ERGONOMICS, 2001,

VOL.

44,

NO.

3, 328 338

A study of the discriminability of shape symbols by the foot

A. J. COURTNEY* and H. M. CHOW
Department of Industrial and Manufacturing Systems Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong SAR, China Keywords: Symbol identi cation; blindness; cutaneous sense; feet; kinesthesis; landmarks. An experiment to test the discriminability of shape symbols using the shod foot was performed with 38 blind people (aged 23 72 years). Ten shape symbols which were 5 mm thick and tted into a 30.5 cm2 tile were presented to subjects to identify by using only their feet. Each subject had 20 trials in which to discriminate the symbols. In each trial, a symbol was selected randomly and presented to the subject in randomized orientation. The subject was instructed to step on the symbol and to identify it using their own method. Time to discriminate a symbol and the accuracy of identi cation were recorded. A very high accuracy (93% on average) was obtained, which is comparable to the accuracy of tactile symbol discrimination using the hands. Average time to discriminate a symbol was 16 s with a standard deviation of 12.15 s, which indicated the high variability of the results. Owing to the high accuracy of identi cation, tactile foot-discriminable symbols have great potential as landmarks for blind people and if applied to a tactile guide path they could provide information for orientation and navigation.

1. Introduction Sometimes sighted people may not be fully aware of the importance of the sense of touch because vision will often dominate other senses in, for example, the perception of shape. Sighted people seem to attend to vision more than to the sense of touch because the hands are assumed to be performing organs rather than sensing organs (Gibson 1962, 1966, Heller and SchiV 1991). Obviously, tactual sense is important for blind people and is necessary in their daily life for reading Braille, in their use of a cane to detect obstacles, in recognizing objects by using their hands, and so on. Their blindness means that they must rely on the sense of touch as an important channel for exploring the world. One application of the sense of touch is in tactile guide paths which can greatly assist blind people to follow a particular route by using their shod feet to detect and distinguish raised patterns. Two kinds of tiles ( gure 1) are generally used to form a guide path, a raised strip pattern and a dot pattern, as for example, de ned in Australian Standards (1992). Two meanings are imposed on these two kinds of tiles. The strip pattern is used to indicate the direction of travel to be taken and the dot pattern is used to indicate a potential hazard such as a change of direction, location of a junction point or the position of stairs. Although tactile

*Author for correspondence. e-mail: acour@hkucc.hk

Ergonomics ISSN 0014-013 9 print/ISSN 1366-584 7 online 2001 Taylor & Francis Ltd http://www.tandf.co.uk/journals DOI: 10.1080/0014013001001208 9

Discriminability of foot symbols

329

(a)

(b)

(c)

Figure 1.

The tactile guide path showing tile patterns: (a) a tile with strip pattern; (b) a tile with dot pattern; (c) a junction point of the tactile guide path.

guide paths are already commercialized, research about the detection capabilities of users does not seem to have been undertaken. Almost all of the research available on tactual discrimination has used human hands for the purpose of discriminating symbols, forms, letters or objects. For example Austin and Sleight (1952a, b) investigated accuracy, speed and related factor for tactual discrimination; Nolan and Morris (1971) and Gill and James (1973) examined the discriminability of some tactual symbols. Gibson (1963) and Davidson (1972) studied the movements of hands and ngers when people were exploring an object actively by touch. Similar experiments for the feet do not seem to be available but it is likely that the mechanism of tactual perception for the hands and feet are similar with haptic touch being involved for both.

330

A. J. Courtney and H. M. Chow

Haptic touch incorporates cutaneous sensations and kinesthesis (Revesz 1950). Tactual perception of texture, roughness, shape and information of spatial localization can be generated by combining the information from these two kinds of sense information together, so much so that the contribution of cutaneous sense and kinesthesis in tactual perception cannot be separated clearly. Performance using touch is increased if people are allowed to explore objects actively. Gibson (1962) and Heller and Myers (1983) reported that the recognition accuracy was increased progressively with the degree of movement allowed between object and perceiver, and high accuracy of form recognition (95% ) was obtained when `active exploration was allowed. Similar results were obtained by Heller and Myers (1983) and by Heller (1984) . The practical nature of the experiment reported here meant that it was not possible to control many of the variables involved. The purpose of this study was to test discriminability of a range of symbols using shod feet and to look at the strategies used during the discrimination process. The test was the rst stage in nding a set of symbols for discrimination by the foot that would not be confused easily and which may be used to provide orientation and navigation information for blind people using guide paths. 2. Method 2.1. Subjects Thirty-eight blind subjects (23 males and 15 females) aged from 23 72 years volunteered to take part in the study. Characteristics of the subjects are summarized in table 1. All of them were recruited at the Hong Kong Society for the Blind, Rehabilitation Training Centre (HKSB-RTC). Therefore, many of them were learning mobility, Braille, cookery or other skills necessary for their daily life in the Centre. The subjects had no experience or only limited experience of tactile graphs or maps. However, almost all of the subjects had some experience with tactile guide paths because the underground Mass Transit Railway (MTR) stations in Hong Kong have tactile guide paths installed. The only subjects who could read Braille well were the early blind; the others read very slowly or could not read it at all. People who are early blind were not born blind but lost their sight at an early age. The late blind are people who remember the experience of vision in their life. 2.2. Apparatus The 10 symbols shown in gure 2 were embossed on a 22 cm diameter cardboard disk. The symbols tted approximatel y into a 3 cm square and were 0.15 cm high,

Figure 2.

The embossed symbols on the cardboard disk.

Discriminability of foot symbols

Table 1. Subject no. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 **29 30 *31 32 33 34 35 **36 37 *38 Gender F M M M F M F M M M M M F M M F M F M F M M F M M M M F M M F F F M M F F F Age (years) 35 62 56 25 52 48 50 45 60 40 23 40 26 60 25 30 55 40 35 64 52 46 50 57 52 72 36 50 36 32 40 35 40 40 40 50 40 50 Characteristics of the blind subjects. Early blind Yes No No Yes No No No No No No No No Yes No No Yes No Yes Yes No No No No No Yes No No No No Yes No No No No No No No No Degree of blindness LP TD TD LP LP TV HM HM LP LP LP LP TD LP HM LP TD LP LP TD HM LP TD TD TD TV TD LP HM TD LP LP HM LP LP TV TD LP Cause Born without eyes Retinal detachment Diabetes Optic nerve degeneration RP RP RP Glaucoma RP RP Optic nerve degeneration RP Cataract RP Optic nerve degeneration Developmental defect RP Cataract RP Parasitic disease RP RP Stevens Johnson Syndrome (allergic drug reactions) Retinal infection Allergic drug reactions RP Chemical injury RP Diabetes Cataract Diabetes RP RP RP Penetrating injury RP Optic nerve degeneration RP

331

Education S P S S S S S P P P S S S T S S T S S P P T S S P P S S T S S S S S P S P

F = female; M = male; LP = light perception; HM = can detect hand movement; TV = tunnel vision; TD = total darkness; RP = retinitis pigmentosa; P = primary education; S = secondary education. *The experiment was not completed. **The subject did not complete the experiment.

which was equal to the thickness of the cardboard background that they were on. This arrangement allowed the subjects to hold the cardboard and familiarize themselves with the symbols by hand. The same set of symbols was used for discrimination by foot, but they were much larger than the symbols on the cardboard. The foot symbols were of synthetic rubber sheeting which was cut to the shapes required and stuck onto 30.5 cm2 (1 ft2 ) plastic

332

A. J. Courtney and H. M. Chow

oor tiles. The symbols were approximately 18 cm square and 0.5 cm thick as shown in gure 3. Eight of the oor tiles were stuck on the oor to form a frame around a centre stimulus tile which could be removed. Therefore only one foot symbol at a time could be placed in the centre for the experiment. 2.3. Experimental procedures The 38 subjects were tested individually. First, the experimental procedure was explained to the subjects by the experimenter. After signing a consent form, their age, gender, educational background, cause of their blindness, speed of reading Braille and experience with tactile symbols or maps or tactile guide paths were recorded. If the subjects had residual vision or tunnel vision, they were blindfolded. Next the subjects were given 5 min for familiarization by using the hands to feel the tactile symbols that were embossed on cardboard. After familiarization with the symbols, the subjects were assisted to stand on each of the foot symbols on the oor. Subjects used their own method to discriminate the diVerent enlarged symbols using only their feet. During this initial contact with the symbols the subjects were asked to describe the symbols verbally and feedback was given by the experimenter to assist the learning process. After this initial discrimination process, the test was started and videotaped. The experimenter selected a symbol randomly and placed it on the oor in front of a subject. Orientation of the symbols was also randomized. The subject was requested to step on and identify the symbol using only their feet. When the subject identi ed the symbol, he/she stepped backwards. The experimenter recorded the time to identify the symbol, which was the time from the initial step onto the symbol until the backward step. The subject then identi ed the symbol and described it to the experimenter by pointing to the tactile symbols on the cardboard. The accuracy was recorded but no feedback was given. Each subject had 20 trials in total in which to discriminate between the symbols. The subjects were interviewed to elicit the methods that they used, how they felt about the process and the nature of guide path symbols in general.

cro

ddl

dot

cir

pi

tri

str

dl

div

Figure 3.

The stimuli on oorboards.

Discriminability of foot symbols

333

3. Results Two of the subjects (31 and 38 in table 1) could not complete the experiment because they were in a hurry and they did not have enough time to complete the whole experiment. Nevertheless, their results were still included. Two subjects (29 and 36 in table 1) withdrew from the experiment because they found that the discrimination process was very diYcult for them even after the learning process. Thirty-four out of 38 subjects completed the experiment. There were 20 trials to discriminate 10 symbols. The symbols were chosen randomly, therefore a symbol may not be shown at all, or could be repeated or could appear sequentially in the 20 trials. However, many of the subjects tried each symbol twice in the 20 trials. Table 2 shows the results of the experiment. The total number of discriminations for each symbol was about 70 for the 34 subjects. The accuracy (number of correct discriminations over total trials) for the symbols ranged from 85 to 99% and the overall accuracy was 93% . There were 8 out of 10 symbols with an accuracy of discrimination of more than 90% . Average correct discrimination time ranged from 9.75 s for the symbol `cir to 19.66 s for `cro and the overall discrimination time was 16 s. The standard deviations were high as expected for such an experiment. The standard deviations of discrimination time for each symbol and for the subjects were also high. The symbol `cro had the lowest accuracy (85% ) and the longest discrimination time (19.66 s). `Dot had the highest accuracy (99% ). t-tests were performed to compare the discrimination time for each symbol with the overall discrimination time. The discrimination times for `cir (9.75 s t35 = 7.113, p < 0.001) and `dot (10.37 s, t35 = 3.551, p < 0.01) were signi cantly faster than times for the other symbols. The correlations between the characteristics of the subjects and the discrimination accuracy or time were not signi cant. The detailed information is shown in table 3. Within the total of 702 trials, there were 39 trials where the symbols were confused. The confusion matrix for the symbols is shown in table 4. About 90% of errors in this experiment involved the three symbols `cro (11 errors) `pi (8 errors) and `div (7 errors). The symbol `cro was confused with `div (6 errors) and `y (4

Table 2.

Result of the experiment, by symbols.

Total Symbols trials dot y tri dl str ddl cir div pi cro Overall 69 72 71 62 69 65 77 79 67 71 702

No. of Average No. of No. of cannot be correct S.D. of correct incorrect discriminated discrimination discrimination trials trials trials time (s) time Accuracy 68 70 69 60 66 61 71 71 57 60 652 1 2 0 2 2 3 3 7 8 11 39 0 0 2 0 1 1 4 1 2 0 11 10.37 16.54 15.99 19.45 17.15 17.82 9.75 16.30 19.51 19.66 16.06 9.30 11.68 12.07 11.97 11.46 13.92 6.09 10.44 14.85 13.33 12.15 0.99 0.97 0.97 0.97 0.96 0.94 0.92 0.90 0.85 0.85 0.93

334
Table 3.

A. J. Courtney and H. M. Chow

The correlation between characteristics of the subjects and discrimination accuracy and time. *Age Accuracy Time Gender Accuracy F(1,3) = 0.412 0.525 Time **Education Accuracy Time Early/late blindness Accuracy Time

F value p

F(5,30) = F(1,3) = 0.412 1.137 0.655 0.362

F(1,3) = F(2,32) = F(2,32) = F(1,33) = F(1,33) = 0.084 0.202 0.497 0.110 0.211 0.773 0.818 0.613 0.743 0.649 60

*The subjects were grouped into six age ranges: < 20, 20 29, 30 39, 40 49, 50 59 and years. **Three groups: primary, secondary or tertiary education.

Table 4.

Confusion matrix: number of times that symbols in horizontal rows were erroneously selected for stimuli in vertical columns. Answer given

Symbol shown (no. of trials) cro (71) pi (69) div (62) cir (69) ddl (72) dl (77) str (65) y (79) dot (67) tri (71)

div 6 1 1 1 1 1

y 4 4 1

dot 1 3 1 1

cro 1 2

ddl 1 1 1

tri 2 1

dl

cir

pi

str

Total 11 8 7 3 3 2 2 2 1 0

1 1 2

Total no. of confusions is 39.

errors). Also, `pi and `y were confused (4 errors) and `div and `dot were confused (3 errors). Most of the errors were made because the subjects did not detect some components in a symbol or perhaps because the separations between the components were not large enough. These phenomena are illustrated in gure 4. On the other hand `tri was never confused with another symbol and `dot was only confused once with symbol `div. The symbols `cir, `pi and `str were never given as wrong answers. 4. Discussion The practical nature of the experiment meant that no control was exercised over the type of shoes worn by the subjects although it is very likely that the type of shoe would aVect the cutaneous sense for the feet. The subjects came along in the shoes that they normally wore and no special attention was paid to footwear. Nevertheless, a very high discrimination accuracy (93% overall) was obtained in this experiment. There does not seem to be any other research available on the capability of blind or sighted people in identifying any discriminate symbols using the shod feet, therefore the results cannot be compared with any previous data. The accuracy obtained here however can be compared with capability to discriminate symbols using the hands. Heller (1989) tested tactual-tactua l matches for 10 embossed shapes for 22 early and late blind subjects obtaining an accuracy of 84.5% . Austin and Sleight (1952a) using

Discriminability of foot symbols

335

Figure 4.

Examples of errors made in identifying symbols.

16 sighted subjects to discriminate 43 gures and geometric forms found that for 35 of them accuracy exceeded 90% and average accuracy was 85% . For active exploration of objects, similar results have been obtained by a number of researchers with accuracies of about 95% (Heller and Myers 1983, Heller 1984, Klatzky et al. 1985, Heller and SchiV 1991). There are several factors that may have contributed to the high accuracy achieved in the present study for the foot when compared to previous studies for the hand. In previous experiments the subjects were not allowed to practice. Here, before the test, the subjects practiced with each symbol once and feedback was given. Some learning on how to discriminate the symbols could occur therefore during the

336

A. J. Courtney and H. M. Chow

Positioning
O i ent at i on r Moving the foot over a large area

Figure 5.

Two strategies in the foot symbol discrimination process.

practice trial, thereby improving subsequent performance. This eVect was shown in an experiment by Austin and Sleight (1952b), who found that the accuracy of tactile discrimination for the hands would increase with an increase in the number of trials. The mean accuracy increased from 87.6% (trial 1) to 100% (trial 7). OVsetting the bene cial eVect of practice to some extent was the fact that the symbols or forms presented to the subjects in previous experiments were in a xed orientation, but here the symbols were presented in random orientation thereby increasing diYculty for the subjects. Moreover, the symbols tested in the present experiment were speci cally designed by the experimenters. Owing to the low sensitivity of the shod foot, simple con gurations were used. The simplicity may have contributed to the high accuracy obtained. The strategies for identifying symbols were expected to be complex and diYcult to isolate clearly into categories because of overlap. By the observation of the experimenter, from feedback given by the subjects after the experiment and reviewing the video record, the strategies used to discriminate the tactile symbols by foot appeared to be as shown in gure 5. At rst, the subject detected the position of diVerent components of the symbols roughly by covering a large area with the sole of their shoe. In this way, they could get an idea of the area that needed to be explored. Then two strategies seemed to be used to explore the shape of the symbols. Here, these strategies will be called `scanning and `stepping . Scanning was a method where the subjects moved one foot to press against and follow the edges of the symbol. If the components of the symbols were not continuous, they were scanned one by one. Then, the shape of the symbols was presumably generated by gathering information from the various foot movements involved. Kinesthesis must have played an important role and contributed to detection of lines in symbols such as `str and to arcs in symbols such as `cir. This is probably the reason why the subject can discriminate the symbols even when a pair of shoes with thick soles was worn.

Discriminability of foot symbols

337

`Stepping was a method where the subjects stepped repeatedly on a symbol and moved their weight from one foot to another, increasing and decreasing the pressure on the sole of their shoe. The shape of the symbol can be gathered in this way from cutaneous information from the sole of the foot. Sometimes, the subject would move round the symbol and step on the symbol from diVerent orientations in order to verify the shape that they were sensing. Figure 4 shows how the more complex symbols could be confused with their subset symbols. The subjects missed detecting some components of the symbols. This occurs perhaps because tactile perception is a sequential process and shape is generated by an accumulation of sensations. The subjects probably only detected some components of a symbol and then did not continue scanning and stepping the shape, thereby confusing it with other symbols. As a result of comments by the subjects, some general recommendations for designing distinguishable foot symbols are suggested here. The symbols should be simple with enough space between the components to allow error-free identi cation. The area of a symbol should be smaller than the area that the foot can easily cover. Symmetrical symbols permit easy veri cation by comparison between sensations from the left and right feet. Dots can be identi ed very rapidly but also they are easy to miss. To reduce the chance of the user making mistakes, symbols with joined components are preferable to symbols with discrete components. Thin lines and small dots may increase cutaneous sensations because of the increased pressure on the soles due to the smaller areas stimulated. Symbols presented to the user in a xed orientation should reduce cognitive demand. This may be achieved by use of an orientation symbol. There was large variation in the results found here. For example, the fastest mean recognition time for a subject was 3.3 s with a 100% accuracy. In contrast, two subjects found that the symbols were extremely diYcult to discriminate and withdrew from the experiment. Also, the standard deviations for correct discriminations were large (table 2). One factor that may have contributed to between-subject diVerences was that no control was exercised over the shoes worn by the subjects. DiVerent thickness and material of shoe soles and heels would aVect cutaneous sensations to some extent. However, a more important reason may be the diVerences between the blind individuals. Heller and SchiV (1991), for example, said the large diVerences between individuals is a common diYculty for researchers working with blind subjects. This indicates the need for appropriate training, matched to individual needs in order to increase the accuracy and speed of discrimination for the foot symbols, and to reduce the variations in discrimination time and reduce errors. Thin-soled shoes may have helped with discrimination of the symbols but it seems unrealistic to expect the users of guide paths to wear speci c footwear given the other practical considerations governing the choice of footwear. The symbols may be more easily discriminated if the patterns are raised more but then there is likely to be an increased risk of tripping for everyone walking over the symbols. A trade-oV between discriminability, symbol height, type of footwear and the risk of tripping will be necessary. 5. Conclusions This experiment was an initial attempt to investigate the discriminability of foot symbols for the blind. The subjects showed their capability, whilst wearing shoes, to discriminate randomly orientated foot symbols. An unexpected high accuracy (93% )

338

A. J. Courtney and H. M. Chow

was obtained and average correct discrimination time was 16 s. How to improve this capability is not yet known. Further study is necessary on the eVect on performance factors of, for example, the design of the foot symbols, the eVect of shoe type in the discrimination process and the eVect of training. However, the high accuracy obtained has demonstrated that it is feasible to use foot symbols in tactile guide paths to act as landmarks or provide other information for blind people in order to facilitate independent mobility and improve user-facility interaction. Acknowledgements The authors would like to thank the Rehabilitation and Training Centre at the Hong Kong Society for the Blind for their assistance in the recruitment of subjects. The study was supported by the Committee on Research and Conference Grants fund of the University of Hong Kong. References
AUSTIN, T. R. and SLEIGHT, R. B. 1952a, Accuracy of tactual discrimination of letters, numerals, and geometric forms. Journal of Experimental Psychology, 43, 239 247. AUSTIN, T. R. and SLEIGHT, R. B. 1952b, Factors related to speed and accuracy of tactual discrimination, Journal of Experimental Psychology, 44, 283 287. AUSTRALIAN STANDARDS 1992, AS 1424.4. Design for Access and Mobility, Part 4: Tactile Ground Surface Indicators for the Orientation of People with Vision Impairment (Sydney, NSW: Standards Australia). DAVIDSON, P. W. 1972, The role of exploratory activity in haptic perception: some issues, data, and hypotheses, American Foundation for the Blind Research Bulletin, 24, 21 27. GIBSON, J. J. 1962, Observations on active touch, Psychological Review, 69, 477 491. GIBSON, J. J. 1963, The useful dimensions of sensitivity, American Psychologist, 18, 1 15. GIBSON, J. J. 1966, The Senses Considered as Perceptual Systems (Boston, MA: Houghton MiZin). GILL, J. M. and JAMES, G. A. 1973, A study on the discriminability of tactual point symbols, American Foundation for the Blind Research Bulletin, 26, 19 34. HELLER, M. A. 1984, Active and passive touch: the in uence of exploration time on form recognition, Journal of General Psychology, 110, 243 249. HELLER, M. A. 1989, Picture and pattern perception in the sighted and blind: the advantage of the late blind, Perception, 18, 379 389. HELLER, M. A. and MYERS , D. S. 1983, Active and passive tactual recognition of form, Journal of General Psychology, 108, 225 229. HELLER, M. A. and SCHIFF, W. 1991, The Psychology of Touch (Hillsdale, NJ: Lawrence Erlbaum). KLATZKY , R. L., LEDERMAN, S. J. and METZGER, V. A. 1985, Identifying objects by touch: an expert system, Perception & Psychophysics, 37, 299 302. NOLAN, C. Y. and MORRIS , J. E. 1971, Improvement of Tactile Symbols for Blind Children (Louisville, KY: American Printing House for the Blind). REVESZ, G. 1950, Psychology and Art of the Blind (London: Longmans, Green & Co.).