#AUTHOR: Foulke, Emerson
#TITLE: Transfer of a Complex Perceptual Skill
#CATEGORY: Perception
#PUBLICATION: Perceptual and Motor Skills, 1964, 18, 733-740
#ABSTRACT: Communication systems that depend upon stimulation of the skin will be more flexible and useful if it is possible to shift to new sets of loci in accordance with the needs of particular situations.
TRANSFER OF A COMPLEX PERCEPTUAL SKILL
EMERSON FOULKE
University of Louisville
Summary: Communication systems that depend upon stimulation of the skin will be more flexible and useful if it is possible to shift to new sets of loci in accordance with the needs of particular situations. However, the feasibility of making such shifts will depend upon the amount of transfer that can be expected. To explore this problem, an experiment was performed in which braille readers served as Ss. They were required to read lines of braille characters with each of eight fingers. Performance was best when the forefingers were used and fell off sharply as the little fingers were approached. Explanations of the results in terms of anatomical, physiological and experiential factors were discussed. Some implications for cutaneous communication systems in general were suggested.
For some time there has been an interest in exploiting the sensing capabilities of the skin for communicative purposes. The skin is sensitive to a great variety of stimuli, some of which can be controlled and varied with sufficient ease to recommend their use in a communication system. Geldard (1957) has reported the successful demonstration of a code based on vibratory stimuli. Gilmer (1961) and Hawkes (1959) have both advocated communication by electrical stimulation of the skin and have carried out a good deal of research prerequisite to such an attempt. This writer is now conducting a research project in which Ss are taught an alphabet formed from electrical stimuli that vary in duration, intensity, and locus. Training is to be continued over a long period of time so that it will be possible to gauge the upper limits of such a system with respect to speed and accuracy. At the present time, our best S can receive 18 words per minute of unfamiliar prose.
One problem that may arise in communication systems that depend upon the skin is that of transfer. As compared with other sense organs, the skin provides an enormous area sensitive to stimulation. It is not all equally sensitive or accessible. However, there is enough flexibility to permit considerable leeway in the choice of loci for the application of electrodes or vibrators. The particular set of loci chosen will depend in part upon the circumstances under which communication is to take place, and they may vary from situation to situation. This raises the question of transfer. When a cutaneous code is learned using a given set of loci, how much of this learning will transfer when stimuli are applied at a different set of loci?
A related problem has to do with the relationships that must be preserved in moving to a new set of loci. Undoubtedly, spatial relationships among loci must be preserved to a considerable degree, but is it necessary to preserve absolute distances or relative distances, or only relative positions?
Answers to these questions are not yet available. Ss now being trained to receive electro-cutaneous codes may be able to provide some answers before too long. However, there is another group of people who have had extensive experience with a different kind of cutaneous communication. The blind have put braille to good use for more than a century. Although the tips of all four fingers on each hand could be brought into a proper relationship with the braille page for reading, the index fingers are used almost exclusively, and one index finger customarily assumes a greater sensing burden than the other (Fertsch, 1946). The remaining fingertips, though available, go unused. The report that follows de scribes an experiment to determine the transfer of braille reading ability to normally unused fingers.
METHOD
Subjects
Twelve adult braille readers of both sexes served as Ss in the experiment. They were all legally blind and relied exclusively upon touch for the perception of braille characters. Most of them made extensive use of braille in their daily lives. Some were proof-readers of braille books. Table 1 shows their age, years of experience in reading braille, and their estimated daily use of braille.
#TABLE1:
Experimental Materials
A table of random permutations was used to arrange the letters of the alphabet in random order. Thus, no letter recurred until all letters had been used.
Eight lines of these letters were reproduced in braille. Each line was divided into five groups of letters, with five letters in each group. The letters were organized in this way to minimize the problem of losing one's place. Ss were required to read a different line of letters with each of the eight fingertips.
Procedure
Since Ss were employed at a variety of places during the day, and could not report to a single testing site conveniently, they were tested in their homes at night. Each S was seated at a table and asked the questions that elicited the information displayed in the first seven columns of Table 1. Then, in order to get an estimate of reading rate, each S was given a 200-word passage to read. He was told to read the passage silently and in his accustomed manner. Time spent in reading was measured by a Standard Electric timer, Type 5-1. Reading rates for each S, expressed in words per minute, are shown in column eight of Table 1. Next, each S received instructions for participation in the experiment. He was asked to read the first line of braille letters in the booklet that had been given to him with the index finger of his preferred braille reading hand. He read the next line with the corresponding finger on the non-preferred hand. The third line of letters was read with the middle finger on the preferred hand, and so on until a line of letters had been read with each of the eight fingers. Record was kept of the time required to read each line, and of the number of errors per line. Collection of data in this manner permitted an analysis of variance of time and error scores, with 55, hands, and fingers as sources of variation.
RESULTS
The curve with open circles in Fig. 1 describes the relation between time spent in reading and finger used. The reading times for all Ss were averaged for each finger. Further, since the analysis of variance did not show the hand used to be a significant source of variation, values for corresponding fingers on the two hands were averaged.
#FIGURE1:
The curve in Fig. 1 with filled triangles shows the relation between number of errors and finger used. In determining the points in terms of which this curve was plotted, error scores were combined in the same way as were the time scores mentioned previously.
#TABLE2:
Table 2 presents the results of the analysis of variance of time scores. The significant F ratio in Row 1 of this table confirms the relation suggested in Fig. 1. The failure of the Fratio in Row 2 to achieve significance indicates that there was no consistent relationship for the group as a whole between reading time and the hand used to read. Many braille readers do express strong preferences for one hand over the other, and their preferences are often consistent with their reading behavior. However, others show little or no preference. Those who do express a strong preference may base it on something other than reading facility. For instance, the person may prefer a given hand for reading because it is the hand his teacher encouraged him to use. It seems reasonable to assume that an individual will read more often with his preferred hand and rely more upon his preferred hand when reading with both hands. Granting this, the insignificant Fratio in Row 2 of Table 2 suggests either that additional practice received by the preferred hand was of no benefit or that it benefitted the non-preferred as well.
The significance of the individual differences in performance is shown in Row 3 of Table 2. In this connection, it is also interesting to inspect the distribution of reading rates shown in Column 8 of Table 1.
The lack of significance of the hand X finger interaction, as shown in row 4 of Table 2, indicates that the amount of transfer did not depend upon the hand used to read. In the absence of such evidence, one might have expected a difference in the amount of transfer shown by the two hands.
The significance of the interaction shown in Row 5 of Table 2 means that some Ss showed more transfer than others. This is of course, not surprising, and it was suggested strongly by inspection of the raw data.
There was a significant interaction between S and hand as shown in Row 6. This finding is consistent with the experience of those who have observed the reading behavior of braille readers. Some readers read much better with one hand than the other. Others read equally well with either hand. Such individual differences are likely to be manifested in other cutaneous codes. (The second order interaction in Row 7 was used as an error term.)
Table 2 also shows the results of the analysis of variance of error scores. The entries, derived from error scores, support, in the main, the conclusions drawn from the analysis of variance of time scores, with one exception. The significant interaction between S and hand found when time scores were used was not found for error scores. The failure of the two analyses to agree completely may suggest a difference between error scores and time scores that is worthy of further exploration. However, since the disagreement occurred in only one instance, no inferences seem warranted at this time.
The results indicate quite clearly that there is a substantial loss in reading ability as the little finger is approached. Several factors may be involved in the explanation of this outcome. There are differences in the musculature associated with each of the fingers. The little finger and ring finger are not usually con trolled and manipulated as well as the middle and forefingers. Also, because of the way the forefinger is used in opposition to the thumb, it has undoubtedly received more practice in the execution of closely coordinated movements. This training should be useful in maintaining proper orientation to the braille characters and in following a braille line.
Another possibility is that there may be differences in the supply of sensory nerve endings to the various fingertips. Or, there may be differences in the amount of cortical representation for each fingertip. A literature search has so far revealed no investigations that would shed light on these possibilities.
One study has been reported in which practice appeared to be an important factor in determining the reading ability of a particular finger (Funchess, 1934). In this study, braille readers were given practice in reading connected prose pas sages with normally unused fingers. The author reported rapid improvement in reading ability with practice. This study does indicate that a solely neural explanation is not sufficient. However, it was scarcely more than a pilot study, and its design does not permit definite conclusions regarding the role played by prac tice.
In any case, anatomical, physiological, and experiential factors are likely to be involved in the reception of cutaneous codes, such as the electrical code under investigation at the University of Louisville. The results of the present study suggest that if, after an electro-cutaneous code has been learned at one set of loci, the electrodes are shifted to a new set of loci, there will be a loss in ability. The magnitude of this loss may depend, at least in part, upon the distance separating the new loci from the original ones, and there is likely to be a considerable loss, judging by the performance of the braille readers in this study. The amount of additional training needed to restore proficiency is of course, unknown. Results of the study by Funchess referred to previously, suggest that the amount may not be large.
On the basis of the finding that the hand used was not a significant source of variation, another expectation seems warranted. If a new set of loci happens to be the bilateral counterpart of the old set, transfer will probably be quite high.
Transfer of training is often accounted for in terms of the generalization of specific factors, such as the excitatory tendencies acquired during learning, and general factors, such as principles and set. In typical reading tasks, the reader is greatly assisted by the transfer of previously learned expectations regarding letter and word sequences. In the present study, Ss were required to read random sequences of characters. Thus, it is safe to regard this kind of transfer as having been eliminated. Those general factors relating to the skill of reading braille were of course still operative. However, it may be interesting to regard the relation shown in Fig. 1 as descriptive of primary stimulus generalization. If this experiment is viewed as a demonstration of stimulus generalization, one is reminded of its similarity to the experiment by Bass and Hull (1934), in which the generalization of a tactile conditioned stimulus was demonstrated. There is also some resemblance to the kind of generalization shown by Brown, Bilodeau, and Baron (1951). A response was associated with a stimulus applied at a particular point in visual space in one case and "tactile space" in the other, and then other stimuli, similar to the original conditioned stimulus, were applied at different spatial points. The magnitudes of the responses occurring to these new stimuli varied inversely as a function of their separation from the original CS, and the curve expressing this relationship was regarded as descriptive of primary stimulus generalization.
In the present experiment, stimuli were applied at different points in "tactile space". Because of the responses used, no direct indication of response magnitude was possible. However, rate of identification and errors in identification of characters provided the kind of indirect measures that are frequently used in studies of generalization of human verbal learning. From this point of view then, the curve in Fig. 1 may be regarded as a curve of stimulus generalization. If so, it is interesting to examine the steepness of the curve in connection with what is known regarding the relationship between the extent of stimulus generalization and the amount of prior learning. Although evidence on this point is not complete, there is a strong suggestion that, as the amount of learning increases, the generalization gradient is at first broad but becomes quite steep (Beritoff, 1924; Hovland, 1937; Pavlov, 1927; Speiger, 1956). Ss in the present experiment have read braille for years. Therefore, in terms of the reasoning presented here, they should be expected to manifest relatively narrow generalization, and the curve in Fig. I attests to this expectation.
Looking at the outcome from a somewhat different point of view, Dr. Webb (W. B. Webb, University of Florida, Personal communication) is impressed with the similarity between the experiment reported here and the experiments performed by Wickens (1943) to demonstrate response generalization. Ss in the present experiment experienced a group of stimuli which were, at least with respect to their physical characteristics, always the same. However, as the finger used in reading was changed, the point of application of these stimuli was of course varied too. In addition, to the extent that the fingers used in reading are involved in the responses evoked in a task of this sort, one might also argue quite reasonably that the results afforded an instance of response generalization as well.
The results just presented should, of course, be viewed with caution. Many variables that might reasonably be expected to have an influence on the outcome of the experiment could not be controlled. The methods used to teach braille to those who served as Ss in the experiment are not known, but it is known that braille teaching methods differ widely. Though an effort was made to determine the number of years each S had read braille, and his typical daily use of braille (see Table 1), only a rough estimate can be made of the amount of practice each S has actually received in the reading of braille. Furthermore, although it is reasonable to suspect such practice to be a relevant variable, it is not yet possible to gauge its effect precisely. Although Ss did express hand preferences, there is really no reliable information about the hand or fingers used in practice or the conditions under which practice occurred. One can safely assume that practice occurring since the termination of formal training has been largely unsupervised and probably relatively inefficient as far as feedback is concerned. Although the author recognizes these limitations, the experiment reported here is felt to be a reasonable first step in the examination of some of the perplexing aspects of perceptual skills.
REFERENCES
BASS, M. J., & HULL, C. L. The irradiation of a tactile conditioned reflex in man. J, comp. Psychol., 1934, 17, 47-66.
BERITOFF, J. S. On the fundamental nervous process in the cortex of the cerebral hemispheres: I. The principal stages of the development of the individual reflex: its generalization and differentiation. Brain, 1924, 47, 109-148.
BROWN, J. F., BILODEAU, E. A., & BARON, M. R. Bidirectional gradients in the strength of a generalized voluntary response to stimuli on a visual-spatial dimension. J. exp. Psychol., 1951, 41, 52-62.
FERTSCH, P. An analysis of braille reading. Outlook for the Blind, 1946, 40, 128-131.
FUNCHESS, L. V. The psychology of reading braille with eight fingers. Unpublished master's thesis, Louisiana State Univer., 1934.
GELDARD, F. S. Adventures in tactile literacy. Amer. Psychologist, 1957, 12, 115.
GILMER, B. v. H. Toward cutaneous electro-pulse communication. J. Psychol., 1961, 52, 211-222.
HAWKES, G. R., & WARM, G. Communication by electrical stimulation of the skin: I. Absolute identification of stimulus intensity level. Army Med. Res. Lab., Sept. 16, 1959. (Rep. No.400)
HOVLAND, C. I. The generalization of conditioned responses: IV. The effects of varying amounts of reinforcement upon the degree of generalization of conditioned responses. J. exp. Psychol., 1937, 21, 261-276.
PAVLOV, I. P. Conditioned Reflexes. Cambridge: Oxford Univer. Press, 1927.
SPETGER, C. C. The stimulus generalization gradient as a function of the intensity of stimulus lights. Child Develpm., 1956, 27, 85-98.
WICKENS, D. D. Studies of response generalization in conditioning: I. Stimulus generalization during response generalization. J. exp. Psychol., 1943, 33, 221-227
Accepted March 24,1964.