiments showing that, for touch, material properties such as texture and hardness are more salient than object properties such as form and size (Klatzky et al 1987). Locher and Simmons (1978) studied the influence of symmetry on haptic perception using eight large (15 cm wide619 cm long) raised symmetric and asymmetric polygon surfaces that varied in complexity (defined in terms of their number of turns). Observers visually inspected the shapes before performing a haptic symmetry-discrimination task. They found that the discrimination of asymmetric polygons was more accurate and faster than the discrimination of symmetric ones at all complexity levels. This finding was supported in another study (Simmons and Locher 1979). Symmetric shapes required twice as much exploration time as asymmetric shapes. Although scanning times decreased and accuracy increased with experience, specific training with the polygons increased the sensitivity for symmetric polygons. However, the discrimination of asymmetric polygons required less time. Furthermore, the benefit of training with specific shapes did not generalise to new shapes. The haptic system is a complex perceptual system that encodes inputs from cutaneous and kinesthetic receptors (Loomis and Lederman 1986). Active touch is very accurate in perceptual identification and recall of familiar objects (eg Klatzky et al 1985; Reales and Ballesteros 1999) and novel 3-D objects (eg Ballesteros et al 1999). In clear contrast, touch is much less accurate and slower in dealing with raised-line displays, according to Lederman and Klatzky (1987). Ballesteros et al (1997) assessed the accuracy of the haptic system to discriminate bilateral symmetry in small (2 cm62 cm) raised-line shapes. They were constructed by connecting 5 or 6 dots in a 363 dot matrix. The novel wooden 3-D objects differed in form and were larger (6 cm67 cm66 cm) than the raised-line shapes. Touch was faster and more accurate in dealing with 3-D objects than 2-D raised shapes (see also Lederman and Klatzky 1987; Klatzky et al 1993). Results reported by Ballesteros et al (1997) are relevant to the purpose of our present study. In general, the haptic system was far more accurate with our 3-D objects than with the raised-line shapes. Furthermore, bimanual exploration was superior to unimanual exploration due to parallel shape information extraction and to the use of the observer's body midline as a body-centred reference frame. A clear-cut picture emerged when considering performance with the raised-line shapes. First, haptic perception was moderately sensitive in detecting bilateral symmetry in small unfamiliar 2-D haptic displays. Second, observers were significantly more accurate with asymmetric than with symmetric displays. Third, accuracy did not differ much across exploration times, although at all exploration times observers were more accurate for asymmetric than for symmetric shape judgments. Fourth, the left hand was not more accurate than the right hand. However, in congruence with the reference-frame hypothesis, bimanual exploration was consistently more accurate than unimanual exploration, either with the left or with the right fingertip. In contrast, the novel 3-D objects showed a quite different pattern of results. First, object discrimination of symmetry was far more accurate than performance with raised-line shapes over a wide range of exploration-time conditions. Second, in contrast to raised-line shapes, the haptic system was more accurate with symmetric objects than with asymmetric ones at all exploration times. Furthermore, subjects were more accurate with symmetric objects than with asymmetric ones, both when they were explored with the preferred hand and when an enclosing hand movement only was allowed. Findings from a new series of experiments are also relevant to the present study (see Ballesteros et al 1998). Stimuli were small open and closed raised-line shapes (half symmetric and half asymmetric) and observers had to detect whether each raised

Visual and haptic discrimination of symmetry

317

shape was `open' or `closed'. Bilateral symmetry was an incidental encoding property in vision, but could also be shown in our incidental tactual task when sufficient reference information was provided at encoding. The results showed for the first time in haptic perception something that is well known in vision: shape symmetry facilitated tactual processing when reference information for spatial coding was provided at encoding. Note, however, that the two stimulus sets were different in many ways (eg size, shape, and complexity) and any comparison had to be considered tentative (Ballesteros et al 1997). Further research that focused directly on studying performance with comparable sets of stimuli was needed. This is because performance on the symmetry-discrimination task depends critically on the stimuli used (see Millar 1994). According to Millar (1994), the range and type of information needed for shape perception depends on the size and the type of stimuli. For example, small raised-line shapes are difficult to code by reference to body-centred spatial coordinates. On the other hand, the perception of 3-D objects manipulated by both hands depends on touch and movement information. The goals of the present study were threefold. The first was to study performance in symmetry discrimination by using a new set of stimuli consisting of 2-D shapes (raised-line shapes and raised surfaces) and 3-D objects (short and tall objects). The 3-D objects were constructed by extending the 2-D shapes in the z -axis. The elongation of the stimulus shapes should permit a better and more informative exploration of objects by touch, facilitating symmetry judgments. Two quite different elongations were selected: 0.5 cm and 6 cm. The 0.5 cm height, though very short, is sufficient to afford the feeling of 3-D. The 6 cm height matched our previous 1997 study. The rationale was to find out whether extending the stimuli even a little in height would influence performance compared with the 2-D displays. The second goal was to compare visual performance with haptic performance by presenting the same stimuli to vision and touch, instead of using 2-D pictorial representations of 3-D objects. As far as we know, this is the first study of symmetry that presented the same displays to vision and touch. Third, we further explored the reference-frame hypothesis. The effect of providing reference information should be more effective for raised shapes than for objects, since reference information is very poor for those stimuli when they are explored with one finger. In experiment 1, participants made visual-symmetry judgments. In experiment 2, they performed the same task by exploring the stimuli with their preferred hand (unimanual condition) or using both hands (bimanual condition). To test the reference-frame hypothesis, hands were aligned to the midbody axis to provide reference information. This hypothesis predicts that exploration with both hands aligned to the midbody axis will facilitate the discrimination of bilateral symmetry of 2-D shapes. In summary, the novelty of the present research was that the symmetry-discrimination task was performed with four sets of materials designed to investigate the influence of the third dimension in the discrimination of this spatial property of shapes and objects. 2 Experiment 1 In experiment 1 we investigated the accuracy and response time of participants in the visually presented symmetry ^ asymmetry-discrimination task. According to the literature, it was expected that visual judgments would be highly accurate and very fast. 2.1 Method 2.1.1 Participants. Sixteen students at the Universidad Nacional de Educacio¨n a Distancia voluntarily participated in an experimental session that lasted approximately half an hour. All participants had normal or corrected-to-normal vision and were experimentally na|« ve.

318

S Ballesteros, J M Reales

2.1.2 Materials. The target stimuli were the four sets of materials carefully designed to be suitable for vision as well as for touch. Each stimulus set was comprised of 9 symmetric and 9 asymmetric stimuli matched in shape, size, and complexity defined by the number of turns (see Locher and Simmons 1978). Stimuli were parametrised by the number of sides (turns) in the perimeter of the shape. The symmetric and asymmetric shapes had 3, 4, 6, 8, 12, 14, 18, 24, and 30 sides each. These more complex stimuli were prepared from the Locher and Simmons' (1978) shapes by reducing their size. Examples of the four sets of stimuli that were used in experiments 1 and 2 are displayed in figure 1. A total of 72 displays were prepared by ONCE (the Spanish National Blind Organisation). Sets 1 and 2 (raised-line shapes and raised surfaces) were prepared from a special paper easily raised by a thermoforming procedure. The solid objects (sets 3 and 4) were constructed from cubes of rigid polyethylene of high molecular weight, low coefficient of friction, and good resistance to abrasion. Observers felt very comfortable touching the raised displays as well as the objects. Each stimulus was presented once only.

Figure 1. Examples of the four sets of unfamiliar stimuli used in experiments 1 and 2. The stimuli ranged from 2-D shapes (polygons with raised outlines and raised surfaces) to 3-D objects (short and tall objects).

Stimuli in set 1 consisted of 18 raised-line shapes, 9 symmetric and 9 asymmetric. The shapes appeared at the centre of a 10 cm613 cm card and measured approximately 6 cm66 cm. The raised lines were 1.3 mm wide60.5 mm high. Stimuli in set 2 consisted of raised surfaces. The surface inside the outline of each raised shape was elevated 1.3 mm. Set 3 comprised objects made of white polyethylene that were prepared from a solid piece approximately 6 cm long66 cm wide60.5 cm high. Finally, set 4 was formed from polyethylene objects prepared from a cubic piece approximately 6 cm66 cm66 cm. Each object was glued at the centre of a card to prevent participants from moving the object and changing its orientation during haptic exploration.

Visual and haptic discrimination of symmetry

319

2.1.3 Equipment. The apparatus used for stimulus presentation was a visual ^ haptic object tachistoscope with a piezoelectric board that acted as the stimulus-presentation platform. The board had a piezosensor located underneath at the centre of the platform, below the position at which the stimulus was presented. The tachistoscope was equipped with a liquid-crystal window located at eye level on the vertical panel facing the seated participant. This window allowed visual object presentation. The apparatus was interfaced with an IBM computer. Special software allowed data collection and provided a different random-order presentation of trials for each participant. 2.1.4 Design and procedure. The design was a 4 stimulus sets (raised-line shapes, raised surfaces, short objects, and tall objects)62 symmetry conditions (symmetric versus asymmetric) repeated-measures factorial design. Participants were tested individually in a quiet room. A sound from the computer alerted them that the liquid-crystal window would allow them to see the stimulus. On each trial, the experimenter presented the appropriate stimulus at the presentation board. All participants performed the symmetry-discrimination task with the four sets of materials. Set presentation order was counterbalanced across participants to avoid practice effects. Observers were asked to look at the stimulus through the tachistoscope window. Each stimulus was presented at the same orientation. The midaxis of the stimulus card was aligned to the body midaxis of the participant in the midtransverse plane. Participants were given examples of symmetric and asymmetric stimuli. On each trial the participant placed his or her two forefingers on the apparatus placeholder. When the fingers were raised from the holder, and following a random delay of a maximum of 1.5 s, the window allowed visual inspection of the stimulus for 250 ms. There was no fixation point. The visual angle subtended by the stimuli was approximately 10 deg by 10 deg. A voice key was used to stop the computer's internal clock. Before the experimental trials started, the participant performed six practice trials. Data from these trials were not included in the analysis. Latencies were recorded from the time that the liquid-crystal window allowed sight of the object until the vocal response. Participants were asked to respond as quickly and accurately as possible whether the stimulus was symmetric or asymmetric. 2.2 Results and discussion 2.2.1 Accuracy analysis. The data were analysed with the SPSS programme. Overall accuracy was high (94.48% mean correct). Symmetric stimuli were as accurate as asymmetric stimuli (94% and 95%, respectively). The visual system appeared to be equally accurate for all four stimulus sets. The ANOVA conducted on accuracy scores showed that neither the effect of symmetry (F1, 15 ˆ 0:28, p ˆ 0:6) nor the effect of stimulus set (F3, 45 ˆ 0:90, p ˆ 0:45) was statistically significant. The interaction between symmetry and stimulus set was also not significant (F 5 1). 2.2.2 Response latencies. The overall mean response time for correct trials was 1164 ms. Symmetric judgments took 13 ms longer than asymmetric judgments (1171 ms versus 1158 ms, respectively) but the difference was not significant (F 5 1). The effect of the stimulus set was statistically significant (F3, 45 ˆ 3:69, p 5 0:05). A posteriori comparisons showed that latencies for tall objects were longer than for raised-line shapes and raised surfaces (both ps 5 0:02)ösee figure 2. The interaction between symmetry and stimulus set was not significant (F 5 1). Unlike visual studies which showed an advantage for symmetric stimuli, our results for symmetric and asymmetric displays did not differ in terms of accuracy or in terms of response time. However, the participants took longer for 3-D objects than for 2-D raised-line and raised-shape sets.

320

S Ballesteros, J M Reales

1300

Response time=ms

1240

1180

1120

1060

1000 RL

RS

Stimulus set

SO

3-D objects

Figure 2. Mean response times as a function of stimulus set. Error bars show an estimate of the standard error of the mean (experiment 1). RL ˆ raised line, RS ˆ raised surface, SO ˆ short object.

3 Experiment 2 Experiment 2 was conducted to investigate the accuracy and latency performance of the haptic perceptual system in the symmetry-discrimination task under unimanual and bimanual conditions. According to previous findings (Ballesteros et al 1997), we expected that performance with symmetric displays would improve from 2-D shapes to 3-D objects. We also expected that the haptic system would be faster with 3-D objects than with 2-D shapes. 3.1 Method 3.1.1 Participants. Thirty-two new students from the same pool voluntarily participated in one experimental session of approximately 45 min. All participants were right-handed and explored the stimuli freely with their preferred hand or with both hands, according to the experimental condition. 3.1.2 Stimuli, equipment, and design. The stimuli and the equipment were the same as in experiment 1. The experimental design was a mixed-factorial design with 2 modes of exploration (unimanual versus bimanual)64 stimulus sets (raised-line shapes, raisedsurface shapes, short objects, and tall objects)62 symmetry conditions (symmetric versus asymmetric). Mode of exploration was a between-subjects variable, whereas stimulus set and symmetry condition were within-subjects variables. Sixteen participants were randomly assigned to each experimental condition. 3.1.3 Procedure. On each trial, the participant never saw the displays and held both forefingers on the two finger holders of the apparatus. The stimulus set was counterbalanced across participants. The computer program indicated to the experimenter which stimuli from number 1 to 18 had to be presented on each trial. The experimenter placed the randomly selected stimulus at the centre of the presentation platform and secured it with the clamps in the correct position. A tone from the computer alerted the participants that a stimulus was located on the platform. After hearing the starting signal, participants moved their preferred hand or both hands, according to the experimental condition, from the starting position to the haptic display. Latencies were automatically recorded from the time the hand made contact with the stimuli to the vocal response. The experimenter recorded responses by pressing the appropriate key on the computer keyboard. The experiment started with 6 practice trials that were not included in the analysis.

Visual and haptic discrimination of symmetry

321

3.2 Results and discussion 3.2.1 Accuracy analysis. Accuracy was higher when the stimuli were explored bimanually (93%) than when they were explored unimanually (84%). Asymmetric stimuli were judged more accurately than were symmetric stimuli under unimanual (89% versus 80%) as well as bimanual exploration (95% versus 91%). Performance was higher for asymmetric stimuli (91%) than for symmetric ones (86%)ösee figure 3. These results were confirmed by a 3-factor mixed ANOVA with mode of exploration (unimanual, bimanual) as between-subjects factor, and symmetry (symmetric, asymmetric stimuli) and stimulus set (raised-line shapes, raised surfaces, short objects, tall objects) as within-subjects factors. Touch was more accurate under bimanual exploration. The main effect of symmetry was significant (F1, 30 ˆ 11:791, p 5 0:01). Stimulus set was also significant (F3, 90 ˆ 3:13, p 5 0:03). A posteriori tests indicated that accuracy was higher for tall objects (91%) than for raised-line shapes (83%, p 5 0:04). There was a significant interaction between symmetry and stimulus set (F3, 90 ˆ 10:228, p 5 0:001). This interaction emerged because the extension of the stimuli along the third dimension improved performance with symmetric stimuli but not with asymmetric ones (all ps 5 0:01). No other interaction effect approached significance.

Mean accuracy=% correct

100

90

80

unimanual bimanual

70 symmetric

asymmetric

60 RL

RS SO 3-D objects Stimulus set

RL

RS SO 3-D objects Stimulus set

Figure 3. Mean accuracy for symmetric and asymmetric stimuli as a function of stimulus set under unimanual and bimanual exploration. Error bars show an estimate of the standard error of the mean.

3.2.2 Sensitivity and response bias. Sensitivity measures on the symmetry ^ asymmetrydiscrimination task under unimanual and bimanual conditions were computed from the individual symmetric and asymmetric judgments for each participant as a function of stimulus set and mode of exploration. Individual estimates were averaged. The results from this analysis are shown in table 1. The sensitivity, d 0, and bias, c, indexes were calculated from the z -score transformation of the mean proportion of hits, defined as saying ``symmetric'' to symmetric stimuli, and false-alarms, defined as saying ``symmetric'' to asymmetric stimuli. To calculate these parameters for each participant, we used the STD SP software (Reales and Ballesteros 1994) which has been recently revised to work in the Windows environment of IBM-compatible computers (Reales and Ballesteros 2000). Table 1 summarises the sensitivity, d 0, and bias, c, results for unimanual and bimanual conditions. An ANOVA was conducted on d 0 and c parameters corresponding to the four stimulus sets in both exploration conditions. The main effect of stimulus set was significant (F3, 90 ˆ 0:663, p 5 0:03) for the variable sensitivity. A posteriori comparisons 0 0 showed that raised-line shapes (davg ˆ 3:111) and raised surfaces (davg ˆ 3:131) did not

322

S Ballesteros, J M Reales

0 Table 1. Signal-detection analysis: sensitivity (davg ) and response bias (cavg ) in the symmetrydiscrimination task in unimanual and bimanual exploration conditions.

Raised-line stimuli Unimanual exploration 0 2.48 davg 0.53* cavg Bimanual exploration 0 3.74 davg 0.34* cavg

Raised surfaces

Short 3-D objects

3-D objects

2.82 0.26

3.12 0.14

3.19 ÿ0.09

3.44 0.19

3.84 0.39*

4.04 ÿ0.20

Note: * A bias toward an `asymmetric' response ( p 5 0:05). 0 differ from each other but both differed significantly from tall objects (davg ˆ 3:604; ps 5 0:03). The main effect of stimulus set for the response-bias data was significant (F3, 90 ˆ 6:536, p 5 0:001). A posteriori comparisons showed that the tall-object set (c ˆ ÿ0:055) differed from the other three stimulus sets (raised-line shapes, c ˆ 0:437; raised surfaces, c ˆ 0:230; short objects, c ˆ 0:266; all ps 5 0:03). The analysis showed that there was a bias to respond ``asymmetric'' to the raised-line shapes. Exploration condition was also significant (F1, 30 ˆ 6:854, p 5 0:02); bimanual exploration resulted in higher sensitivity than unimanual exploration. There was a response bias for raisedline shapes as well as less sensitivity, compared with 3-D objects. The interaction effects for both analyses were not significant (F 5 1).

3.2.3 Response-time analysis. It took less time to discriminate whether a stimulus was symmetric or asymmetric under bimanual (9.75 s) than under unimanual exploration (11.12 s). Statistical analysis confirmed this observation. The ANOVA on response times for correct responses, with mode of exploration as a between-subjects factor and stimulus set and symmetry as within-subjects factors, showed that the main effect of mode of exploration was highly significant (F1, 30 ˆ 195:786, p 5 0:0001). Bimanually exploring the displays was much faster than unimanual exploration (see figure 4). The main effect of symmetry was also significant; response time was shorter for asymmetric (10 s) than for symmetric (10.8 s) stimuli (F1, 30 ˆ 5:853, p 5 0:02). Stimulus set 17

symmetric

asymmetric

Response times=s

15 unimanual

13

bimanual

11 9 7 5 RL

RS SO 3-D objects Stimulus set

RL

RS SO 3-D objects Stimulus set

Figure 4. Mean response times for correct judgments corresponding to symmetric and asymmetric stimuli as a function of stimulus set and exploration condition (unimanual and bimanual). Error bars show an estimate of the standard error of the mean.

Visual and haptic discrimination of symmetry

323

was significant (F3, 90 ˆ 41:476, p 5 0:001). Observers were faster as the displays gained in height (13 s, 12.7 s, 8.7 s, and 7.3 s for raised-line shapes, raised surfaces, short objects, and tall objects, respectively). Subjects were slower for raised-line shapes, raised surfaces, and short objects than for tall objects (all ps 5 0:001). None of the interactions reached significance. One reviewer cogently pointed out the importance of considering stimulus complexity as the stimuli varied from 3 to 30 sides. Performance was superior for low-complexity stimuli (93.56% correct) than for medium (89.43% correct) and high-complexity (82.81% correct) stimuli. A 4-way ANOVA was conducted on the number of correct responses, with 3 levels of stimulus complexity (1) (low, medium, high), 4 stimulus sets, 2 symmetry conditions, and 2 modes of exploration. It showed a significant effect of stimulus complexity (F2, 12 ˆ 6:266, p 5 0:03). Paired comparisons indicated that accuracy was significantly higher for low-complexity stimuli than for high-complexity stimuli ( p 5 0:01). Medium-complexity stimuli also differed significantly from highcomplexity stimuli ( p 5 0:05). The difference between low and medium complexity was not significant ( p ˆ 0:2). The main effects of mode of exploration (F1, 12 ˆ 65:07, p 5 0:01) and symmetry (F1, 12 ˆ 6:169, p 5 0:03) were both significant. The effect of stimulus set did not reach significance ( p 4 0:05). As shown in previous analyses, accuracy was higher under bimanual exploration, and for asymmetric stimuli. The interaction mode of exploration6complexity was significant (F2, 12 ˆ 5:655, p 5 0:019). The interaction means that high-complexity stimuli were far more accurate under bimanual exploration compared with unimanual exploration than low-complexity and medium-complexity stimuli. The ANOVA on time scores for correct responses showed that the effect of complexity was significant (F2, 12 ˆ 16:502, p 5 0:001). High-complexity stimuli required longer exploration time (13.7 s) than medium-complexity stimuli (9.9 s) ( p 5 0:001), and low-complexity stimuli (8.1 s) ( p 5 0:001). The main effects of mode of exploration (F1, 12 ˆ 29:279, p 5 0:001), and stimulus set (F3, 36 ˆ 107:827, p 5 0:001) were both significant. The interaction between stimulus set and complexity was also significant (F2, 12 ˆ 16:502, p 5 0:001). Although more time was required at all the stimulus sets as the stimuli increased in complexity, differences were statistically significant for 2-D shapes (sets 1 and 2) but not for 3-D objects (sets 3 and 4). That is, level of complexity had an effect on the time required to discriminate raised-line shapes and raised surfaces ( ps 5 0:01) but not objects (short and tall). Moreover, the time required to discriminate low-complexity stimuli did not differ across stimulus sets but, as stimulus complexity increased, more time was required for 2-D shapes than for 3-D objects (F 5 1). 4 General discussion The present results are relevant for understanding symmetry discrimination. The question we asked was whether moving from 2-D shapes to 3-D objects with the same contour but extended in the z-axis influenced visual and haptic perception of symmetry. We briefly summarise the main findings and discuss four issues. First, both modalities, vision and active touch, allowed observers to extract accurate information about the bilateral symmetry of shapes and objects. Second, unimanual exploration produced more accurate perception of asymmetric than symmetric 2-D shapes. Third, bimanual exploration was more accurate and faster than unimanual exploration. Fourth, performance with symmetric stimuli improved from flat to 3-D (1) As each stimulus was presented only once, to achieve liability the 9 levels of complexities (one for each symmetric and asymmetric stimuli) were categorised in 3 levels of complexities corresponding to low (3, 4, and 6 sides), medium (8, 12, and 14 sides), and high (18, 24, and 30 sides) stimulus complexity. There were 3 symmetric and 3 asymmetric stimuli at each complexity level.

324

S Ballesteros, J M Reales

objects, but performance did not change with asymmetric stimuli. Fifth, while touch latencies decreased as the stimuli gained in extension in the third dimension, visual latencies increased. Sixth, stimulus complexity influenced performance, which was better for less complex than for more complex stimuli. 4.1 Touch as a fast and accurate information-processing system Klatzky and her colleagues (1985) showed that the haptic system is very accurate and fast in identifying familiar objects. We also found (Reales and Ballesteros 1999) that haptic explorers named familiar objects presented the first time in approximately 2 s but when the same objects were presented again later, after performing a distraction task, the time required for identification was significantly reduced (by 0.28 s). Here, as well as in earlier results (Ballesteros et al 1997, 1998; Locher and Simmons 1978), we show that touch is also accurate and fast in detecting bilateral symmetry, an important attribute of the spatial layout of tangible unfamiliar displays. These findings confirm and extend previous results by showing that touch is an accurate perceptual system in detecting this spatial property in shapes and objects. Although touch is quite sensitive in dealing with flat displays, it is far more accurate and faster with 3-D objects. As far as we know, the present study is the first to show that extension along the axis of the third dimension influences performance as a whole in terms of accuracy and response time. 4.2 Effects of the third dimension in the discrimination of symmetry Our previous studies demonstrated that although touch was moderately accurate in detecting bilateral symmetry of small raised-line displays (Ballesteros et al 1997, 1998), it was even more accurate in detecting this property in 3-D objects that were explored with the hands (Ballesteros et al 1997). However, as both sets were not comparable in any variable (eg size, shape, complexity) it was not possible to generalise across 2-D and 3-D stimuli. The present study was designed to overcome this shortcoming by constructing four sets of displays with exactly the same contours and different heights. Unlike our previous study, in which only accuracy was recorded, in the present experiments we also recorded response times. Results showed that haptic accuracy increased as the stimuli were extended in their third dimension, but for symmetric stimuli only. 0 Sensitivity assessed by the davg parameter also increased as a function of elongation but only for symmetric displays. On the contrary, response times decreased for both types of displays. 4.3 The reference-frame hypothesis: Bimanual versus unimanual exploration In this study, unimanual exploration was compared with bimanual performance. The results are in agreement with the reference-frame hypothesis. Exploration with both hands aligned to the midbody axis facilitated the discrimination of bilateral symmetry. Under active tactual exploration, asymmetric 2-D shapes were detected better because integration of shape information did not seem to be needed. However, adding spatial reference at the initial coding of the stimuli (eg fingers and hands in relation to the body midline) provides enough information to improve accuracy in detecting symmetric and asymmetric shapes and objects. An important result is that only for symmetric stimuli performance in both the unimanual and the bimanual conditions improved as a function of extending the stimuli in the third dimension. For asymmetric stimuli, however, approaching the task with two parallel hands produced better performance overall, but accuracy did not change as a function of extension in the third dimension. The results from the bimanual condition are in agreement with findings from a study conducted, in collaboration with Millar, with small raised open and closed shapes (Ballesteros et al 1998). In this study, however, we did not align the two forefingers to the body midaxis and did not centre the two forefingers directly above the stimulus

Visual and haptic discrimination of symmetry

325

card to provide body-centred spatial reference. Instead, participants explored stimuli freely with both hands. The only requirement for the subjects was to hold their two forefingers on the holders, parallel to the midbody axis, as the starting position on each trial. Better performance on this spatial task with 3-D objects is explained by the mode of exploration. Bimanual exploration allows one to relate hand locations to the body axis which provides an effective frame of reference for coding bilateral, redundant features of objects. This has been shown in a previous study in which mode of exploration was recorded and analysed, frame by frame (Ballesteros et al 1997). 4.4 Comparing vision and touch While for 2-D shapes vision was more accurate than touch, for 3-D objects accuracy was almost equal. These results confirmed previous findings: touch is highly effective not only in identification of familiar objects (eg Klatzky et al 1985), but also in symmetry discrimination with unfamiliar displays. Bilateral symmetry is considered a very salient property of visual shape (eg Locher and Nodine 1973; Mach 1959; Roger 1981). A recent study with dots as stimuli has shown that the symmetry of these types of visual displays is conveyed only for dots close to the vertical axis of symmetry (eg Dakin and Herbert 1998). Our stimuli, although unfamiliar, are found more frequently in the world. Vision was very accurate in performing the task with symmetric as well as with asymmetric stimuli. Our tactual experiment has shown that touch, as vision, was very accurate with symmetric and asymmetric 3-D objects. However, touch was less accurate than vision for 2-D shapes. This result confirmed previous findings (Ballesteros et al 1997; Locher and Simmons 1978): visual performance with both eyes was very accurate and similar for all types of displays. We found that touch and vision differed not only in accuracy, but also in latency. While performance in vision was faster with flat stimuli than with 3-D objects, an opposite trend was found in touch. The increase in the number of edges and angles of the stimuli made the detection of visual stimuli more complex (eg Attneave 1957; Chipman 1977; Day 1967). The different results observed in touch might be explained by the influence of the motor component of active tactual exploration. This motor component enhances the capabilities of the hands when exploring objects (Lederman and Klatzky 1987). In this study, participants' hands first manipulated objects globally. It has been demonstrated that the initial movements consisted of moulding the hands and fingers to enclose as much as possible of the contour of the object. This very efficient and fast movement is performed to explore familiar objects (Lederman and Klatzky 1987) as well as unfamiliar objects (Ballesteros et al 1997). This movement allows the capture of global spatial information from the parts of the object from the beginning of exploration. The second movement performed in the symmetry-detection task was contour following, to explore the edges of the object (Ballesteros et al 1997). Touch is a `clever' perceptual system when dealing with objects, because the motor information enhances the sensory subsystem. However, it is much less effective in exploring raised-line displays, owing to the limited use of the motor component, thereby losing part of its encoding capacity. Two types of explanation have been proposed to justify the advantage of bilateral symmetry in vision. An early justification for the advantage of this type of symmetry over horizontal and oblique symmetry relies on the bilateral organisation of the brain (eg Corballis and Roldan 1975). According to this explanation, bilateral symmetry does not require mental rotation. More recently, however, Pashler (1990) proposed a reference-frame hypothesis to explain results showing that precuing the relevant spatial axis produced an advantage for the detection of symmetry. According to Pashler (1990),

326

S Ballesteros, J M Reales

in vision the observer adjusts his/her internal reference frame so that the top ^ bottom axis for object recognition coincides with the expected axis of symmetry. Other studies conducted with complex visual stimuli have also shown the advantage of bilateral symmetry (Locher and Wagemans 1993; Wagemans 1995) and favoured the hypothesis that this type of symmetry is detected preattentively. At a first stage, visual symmetry is detected without attention; while at a second stage, cognitively controlled attention processes contribute to task performance (van der Helm and Leeuwenberg 1996; Wagemans 1995). In conclusion, manipulating stimulus height, while keeping shape, size, and complexity constant, significantly improved observers' haptic performance with symmetric stimuli. Furthermore, bimanual exploration produced an advantage over unimanual exploration for both symmetric and asymmetric stimuli in accuracy and response times, supporting the reference-frame hypothesis. Exploration with both hands parallel to the midbody axis facilitated the detection of bilateral symmetry. Moreover, presenting exactly the same stimuli to vision and touch has shown that both modalities are almost equally accurate in detecting the symmetry ^ asymmetry of the stimuli (especially with 3-D objects). However, vision is much faster. Acknowledgments. This study was supported by a UNED (2000) Strategic Research Grant. Portions of this article were presented at the 42nd (2001) Annual Meeting of the Psychonomic Society, Orlando, Florida. We thank the Spanish Organisation for the Blind (ONCE) for preparing the material used in the experiments. We also thank Morton Heller for helpful comments on this article. References Appelle S, 1991 ``Haptic perception of form: Activity and stimulus attributes'', in The Psychology of Touch Eds M A Heller, W Schiff (Hillsdale, NJ: Lawrence Erlbaum Associates) pp 169 ^ 188 Attneave F, 1957 ``Physical determinants of the judged complexity of shapes'' Journal of Experimental Psychology 53 221 ^ 227 Ballesteros S, Manga D, Reales J M, 1997 ``Haptic discrimination of bilateral symmetry in 2-dimensional and 3-dimensional unfamiliar displays'' Perception & Psychophysics 59 37 ^ 50 Ballesteros S, Millar S, Reales J M, 1998 ``Symmetry in haptic and in visual shape perception'' Perception & Psychophysics 60 389 ^ 404 Ballesteros S, Reales J M, Manga D, 1999 ``Implicit and explicit memory for familiar and novel objects presented to touch'' Psicothema 11 785 ^ 800 Barlow H B, Reeves B C, 1979 ``The versatility and absolute efficiency of detecting mirror symmetry in random dot displays'' Vision Research 19 783 ^ 793 Bornstein M H, Ferninandson K, Gross C G, 1981 ``Perception of symmetry in infancy'' Developmental Psychology 17 82 ^ 86 Chipman S F, 1977 ``Complexity and structure in visual patterns'' Journal of Experimental Psychology: General 106 269 ^ 301 Corballis M C, Roldan C E, 1975 ``Detection of symmetry as a function of angular orientation'' Journal of Experimental Psychology: Human Perception and Performance 1 221 ^ 230 Dakin S C, Herbert A M, 1998 ``The spatial region of integration for visual symmetry detection'' Proceedings of the Royal Society of London, Series B 265 659 ^ 664 Day H, 1967 ``Evaluations of subjective complexity, pleasingness, and interestingness for a series of polygons varying in complexity'' Perception & Psychophysics 2 282 ^ 286 Helm P A van der, Leeuwenberg E L-J, 1996 ``Goodness of visual regularities: A nontransformational approach'' Psychological Review 193 429 ^ 456 Horridge G A, 1996 ``The honeybee (Apis melliffera) detects bilateral symmetry and discriminates its axis'' Journal of Insect Physiology 42 755 ^ 764 Julesz B, 1971 Foundations of Cyclopean Perception (Chicago, IL: University of Chicago Press) Klatzky R L, Lederman S J, Metzger V A, 1985 ``Identifying objects by touch: An `expert system' '' Perception & Psychophysics 37 299 ^ 302 Klatzky R L, Lederman S J, Reed C, 1987 ``There's more to touch than meets the eye: The salience of objects attributes for haptics with and without vision'' Journal of Experimental Psychology: General 116 356 ^ 369 Klatzky R L, Loomis J M, Lederman S J, Wake H, Fujita N, 1993 ``Haptic identification of objects and their depictions'' Perception & Psychophysics 54 170 ^ 178

Visual and haptic discrimination of symmetry

327

Lederman S J, Klatzky R R, 1987 ``Hand-movements: A window into haptic object recognition'' Cognitive Psychology 19 342 ^ 368 Lehrer M, Horridge G A, Zhang S W, Gadagkar R, 1995 ``Pattern vision in bees: Preference for radial patterns'' Philosophical Transactions of the Royal Society of London, Series B 347 123 ^ 137 Locher P J, Nodine C F, 1973 ``Influence of stimulus symmetry on visual scanning patterns'' Perception & Psychophysics 13 408 ^ 412 Locher P J, Simmons R W, 1978 ``Influence of stimulus symmetry and complexity upon haptic scanning strategies during detection, learning, and recognition tasks'' Perception & Psychophysics 32 110 ^ 116 Locher P J, Wagemans R W, 1993 ``Effects of element type and spatial grouping on symmetry detection'' Perception 22 565 ^ 587 Loomis J, Lederman S J, 1986 ``Tactual perception'', in Handbook of Perception and Human Performance volume 2, Eds K R Boff, L Kaufman, J P Thomas (New York: Wiley) pp 31-1 ^ 31-44 Mach E, 1959 The Analysis of Sensations (Chicago, IL: Open Court, originally published in 1887) Millar S, 1978 ``Aspects of memory for information from touch and movement'', in Active Touch: The Mechanism of Recognition of Objects by Manipulation: A Multidisciplinary Approach Ed. G Gordon (Oxford: Pergamon Press) pp 215 ^ 227 Millar S, 1994 Understanding and Representing Space: Theory and Evidence from Studies with Blind and Sighted Children (Oxford: Oxford University Press) Palmer S E, 1989 ``Reference frames in the perception of shape and orientation'', in Object Perception: Structure and Processes Eds B S Shepp, S Ballesteros (Hillsdale, NJ: Lawrence Erlbaum Associates) pp 121 ^ 163 Palmer S E, Hemenway K, 1978 ``Orientation and symmetry'' Journal of Experimental Psychology: Human Perception and Performance 4 691 ^ 702 Pashler H, 1990 ``Coordinate frame for symmetry detection and object recognition'' Journal of Experimental Psychology: Human Perception and Performance 16 150 ^ 163 Reales J M, Ballesteros S, 1994 ``STD SP, a program in Pascal for computing parameters and significance tests from several detection theory designs'' Behavior Research Methods, Instruments & Computers 26 151 ^ 155 Reales J M, Ballesteros S, 1999 ``Implicit and explicit memory for visual and haptic objects: Cross-modal priming depends on structural descriptions'' Journal of Experimental Psychology: Learning, Memory, and Cognition 25 644 ^ 663 Reales J M, Ballesteros S, 2000 ``TDS EXPER''A Computer Programme for Signal Detection Theory (Madrid: Universitas) Rock I, 1983 The Logic of Perception (Cambridge, MA: MIT Press) Royer F L, 1981 ``Detection of symmetry'' Journal of Experimental Psychology: Human Perception and Performance 7 1186 ^ 1210 Simmons R W, Locher P J, 1979 ``The role of extended perceptual experience upon haptic perception of nonrepresentational shapes'' Perception & Psychophysics 48 987 ^ 991 Wagemans J, 1995 ``Detection of visual symmetries'' Spatial Vision 9 9 ^ 32 Wagemans J, van Gool L, D'Ydewalle G, 1992 ``Orientation effects and components processes in symmetry detection'' Quarterly Journal of Experimental Psychology A 44 475 ^ 508

ß 2004 a Pion publication

ISSN 0301-0066 (print)

ISSN 1468-4233 (electronic)

www.perceptionweb.com

Conditions of use. This article may be downloaded from the Perception website for personal research by members of subscribing organisations. Authors are entitled to distribute their own article (in printed form or by e-mail) to up to 50 people. This PDF may not be placed on any website (or other online distribution system) without permission of the publisher.