RELATIONS AMONG HANDEDNESS, HEMISPHERE LOCALIZATION, AND PERC‘EPTUAL SKILLS OF THE RIGHT HEMISPIIERE V "(basis for the fiegree 0‘? Ph. 3}. MICHIGAN STATE UIIIVERSITY. CHRISI'GPHER GILBERT 1973 ABSTRACT RELATIONS AMONG HANDEDNESS, HEMISPHERE LOCALIZATION, AND PERCEPTUAL SKILLS OF THE RIGHT HEMISPHERE By Christopher Gilbert Left-handedness has been linked with increased variability of cerebral localization for speech. This condition in turn has been assoc- iated with deficits in some perceptual skills thought to be mediated by the right hemisphere. To test these relationships, four handedness groups of 16 members each were formed: strongly right-handed, strongly left-handed, weakly right-handed, and weakly left-handed. The designation of "weak" indicated decreased lateral asymmetry and was based on both a manual dexterity test and a hand-usage questionnaire. Subjects were tested on three perceptual measures (Block Design, Object Assembly, and facial recognition) intended to assess primarily right-hemisphere abili- ties, and two reaction-time (RT) tasks designed to detect lateral asym- metry in speed of information-processing. .The first RT measure presented letter-pairs by tachistoscope in the left or right visual fields. An identification of letter-pairs which were physically dissimilar but sym- bolically identical (Aa) was considered more language-dependent than identification of letter—pairs which were physically identical (AA). It was predicted that RTs to right-field presentation of "symbolic" pairs would be faster, reflecting the more direct (non-commissural) sensory transmission pathways between the right visual field and the Christopher Gilbert left hemisphere. Physically-identical pairs were expected to yield a left-field superiority. There was little support for either prediction; the groups did not differ significantly in their visual field biases, nor was there any consistent bias across groups, so letter RTs were not useful for determining speech lateralization. The second RT measure was a face-discrimination task in which single faces were presented tachistoscopically for identification (4—face memory set). This measure revealed an overall left-field bias (RT faster to left visual field, ave. 12.5 msec., p‘ .01), which was expected be- cause of right—hemisphere specialization for face processing. No differ- ences were found among the four handedness groups; the face-processing bias was unaffected by the handedness classification. This unexpected result suggested that right—hemisphere perceptual specializations are more firmly located in the right hemisphere than speech is in the left hemisphere. When analyzed with multivariate analysis of variance, the four handedness groups did not differ significantly in their performance on the other three perceptual tests. Since the variables of handedness had very low correlations with each other, each variable was then examined separately: the effects of writing hand, asymmetry of manual dexterity, and consistency of hand-usage produced no significant differences in either the perceptual measures of the reaction-time measures. Sex was a significant effect for facial recognition, with females superior. The presence of left-handedness in the subject's family produced significantly worse performance on the facial recognition test (p‘ .0001) regardless of the subject's own handedness. This result is congruent with other Christopher Gilbert evidence that familial left-handedness is associated with decreased hemisphere specialization for speech, since the partial presence of speech function in the right hemisphere (inferred from handedness data) has been associated with a drop in non—verbal perceptual abilities. J. Levy's "interference" theory of neurological incompatibility of verbal and non-verbal functions was discussed in light of this finding. RELATIONS AMONG HANDEDNESS, HEMISPHERE LOCALIZATION, AND PERCEPTUAL SKILLS OF THE RIGHT HEMISPHERE By -I 90“" ChristopherIGilbert A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1973 ACKNOWLEDGMENTS This research was partially supported by a NIMH Biomedical Science Support Grant, Institute of Biology and Medicine, Michigan State University. It would have been impossible without my valiant and good—humored subjects. I am truly grateful to the members of my committee: Robert Raisler, Lester Hyman, Charles Hanley, and my chairman, David Wessel. Each helped, when needed, in his own unique way, contributing more than they suspect to the development of both this dissertation and its author. ii LIST OF TABLES LIST OF FIGURES INTRODUCTION . Audition . Vision . Reaction time asymmetry Left-handedness Visuo-spatial deficits Facial recognition Lateral face bias . Research plan METHOD . . TABLE OF CONTENTS Apparatus for reaction—time Procedure . Subjects . RESULTS . . Reaction times to faces Reaction times to letters Multivariate analyses . Correlations testing . iii Page vi 10 13 15 15 l6 l9 19 21 26 28 29 30 32 38 DISCUSSION . . . . . . . . Reaction time testing . . . . Multivariate analyses . . . . Correlations . . . . . . APPENDIX A: Hand-usage questionnaire. APPENDIX B: Distribution of questionnaire BIBLIOGRAPHY . . . . . . . iv SCOI‘BS Page 44 44 47 53 S7 58 59 10. 11. 12. 13. 14. 15. LIST OF TABLES Means of reaction times (in milliseconds) to faces in left and right visual fields . . . . . . Average absolute visual—field difference for face reaction times and number of subjects in each group with LVF bias . . . . . . . . . . . . Mean reaction times to letters under PI and NI conditions . . . . . . . . . . . . . Proportion of subjects showing a RVF bias under PI and NI conditions. . . . . . . . . . Average absolute visual—field differences for NI PI conditions . . . . . . . . . . . . Multivariate analysis based on handedness groups Means for subjects separated on the basis of asymmetrical dexterity . . . . . . . . . Means for subjects separated on the basis of hand-usage questionnaire . . . . . . . . Means for subjects separated on the basis of writing hand . . . . . . . . . . . . Means for subjects divided by presence of left-handedness in the family. . . . . . Group means for males and females and males . . Correlation matrix . . . . .' . . . . . Correlations of combined "weak" variables. . . Correlations of handedness variables . . . . Average absolute VF differences in RT for subjects divided on the basis of familial left-handedness (in milliseconds) . . . . . . . . . . Page 29 29 30 31 32 33 34 35 36 37 38 39 40 42 43 LIST OF FIGURES Page Letter-pair stimuli for letter-discrimination task . . . 20 Schematic illustration of stimuli used for half-face bias test . . . . . . . . . . . . . . . . 22 Distributions of handedness-questionnaire scores for IEfC-hBDdCI’B and right-handers o o o o o o o o o 58 vi INTRODUCTION In spite of the lack of strong morphological differences between the two hemispheres of the human brain, there appear to be large func- tional differences in the way they process information. Most of our early knowledge of hemisphere differences came from observing results of brain damage and from electrical stimulation of the living cortex. But the recent appearance of medically-created split—brain subjects (persons with severed cerebral commissures) has allowed the first prac- tical testing of each hemisphere in functional isolation, since stimuli presented to one hand or one visual field of such persons are received principally in the contralateral hemisphere. Split—brain research as a whole has revealed the cerebral hemispheres as fairly specialized for specific perceptual cognitive functions. Localization of speech produc- tion in the left hemisphere seems to be the most basic lateral asymmetry; the most notable recent general discovery is that much non-verbal per— ception, both visual and auditory, is similarly localized in the right hemisphere (see Benton, 1972, for a review). Several authors (Began, 1969; Bakan, 1971; Ornstein, 1972) have characterized the left hemisphere's abilities as analytic and symbol— manipulating, while the non-dominant hemisphere is diffuse and "Gestalt— 1ike" in function. Levy (1969) proposed a reason for the physical separation of symbolic and non—symbolic functions in the cerebrum: 2 possibly because of dissimilar neural organizations, analytic processing may be incompatible with the more diffuse non-symbolic processing. Semmes (1968) has suggested an overall difference between the two hemi- spheres, based on neuroanatomical research finding that the dominant ("speech") hemisphere is organized more discretely, corresponding point- to—point to sensory receptors, while the non—dominant hemisphere's sensory and motor connections are more diffuse and interconnected. This would seem to make the right hemisphere more suited for Gestalt-like perception, while organization of the dominant hemisphere would be more appropriate for discrete symbol manipulation. Semmes' evidence was derived from mapping of the human cortex, and fits well with other evi- dence of functional differences. These broad generalizations to characterize an entire hemisphere's perceptual and motor functions may seem excessive, but they represent the first crude attempts to classify the emerging specializations of each half-brain under a single label. The situation is probably not that simple. But at present, the basic differentiation between symbol manip- ulation and Gestalt-like comprehension, whether mutually antagonistic neurologically or not, seems to be the best general dichotomy available. Mbst of our recent knowledge of individual hemisphere function has come from studying split-brain patients (Sperry, 1964; Gazzaniga, 1967, 1970; Milner, 1968). In the intact human being, one would assume that the cerebral commissures equalize differences between the hemispheres, as far as input and output are concerned. It would be awkward for humans to exist in a state in which only the left hand could recognize objects tactually, or in which one must look at the left of a word to comprehend it, or to the right of a face to recognize it. These 3 restrictions do, however, hold true for split—brain persons; such people are living illustrations of how we would all function perceptually without our cerebral commissures. The overall effect of the commissures seems to be making the human being's sensory capacities more symmetrical, compensating for a degree of asymmetry of function greater than any other animal, and upon which our present intellectual capacities may depend. From recent evidence it seems that this commissural "compensation", and thus the symmetry of our perceptual world, is not perfect. To a certain extent we are all "split-brain" persons, though the effects are normally unnoticeable in daily life. Careful testing of human perceptual capacities has revealed definite asymmetries parallel to those produced when the commissures are severed. Already a large number of studies have extrapolated from split-brain discoveries and have found evidence of the same hemisphere differences in intact human subjects, though much reduced. These "perceptual asymmetry" studies use both visual and auditory modal- ities. I will review the most important of these and then discuss the relations of handedness and visuo-spatial abilities to the topic of sensory asymmetry. Audition Results of dichotic listening experiments have been used to demon— strate hemisphere specialization in auditory information processing. The anatomical connections are not nearly as clear-cut fOr audition as they are for vision; each ear has afferent connections to both hemispheres, and unlike vision there is no way to separate the ipsilateral and contra- lateral channels. To explain asymmetries in auditory perception, experi— menters have assumed that the contralateral channel is somehow more 4 "primary," similar to the greater discrimination contralaterally for kinesthesis. There is some evidence that the contralateral auditory channel actually is more efficient( Rosenzweig, 1951) but the question does not seem entirely settled yet. The classic dichotic listening paradigm has been to use two simul- taneous channels for presenting stimuli. Typically, two different strings of letters or digits are delivered, one to each ear, at a rate of one to three per second. The recordings to the two ears are matched for sound-pressure level, usually close to threshold. Sometimes a white-noise background is used. Periodically the tape is stopped and the subject is asked to recall every letter or digit he heard. Under these circumstances, most persons report more of the stimuli delivered to the right ear than to the left. Some typical early studies reporting this effect are by Milner (1962) and Kimura (1967). Most studies of dichotic listening have tested left-hemisphere specialization only, probably because numbers and digits are easy stimuli to work with. But several studies have shown that the expected right— hemisphere (left ear) superiority for its specialties exists too: for melodies (Kimura, 1964; Shankweiler, 1966) and for non-verbal environ- mental sounds (Curry, 1967; Knox & Kimura, 1970; Bakker, 1970). Examples of the "environmental sounds"used are recordings of trucks, airplane noises, and flushing toilets. Left-ear superiority has also been found for sequences of dots and dashes. Perception of temporal patterns, although verbally encodable, is considered from other research to be right-hemisphere-mediated. The dichotic listening effect has proved compatible with develop- mental data on cognitive development; for example, the auditory 5 asymmetry usually develops in children around age five, and it usually appears earlier in girls than in boys, coinciding with their earlier verbal development (Knox & Kimura, 1970). Also, Geffner & Hochberg (1971), acting on earlier evidence that socio-economic level influences the onset of auditory asymmetry, tested large numbers of children at different ages from "normal" and "depressed" environments. The children from the depressed environment developed auditory asymmetry two to three years later, on the average. Presumably this reflected delayed speech lateralization, and delayed speech lateralization in turn has been linked with reading retardation, much more common in the depressed environment. It should be noted that dichotic listening studies actually measure the subject's perceptual bias for stimuli delivered in one ear over the other ear, rather than measuring absolute discrimination ability. When the ears are tested separately by unilateral presentation, the asymmetry generally disappears (Calearo and Antonelli, 1963; Palmer, 1964). The hemisphere-dominance explanation, therefore (stated most clearly by Kimura, 1967) has stressed the interaction of contralateral and ipsi- lateral neural transmission; competition between the two auditory chan- nels is thought to result in contralateral dominance. So the dichotic listening experiments do no more than determine which of two rival sets of stimuli seem to dominate the subject's conscious attention. This is very different from the asymmetries discovered for detection of stimuli in the right and left visual fields, which are discussed next. Vision Taking advantage of the clear contralateral relationship between the visual fields and the hemispheres, early experimenters sought a superiority for letter detection in the right visual field. The standard procedure is to present single letters or digits tachistoscopically for less than 200 milliseconds in one visual field at a time, using a pre— exposure central fixation point to ensure that the stimulus goes to the desired hemisphere. Sometimes words are used instead as stimuli, pre- sented either horizontally or vertically. The usual result, generalizing from a number of similar studies, is that letters, digits, and words in the right visual field can be identified more accurately (Kimura, 1966; Hines & Satz, 1971; McKeever & Ruling, 1971). The most-used explanation for this phenomenon is that since visual input from the right visual field (RVF) goes directly to the left hemisphere, which dominates for digit and letter recognition, this perceptual "route" is shorter and perceptually clearer than the LVF-right-hemisphere route. This assumes that material appearing in the LVF must be transferred right—to—left across the corpus callosum for letter identification; somewhere in this longer pathway between stimulus and response there is a loss of information. Fedio & Buchsbaum (1971) have demonstrated such a degradation of the evoked potential after being transferred across the commissures. There have been objections to this explanation (see White, 1969, for a review). One consistent finding has been that simultaneous bi- lateral presentation of letters and digits usually results in a left- field superiority for detection (Kimura, 1966). This LVF superiority was thought to reflect the left-to-right reading habits which could 7 lead subjects, when confronted with two letters, to report the left one, even though a pre-exposure central fixation point and short exposure time are used. The left-field bias could have been based on attention or on order of report even in the absence of eye movements. The most recent development in this controversy has been the research of McKeever & Huling (1971, 1972) and Hines (1972). It was found that requiring subjects to report a small digit appearing at the central fixation point abolished the LVF superiority which usually reSulted from bilateral presentation. Subjects now recognized words better in the right visual field. The authors argue that with bilateral presentation, extra control of fixation is necessary to counteract the left-to-right scanning and reporting habits which could lead subjects to unintentionally fixate slightly to the left. One would also expect the left visual field to be superior for recognition of abstract shapes and faces, since split-brain research has revealed this right-hemisphere specialization. The only notable successes so far for recognition accuracy have been with dot patterns (Kimura, 1966, 1969; McKeever & Ruling, 1970) and with judgments of three-dimen- sional depth (Durnford & Kimura, 1971). Geometric shapes have been found to be equally recognizable in either visual field, but they can be easily encoded verbally, while irregular patterns of dots cannot. Reaction time asymmetry, Measurement of choice reaction times to stimuli directed to only one hemisphere is a new technique which has supported and extended the results for early detection studies. The experimental arrangement is generally to have a subject memorize a set of letters or digits. Stimuli 8 are then presented with a tachistoscope to the left or right visual fields; the subject discriminates between the two sets of stimuli (on the basis of the memory set) and responds as fast as possible. Instead of proportion of correct detections, the average response time is used as the index. The expectation is that the symbolic stimuli projected to the right hemisphere will be identified more slowly than will stimuli projected to the left hemisphere, because letters and digits must be processed in the left hemisphere. This extra crossing time should result in slightly longer reaction times. Filbey & Gazzaniga (1969) were the first modern researchers to use this technique, although it had been found and then forgotten in the 1920's. Filbey & Gazzaniga found a RVF superiority for both letters and digits. Bradshaw & Perriment (1970) and Klatzky (1970), besides verifying Filbey & Gazzaniga's results, investigated the variable of which hand is responding, in case this was a possible complication. Their general finding was that there is no difference between ipsilateral and contra- lateral hand—response times unless individual finger control for separate responses is required. Rizzolatti, Umilta and Berlucchi (1971) and Geffen, Bradshaw and wallace (1971) found faster reaction times to letters in the RVF and a faster reaction time to faces in the LVF. In all these studies, subjects were only required to identify a single letter, and this produced reliable RVF superiority on the order of 15 to 20 milliseconds. However, Klatzky & Atkinson (1971) presented evidence that simple identification of letters could be done as well by the right as by the left hemisphere, because the letters were being stored and processed as spatial configurations only. These experimenters found RVF superiority only when subjects were faced with a more complicated 9 task requiring them to match letters to the names of pictures, which was more clearly a verbal task. Their study raised important questions about the real difference between visual symbols and visuo-spatial con— figurations, and when their results were supported by the research dis- cussed in the next paragraph, it became apparent that letters could not be simply considered as verbal stimuli, but that the designation depended on the discrimination required of the observer. An effective technique for examining hemisphere differences for letter processing is an adaptation of a same-different judgment task originated by Posner & Mitchell (1967). Pairs of letters are prepared for two experimental conditions. In both conditions the subject must decide whether or not two letters simultaneously appearing in the same visual field are "the same"-—but this can either mean the same symboli- cally, or the same physically (and also symbolically). Thus in the sym- bolic (name-identity) condition the subject must discriminate between "Aa and "Ae", while in the physical-identity condition the subject must discriminate between "AA" and "AE." The second discrimination is thought to be less dependent on left-hemisphere verbal specialization, while the first discrimination requires a learned knowledge that "A" and "a" are symbolically equivalent. Three studies (Geffen, Bradshaw and Nettleton, 1972; Cohen, 1972; Ledlow, Swanson and Carter, 1972) have demonstrated a right-field superiority for the name-identity condition and in some cases a left-field superiority for the physical-identity condition. 10 Left-handedness A common practice in both visual-field and dichotic-listening studies has been to use all right-handed subjects. This practice maxi— mizes the chances of obtaining the predicted results, for the assumption that verbal function is in the left hemisphere is only probabilistic when applied to an individual. The most accurate information on this matter now available is derived from the procedure of anesthetization of one hemisphere at a time and observing the degree of aphasia resulting (Branch, Milner and Rasmussen, 1964). On a sample of 48 right- and 44 left-handers, 902 of the right-handers were found to have major speech centers in the left hemisphere, and 102 in the right hemisphere. Of the left-handers, 64X seemed to have major speech centers on the left, 202 on the right, and 162 had bilateral representation. Therefore at least 202 of left-handers seem to have "reversed" brains, according to this test. The authors also observed that for left-handers generally, speech seemed more evenly distributed between the hemispheres, or less "latera- lized." So excluding left-handers in perceptual asymmetry studies amounts to excluding the only known variations from the standard hemisphere- dominance pattern. This seems unwise when the experimenter has no other real assurance that hemispheric specialization is being tested at all. Hecaen and Sauguet (1971) provided the most comprehensive recent analysis Of correlations between handedness and brain-damage effects. After surveying hundreds of cases of localized brain damage, they con- cluded that speech impairment occurs fairly often following right- hemisphere damage to the brains of left-handers; this result is rare in right-handers. In a smaller sample, they examined the variables of strength of handedness and left-handedness in the subject's family. 11 H. L. Dee (1971) also studied the variables of handedness in relation to dichotic listening performance, and his conclusions closely match those of Hecaen & Sauguet's: 1) A strongly left—handed person is most likely to have the normal speech-on-the-left cerebral organization. 2) Persons who are less strongly left—handed (showing less strong lateral motor specialization) are the ones most likely to have reversed or less lateralized cortical function, at least for speech. In addition, Hecaen & Sauguet found from their data that "weak" left—handers are also more likely to have left—handedness in their immediate families. So instead of grouping all left-handers together as most studies have done (if left—handers are included at all) it now seems best to separate them into strongly and weakly left—handed, and to pay close attention to the presence of familial left-handedness. The placement into "strong" and "weak" groups can be done both with hand-usage ques- tionnaires and with dexterity tests. H. L. Dee found with such testing that only 30% of left-handers he sampled could be called strong left- handers. The rest were "weak"--exhibiting less lateral dominance than the average right-hander does. Besides Dee's study, several others have included the handedness variable with perceptual-asymmetry testing, although the sample size in these studies is usually small. First, Kimura (1961) tested left-handers with speech located definitely in the right hemisphere (as determined by hemisphere anesthetization) with a dichotic listening procedure, and found reversed laterality, as expected: the left ear dominated for letter and digit recognition. This single study, incidentally, added immense validity to the whole dichotic-listening paradigm. Zurif & Bryden (1969) found a strong difference between familial 12 and non-familial left-handers with the dichotic listening test and also with a letter-recognition test. Familial left—handers had decreased or reversed ear and visual-field differences. Satz, Achenbach and Fennel (1967) and Satz, Fennel and Jones (1969) found a positive relationship between strong left-handedness and deviations from the normal right-ear superiority on dichotic listening tests. Bakker (1970) assessed the degree of laterality development-~"sidedness" as well as handedness--in a large number of school children and found that inconsistent or retarded laterality development was reflected in a dichotic listening task: such children had either weaker asymmetry or none at all. Ledlow et a1. (1972) tested a group of 10 left—handers (simple self-report was the only criterion) in the previously-described physical vs. symbolic letter-pair matching. They found little difference in reaction times between left-field and right—field scores, as expected, while right—handed subjects showed a RVF superiority for name-matching. Hines & Satz (1971) found similar results in a group of left-handers, using accuracy of detection of letters as the index. They also reported that in their sample, right-handers with left—handedness in their families showed less right—field bias for digit detection, supporting the specu- lation that handedness has a genetic component. To summarize the present state of knowledge on this matter, the three most important components of handedness seem to be: 1) Strength of handedness as measured by usage questionnaires (hand usage for various common manual activities) 2) Asymmetric dexterity (difference between hands) as measured by manual skill tests 3) Familial handedness: the presence of left-handedness in the l3 immediate family. The first two components may measure the same thing in different ways; no report of correlations has appeared in the literature yet. Some studies in the past have concentrated on the familial left-handedness variable, ignoring dexterity and usage. Others have carefully focused on finding "pure" left-handers either by actual testing or by question- naire, leaving the familial variable unmentioned. Visuo-spatial deficits Levy (1969) was the first to report that a sample of left-handers had performed significantly lower than a matched group of right-handers on the performance scale of the Wechsler Adult Intelligence Scale( WAIS) with no difference in their verbal scale scores. This finding was pre— dicted; Levy reasoned that left-handers, having more variable locali— zation of speech function, may for this reason be impaired in non—verbal tasks because the language function of the brain interferes with normal right-hemisphere functions. Levy's study has since been verified with a different IQ test (Miller, 1971) although Ledlow et al. (1972) did not obtain such results. Other research with actual right-hemisphere-damaged persons on WAIS per- formance (Lansdell, 1970, Vega, 1971) has shown that although the "perfor- mance" scales as a whole measure right-hemisphere abilities more than left, the greatest right-hemisphere involvement seems to be on the Block Design and Object Assembly subtests. Robert Nebes (1971) published results of his Arc-Circle test, de- veloped to test Gestalt-closure ability in split-brain subjects. The test requires subjects to choose visually from several complete circles the 14 one which best matches a circle segment examined out of sight with the fingers. The validity of this test for assessing right-hemisphere func- tions rests on Nebes' reported testing of many split-brain subjects; the left-hemisphere-right-hand "team" performed very poorly, usually at chance level, while the opposite combination did as well as normal sub— jects. Nebes also tested left-handed and right-handed normal (intact) persons and found left-handers to be deficient, performing significantly worse than the right-handers. Because of the intact commissures, it did not matter which hand was used. The left-handers in these tests, again, were undifferentiated; familial, non-familial, "strong" and more ambi- dextrous were all pooled. So although the studies are useful for demon- strating the involvement of laterality in the tasks, they are less useful for studying handedness itself. To summarize: there are at least two established perceptual de- ficiencies already shown for left-handers as an undifferentiated group: deficit in non-verbal parts of intelligence tests, and lower performance on the Arc-Circle test. In addition, dichotic listening tests show much less ear asymmetry, if any, for lefthhanders, and visual-field testing shows less visual-field bias for left-handers. The two latter tests indicate less complete lateralization of function, and this in turn serves as support for the theory of brain evolution proposed by Levy: that the human's recently developed verbal-symbolic function is income patible with Gestalt-like spatial function because of some fundamental difference in neural organization. When verbal activity intrudes into the non-dominant hemisphere the latter's functions are impaired, although for some reason verbal function is not correspondingly enhanced. 15 Facial recognition Facial recognition is the most strongly established right—hemis— phere specialization, judging from the agreement of brain-damage studies (De Renzi & Spinnler, 1966; warrington & James, 1967; Milner, 1968). Damage in particular to the temporal lobes of the right hemisphere commonly impairs the person's ability to recognize faces, while this impairment is rare with left-hemisphere damage alone. From Levy's research finding a deficit for left-handers in non- verbal skills, it seems logical to generalize the finding to facial recognition ability, since this is also a non-verbal discrimination. Here, in reference to Dee's and Hecaen & Sauguet's recent research, it is most likely that more ambidextrous left-handers would be the most im- paired on this skill, because of their more variable and decreased asym- metry of speech function. This result was obtained (Gilbert, 1973) in a sample of left-handers divided into strongly and weakly left-handed. Compared to a group of right—handers, the weakly left-handed group per- formed significantly lower in a facial recognition test, while the strongly left-handed subjects did not differ from the control group. Lateral face bias One important component of facial recognition is the lateral one; that is, perception of a face in left or right visual fields. The two studies already cited showing a faster reaction time to faces in the left visual field suggest that a right-hemisphere specialization for face processing exists in intact subjects. A unique study by Levy, Trevar- then and Sperry (1972) used very dissimilar faces joined together at the midline to make a chimeric whole face. These were presented, centered 16 in a tachistoscope, to split-brain subjects, thereby putting each half- face into the contralateral hemisphere. When the subjects were asked to choose between the two versions of the face just viewed, they usually chose the complete face whose half was represented in the left visual field. When they were asked to describe verbally what they saw, however, they described the face represented in the right visual field, since the left hemisphere controlled the speech mechanism. When no verbalization was required, the right-hemisphere-left-field link was generally dominant. This study showed a very strong right-hemisphere superiority for face- processing, presumably because of the lack of commissural transfer. The same LVF bias seems to exist in intact subjects, though much reduced (Gilbert & Bakan, 1973). Subjects were found to consider material from the left side of a photographed face more "similar" to their impres- sion of the entire face. This was tested by offering two half-face com- posites for comparison with the whole face. The bias, small but consistent, is not due to specific properties of the right side of the face, but seems to depend only on position of the stimulus material relative to the rest of the face. A group of left-handers showed no bias for the left visual field; this was interpreted as a consequence of their less consistent lateralization. Researchgplan The present study was undertaken to test the idea that decreased cerebral lateralization is associated with slight impairment of right- hemisphere abilities. From the preceding review it is clear that many methods can be used to investigate perceptual asymmetry (and presumably l7 cortical lateralization) so some choice had to be made among these methods. The Arc-Circle test was discarded because a pilot study with 60 subjects failed to replicate Nebes' published results. Unfortunately, the most valid method for determining functional lateralization-—tempo— rary anesthetization of each hemisphere--was not feasible. Handedness is probably the next best predictor of where the primary speech centers are located, but handedness only predicts proportions among groups. If the assumptions behind visual-field reaction-time testing are true, then this measure should indicate functional lateralization in individual cases. So reaction-time testing was chosen as a second measure of later- alization. A compromise was necessary between obtaining a reliable measure of reaction time (ideally, several hours of testing per subject) and using enough subjects in each handedness group to make the laterality predic- tions meaningful. The spatial-ability tests also required a fair sample size. The final plan called for around three hours of testing per sub- ject, in two separate sessions. All testing was to be done individually except for facial recognition. Handedness was the reasonable independent variable because most other laterality tests are correlated with handed- ness instead of with each other. Reaction—time asymmetry along with the half—face bias both served as less certain concurrent measures of later- alization. It was predicted that reaction times for left and right hemispheres would be more similar in the weakly left- and right—handed subjects, and would show more lateral asymmetry for the strongly right- and left- handed subjects. Specifically, reaction time was expected to be faster in the left visual field for faces and for the physical-identity letter l8 discrimination, and faster in the right visual field for the name— identity letter discrimination. The Block Design and Object Assembly subtests and the facial recognition test served as measures of right- hemisphere function; all were predicted to be more difficult for the weakly lateralized subjects, and for those with left—handedness in the family. 19 METHOD Apparatus for Reaction-time Testing, Stimulus materials for the reaction time experiments were presented on a Scientific Prototype model 800 2-field tachistoscope with binocular viewing. Exposure duration was kept constant at 150 milliseconds. Illu- mination was .20 log foot-lamberts at both the pre-exposre field (con- taining the fixation point) and the stimulus field. Eye movements were not monitored, since there is adequate evidence that subjects in similar experiments do fixate properly (Geffen et al., 1972, Cohen, 1972). A switch controlled by the experimenter simultaneously exposed the stimulus field and activated a Hunter Clockounter millisecond timer. The subject's response switch consisted of a lever which stopped the timer when moved either forward or backward. For half of the subjects, a forward move— ment of the lever signified a discrimination of "same" for the letter pairs; for the other half the direction was reversed. Subjects were instructed to use their whole arm for moving the switch, and to keep their wrist and fingers stiff. This was done to minimize individual finger movements, which might have introduced an unwanted contralateral- hemisphere advantage. Face stimuli. Stimuli for the face discrimination task consisted of four small black-and-white portraits, two male and two female. In a pilot study, six portraits were used, with subjects moving the response lever one way for three of the faces and the opposite way for the other three. This proved difficult for most of the subjects to learn quickly. Rather than change to faces more easily discriminated, the size of the set was reduced to four. Each picture was 2.5 x 3.0 cm. in size, mounted 20 on a white card so that the inner edge was 15 mm. either right or left of the central fixation point (l.1°) and the outer edge was 40 mm. (3°) from fixation. Letter stimuli. For the letter discrimination task, letter-pairs 5 mm. high were drawn with India ink on white cards, using a Leroy lettering guide and #2 pen size. The closest edge of the closest letter to the central fixation point was 16 mm. to the left or right (1.22° of visual angle) and the outer edge of the outer letter was 28 mm. (2.l4°) from the center. The letters A, E, G, and R were used; these were chosen because of the differences between their upper and lower-case forms, making a discrimination of symbolic identity maximally different from a discrimination of physical identity. Figure 1 shows the letter-pairs used and their class membership; each pair was presented in both visual fields, making a total of 64 stimulus cards. SAME DIFFERENT Name Physical identity identity Aa AA AE Ae aA as as eA Ee EE EA aE eE ee ea Ea Gg GG GR Gr 86 as gr 1'0 Rr RR RC gR rR rr rg Rg Figure 1. Letter—pair stimuli for letter-discrimination task. 21 Procedure Subjects were familiarized with the same-different discrimi- nation for letters with approximately fifty practice trials. The 64 stimulus cards were then shuffled and presented in random order twice, once through for each hand. Half of the subjects began with the right hand and half with the left. The experimenter gave a verbal "ready" signal approximately one second before the onset of the stimulus. For the face discrimination, subjects studied a card containing duplicates of the four stimulus faces to learn the stimulus-response relationships; two of the faces called for a forward response of the lever and two for a backward one. Subjects underwent fifty practice trials, more if their learning was obviously slow. The cards were then shuffled and presented in random order both with respect to stimulus and to visual.field. Sixty-four observations were collected from each subject, 32 for each visual field, with the responding hand switched halfway through. Errors. For both the letter and face reaction times, errors were treated by informing the subject of the error, recording where the error occurred, and re-inserting the card at a randomly selected position in the remaining cards. Times for errors were not recorded. Half-face Bias Full details of the preparation of these stimuli are in an earlier report (Gilbert & Bakan, 1973). Thirty sets of prepared faces were used. Each set consisted of a normal portrait of a face, 4" x 5", and its two related half-face composites. Each composite contained two prints of either the left or the right side of the whole face, with one 22 print reversed and joined to the other at the midline. This gave the superficial appearance of a normal face, but each composite was perfectly symmetrical and contained material from only one half of the face. Thus a ”left composite" consisted of only the left side of a face (adjacent to the photographed person's left hand), and a "right composite" consisted of only the right side of a face. The composites were the same size as the originals; all three prints of each set were mounted on cardboard and trimmed to give the same clearance on each side between the face and the print's edge (see Figure 2). WHOLE FACE COMPOSITES RIGHT LEFT Figure 2. Schematic illustration of stimuli used for half~face bias test. 23 For each set, the subject's task was to decide which of the two composites looked more like the whole face. The major difference from the original method of presentation lay in not allowing the subject to view the whole face and the composites simultaneously. Whereas the original results were obtained by allowing the subject to look back and forth among the prints for comparison, this time the whole face was presented directly in front of the subject for five seconds, and then covered with the two composites. This was done to determine whether the 602 left-field bias would extend from an "immediate perception" condition (being allowed to compare both composites with the original) to a new situation calling for accurate memory of the whole face. To respond, the subject was asked to point to the composite which he thought more closely resembled his memory of the covered face. Very few sub- jects, when questioned later, had figured out how the composites were made. Picture-sets were always presented in the same order, with the left-right position of the two composites randomly varied to minimize any position biases which the subjects possessed. Each subject's final score was the number of choices of the left—field composites; i.e., from the left side of the whole-face photograph. Half of the whole faces were "original" photographs, depicting the persons as they ap— peared in life; the other half were "reversed," or mirror images, although the previous study established that this is not a confounding variable. Block Design This subtest of the Wechsler Adult Intelligence Scale was administered as described in the WAIS manual. The test consists of 24 nine blocks, each identically colored, which are used to duplicate increasingly complex geometrical desings. Scores depend on the elapsed time for correct construction of a design, and were assigned according to the WAIS norms. Total possible score for the ten designs was 48. Object Assembly This was another subtest of the WAIS in which the subject assembles four simple cardboard puzzles of Common objects such as a hand and a human profile. Following the standard procedure, the pieces for each puzzle were laid out in the prescribed orientation behind a shield. The subject was instructed to assemble the pieces so that they made a recognizable picture. Scores again depended on the elapsed time for correct assembly, with points deducted for errors. The total pos— sible score was 44, with scores assigned from the WAIS norms. Handedness The initial classification of handedness was made according to the subject's writing hand. Within that classification, each subject's degree of left- or right-handedness was assessed in two ways: by the difference between the hands for manual dexterity, and by the score on a standard hand-usage questionnaire. The questionnaire (Crovitz & Zener, 1962) consists of a list of 14 common activities such as using scissors, holding a hammer, and throwing; responses are made on a 5- point scale graded from "left hand always" to "right hand always." The range of scores is from 14 to 70, with 14 representing complete left— handedness for every activity listed and 70 representing complete right- handedness (see Appendix for a capy of the distribution of scores and a copy of the questionnaire). Examination of the score distribution for 25 left-handers suggested a score of 30 as a reasonable dividing point for separating left—handers into "weak" and "strong;" 30 is approximately at the median. A corresponding dividing point of 54 was used for the right—handed group. The dexterity test was the second definer of strength of handedness. As in an earlier study (Gilbert, 1973) the test was impro- vised, modelled after the principles of the Crawford Small Parts Dex- terity Test, since both tests measure fine eye-hand motor coordination. Subjects were directed to quickly place twelve small metal washers on twelve straight pins which protruded from a styrofoam ball, using tweezers, and then to remove them one by one. The elapsed time for this sequence constituted the first score for that hand. In all cases the dominant (writing) hand was tested first, then the non—dominant. Usually two trials for each hand sufficed, alternating hands; if results were un- clear a third and fourth trial were run. If the non-dominant hand completed the task consistently faster or if there were no clear dif-_ ferences between the hands (less than five seconds) then the person was classified as weakly left— or right-handed. It is not clear whether a greater variability of reported hand usage on the questionnaire or a smaller difference in tested dex- terity should be the better definer of "weak" handedness. Therefore, if a subject was classified "weak" on either the questionnaire or the dexterity test, that became his final classification. All other persons were designated either strongly right-handed or strongly left-handed. 26 Facial Recognition The stimuli for this test were the same as those for the earlier study, and presentation was identical except for more closely controlled lighting conditions, which increased the contrast of the slide-projected stimuli and raised the average score slightly. From old university yearbooks, groups of faces were selected and combined to make four arrays of 40 faces each, two all male and two all female. All clothing was blacked out, and prominent earrings, unusual hairdos, and obvious identifying characteristics were causes for rejection of a picture. From each array eight faces were individually photographed and enlarged. These sets were re-photographed onto slide film, as were the 40-face arrays, for group presentation. Subjects were tested in small groups of five or six, mixed with respect to handedness category. After instructions, the first set of eight test faces was projected onto a screen for 25 seconds at a distance of approximately 20 feet from the subjects. Immediately afterward, the corresponding 40-face array was projected. On answer sheets subjects identified the faces they recognized from the first group, with no time limit. This procedure was repeated for the other three sets, always in the same order. The final score was the total correct identifications for the four sets. Subjects In order to fill the four handedness categories, several methods were used to recruit subjects. The study was initially adver- tised as a "Facial recognition" study, sometimes specifying left—handers. Subjects received only class credit for participation. Later subjects 27 were recruited with an offer of $5.00 for participation, with no stipu— lations about their handedness. The weakly-right-handed category, how— ever, was the most difficult to fill. Some subjects were recruited by describing the desired characteristics to large introductory psychology classes ("...if you write right-handed but do a few things better left- handed, or if you feel equally dextrous with both hands..."). Class credit was offered and payment was not mentioned directly, but subjects who passed the initial screening (handedness questionnaire and dexterity test) received $5.00 payment. 28 RESULTS First, the results of the reaction-time tests were analyzed to determine if they could reasonably be used as measures of laterali- zation in individuals. To attain equal cell sizes, four subjects were removed at random from the strongly left—handed group and one was drawn from the weakly left-handed group, resulting in 16 persons for each handedness group. Some extremely long reaction times were considered to be due to extraneous factors, so the following cut—off rules were used for eliminating such times: reaction times longer than 1200 msec. were discarded, and for a given subject and stimulus class, if any single reaction time was isolated by more than 200 msec. from any other in its class it was regarded as spurious and discarded. This follows the general practice of the other studies of this type, although the cut—off point varies. The number of discarded times averaged less than 22 for all groups. Reaction Times to Faces A 2 x 4 analysis of variance with repeated measures on the visual fields was done on the means of all correct responses for each visual field and subject group. Table 1 shows the mean reaction times for each group, for each visual field. In this and in subsequent tables, standard errors are given beside each mean. 29 Table 1. Means of reaction times (in milliseconds) to faces in left and right visual fields. Handedness group LVF RVF s.e.m. s.e.m. SRH 708 (20.2) 715 (18.1) SLH 702 (19.9) 716 (17.9) WLH 710 (16.0) 725 (17.7) WRH 688 (21.2) 702 (24.6) Only the main effect of visual field was significant (F = 1, 60 9.42, p( .01). The lack of interaction between visual field and handed- ness indicates that all four groups had a similar left visual field superiority. The average difference between the LVF and RVF mean times re- gardless of sign was calculated for each group, to check the possibility that the average absolute difference might be smaller in the two "weak" groups, reflecting less hemisphere asymmetry. With this calculation, no assumptions were made as to the location of facial—recognition function. Table 2 contains these differences and also the proportion of subjects in each group having a faster mean time for the left visual field. Table 2. Average absolute visual-field difference for face reaction times and number of subjects in each group with LVF bias. SRH SLH WLH WRH 25.7 24.7 28.1 23.3 11/16 11/16 11/16 12/16 30 The equivalence of LVF biases for the four groups is sup- ported by the equivalence of absolute differences and by the number in each group showing a LVF bias. Reaction Times to Letters Two separate conditions of discrimination were present in this segment: Physical Identity (PI) and Name Identity (NI). The subject means were analyzed using a 2 x 2 x 4 analysis of variance with repeated measures on visual fields and identity condition. Table 3 shows the means for each condition, for all "same" responses. Table 3. Mean reaction times to letters under PI and NI conditions. Physical Identity Name Identity (AA) (A8) LVF RVF LVF RVF s.e.m. s.e.m. s.e.m. s.e.m SRH 627 (13.2) 617 (12.5) 747 (20.2) 725 (14.6) SLH 623 (22.1) 619 (17.5) 759 (29.6) 764 (28.3) WLH 619 (24.1) 623 (25.3) 713 (29.8) 734 (23.6) WRH 620 (19.6) 615 (24.7) 730 (28.4) 728( 26.9) nificant, and no interaction was significant. The main effects of groups and visual fields were not sig- The PI condition was significantly faster overall than the NI condition (F1,60 - 396, p‘ .001). Another way to analyze these reaction times is to compare the differences within each hemisphere between the PI and NI conditions. Cohen (1972) used this method, reasoning that if the PI condition favors the right hemisphere and the NI condition favors the left hemisphere. 31 (although the PI discrimination is nearly always faster than the NI regardless of hemisphere) then the difference between the two conditions should be larger in the right than in the left hemisphere. Following Cohen's example, a Wilcoxon signed-rank test was performed on the within-hemisphere differences for all 64 subjects. This is a non— parametric test which takes account of both the direction and degree of difference. There was no support for Cohen's prediction; the two hemispheres did not differ significantly in their degree of PI—NI difference. Cohen's data did, however, show significant differences between hemispheres using conventional analysis of variance, and her Wilcoxon analysis supported that analysis. In the present study a separate Wilcoxon analysis was done on the strongly rightéhanded subjects only, but results for this group alone were also non-significant. Table 4 shows the proportion of subjects in each handedness group who exhibited a right-field bias in either condition. Table 4. Proportion of subjects showing a RVF bias under PI and NI conditions. SRH SLH WLH WRH PI 7/16 7/16 9/16 9/16 NI 11/16 7/16 4/16 9/16 Again, the average absolute differences were computed for each handedness group for PI and NI conditions (Table 5). No strong pattern was apparent except that the NI condition produced consistently Q larger mean differences. 32 Table 5. Average absolute visual—field differences for N1 and PI conditions. SRH SLH WLH WRH PI 33 27 24 36 NI 48 38 43 45 Multivariate analyses To test for the presence of group differences on the remaining measures, a multivariate analysis of variance was carried out. Table 6 gives the group means for each dependent variable, standard errors of the means, and the univariate F-ratios. The cell entries for the first three variables are mean test scores; for the half-face variable, any number above 15.0 indicates a left—field bias; the reaction-time entries are in milliseconds. There were no significant effects due to handed- ness group. The multivariate F-ratio was 1.10, which was insignificant. 33 Table 6. Multivariate analysis based on handedness groups. SRH SLH WLH WRH F-ratio less than: Facial Recognition 20.56 21.12 20.19 18.75 1.87 .14 (.68) (.74) (.64) (.87) Block Design 40.81 38.00 41.56 36.37 1.99 .12 (1.46) (1.80) (1.28) (2.22) Object Assembly 35.50 35.12 33.87 32.75 .64 .59 (1.39) (1.38) (1.58) (1.87) Half-face Bias 15.94 14.12 15.62 16.25 1.11 .35 (.73) (1.01) (1.05) (.71) RT to Faces* +6.75 +13.81 +14.87 +14.l9 .20 .89 (8.51) (6.92) (9.36) (7.48) BUT to P1 Letters —10.29 -4.50 +4.37 -5.24 .55 .65 (10.8) (8.1) (7.6) (11.8) RT to N1 . ‘Letters +21.69 —4.94 -21.34 +2.31 1.80 .16 (14.3) (13.6) (11.1) (13.9) * For the reaction-time measures, negative numbers indicate a right-field bias for the fifth and sixth variables and a left-field bias for the last variable. In each case, therefore, a negative number denotes a mean contrary to prediction. The initial classification of persons into handedness groups ‘based on the dexterity test and hand-usage questionnaire showed no Promise for demonstrating differences in right-hemisphere performance. Irherefore the next logical step was to examine each component of 34 handedness separately in case one were masking the effects of another. The components examined were dexterity asymmetry, hand—usage question- naire, writing hand, and presence of left-handedness in the family. Sex of subject was also examined as a possible variable. Dexterity Subjects were classified into two groups: one group contained those whose writing hand was superior on the dexterity test, and the other group contained those whose hands were equal or whose writing hand performed more poorly. The mean scores on the seven dependent variables are shown in Table 7. No differences were significant at the .05 level. Table 7. Means for subjects separated on the basis of asymmetrical dexterity. Half- Face PI NI FR BD 0A face RT RT RT Writing Hand Superior (N - 42) 20.55 40.04 34.88 15.17 +10.19 -11.08 -.53 (.43) (1.07) (.84) (.55) (4.36) (5.80) (8.24) Writing Hand Inferior or Equal (N - 22) 19.41 37.54 33.23 16.09 +15.23 +7.04 -.30 (.52) (1.06) (1.1) (.55) (5.87) (5.81) (8.73) F-ratio 2.08 1.85 1.02 .96 .36 3.30 .00 p-value less than: .15 .18 .32 .33 .55 .07 .99 Multivariate F: 1.20, p‘ .32 35 Handednessjguestionnaire Subjects were divided next into two groups based on the weak-strong distinction: one group contained those whose question— naire scores were either between 14 and 30 or 54 and 70. These ranges defined a strong, consistent lateral preference for manual activities in contrast to those in the second group, whose scores were between 30 and 54 (see Table 8). No differences were significant at the .05 level. Table 8. Means for subjects separated on the basis of hand—usage questionnaire Half— Face PI NI FR BD 0A face RT RT RT "Strong" (N - 43) 20.35 38.05 34.40 15.12 +9.86 —7.21 +5.98 (.48) (1.06) (.98) (.53) (4.70) (5.8) (8 .9) "Weak" (N - 21) 19.76 41.52 34.14 16.24 +16.14 .00 -13.58 (.59) (1.47) (1.27) (.82) (7.46) (8.7) (9.0) F-ratio .53 3.59 .02 1.40 .54 .49 1.86 prvalue less than: .47 .06 .88 .24 .54 .49 .18 Multivariate F-ratio: 1.50, pc .19 Wtitingrfiand Table 9 shows the mean scores for subjects divided into two groups based on their writing hand alone. No differences were signifi- cant at the .05 level. 36 Table 9. Means for subjects separated on the basis of writing hand. Half- Face PI NI FR ED CA face RT RT RT Right- handed (N - 32) 19.66 38.60 34.12 16.09 +10.47 -9.63 +12.25 (.57) (1.35) (1.17) (.50) (5.6) (7.9) (9.9) Left- handed (N - 32) 20.60 39.78 34.50 14.87 +13.37 -.07 -l3.13 (.49) (1.15) (1.06) (.73) (5.7) (5.5) (8.8) F—ratio 1.77 .45 .06 1.88 .13 .98 3.64 p_1ess than: .18 .50 .81 .17 .72 .33 .06 Multivariate F-ratio: 1.55, p< .17 Familial Left-handedness Regardless of their own handedness, subjects were divided on the basis of whether or not at least one parent or full sibling was left-handed. The analysis revealed a strong difference in facial recognition ability between the groups; persons with left-handedness in the family had significantly lower scores “1,62 - 17.8, p< .0001). No other differences were significant (see Table 10). 37 Table 10. Means for subjects divided by presence of left—handedness in the family. Half- face PI NI FR ED CA face RT RT RT No familial left- handedness (N I 27) 21.81 40.07 35.44 15.44 +13.00 —12.34 +6.92 (.46) (1.26) (1.0) (.66) (6.7) (6.5) (11.9) Familial left- handedness (N - 37) 18.94 38.54 33.49 15.51 +11.13 +.62 -5.81 (.47) (1.21) (1.1) (.61) (4.9) (6.8) (7.9) F-ratio 17.78 .74 1.56 .00 .05 1.78 .86 p less than: .0001 .39 .22 .94 .81 .19 .36 The multivariate F—ratio was 2.71, p‘.01. Sex of Subject When subjects were separated on the basis of sex (see Table 11), females were found to show superior performance on the facial recognition test (Fl,62 - 7.14, p‘ .01). No other differences were significant at the .05 level. 38 Table 11. Group means for males and females. Half— Face PI NI FR ED CA face RT RT RT Males (N - 29) 19.10 40.55 33.00 15.44 +12.21 —1.42 +6.55 (.60) (1.35) (1.16) (.55) (4.8) (6.4) (9.0) Females (N - 35) 21.03 38.06 35.40 15.69 +11.68 -7.69 -6.23 (.43) (1.13) (1.02) (.69) (6.2) (7.0) (9.8) F-ratio 7.04 2.03 2.41 .24 .00 .41 .87 p less than: .01 .16 .12 .62 .95 .52 .35 Multivariate F-ratio: 2.44, p< .03. Correlations Many relationships revealed by the correlation analysis were already obtained from the multivariate analyses. Also, some of the correlations are not very meaningful; for example, the correlation between handedness group and dexterity is .71 because dexterity is one of the definers of the handedness groups. Yet several other new relationships emerged. For 64 subjects, a correlation of .24 is sig- nificant at the .05 level, and a correlation of .31 is significant at the .01 level. Table 12 presents the entire correlation matrix. Table 13 shows the correlations with three "combined" handedness variables, which were examined to test the possibility that a combination such as equal dexterity and familial left-handedness contributes a unique effect. 39 zoos n N .wsouum u H Hmovo u N .Hmavos: u H NNHN N>N - NNHH NNHN NNH - NNHN NNHN N>H - HNHN NNHN NNH - NNHH no» u N .0a a H OHmaom u N .mHma a H NNNH - N .NNNHN - H mes-N .NHs-N .NHN-N .NNN-H HmBHooa .oouosHsHHo our nusHoa .NN - z .Ho>uH Ho. «as on acoOHustHu NH Hm. “Ho>oH no. sou us assOHstme mH «N. we soHuwHouuoo o - NN N- N N NH N- NN N- N N- NN NN NH .Nmosc NN - o NN N NH NH- NH- NH- N N NH- HN NH NuHumuxmn N- o - NN- N N N NH NH NH- NH- NN- NH- HH NN Hz N NN NN- - o N- NH- N- HH- NH N- NH N NH NN Hm N N N o - N N N- N- N- H- N N N am «can NH NH N N- N - NH- N NH H N NH- N N moaN-NHNN N- NH- N NH- N NH- - NN NN NH- NH N NH- N Na NN NH- NH N- N- No NN - NH HH- NH- N NH- N NN N- NH- NH HH- N- NH NN NH - NN- NN NH NN- N am N N NH- NH N- H NH- HH- NN- - HN- N- oH N .H.H .aNN N N NH- N- H- N NH NH- NN HN- - NH N N sum oN NH- NN- NH N NH- N N NH N- NH - o N New: NaHuHuz NN HN NH- N N N NH- NH- NN- NH N- o - H NaouN Nana NH NH HH NH N N N N N N N N H .xHuuma GOHuMHouuou .NH OHnoH Table 13. Correlations of 40 combined "weak" variables.* l=SRH, 2=SLH, 3=WLH, 4=WRH l = right, 2 = left 1 = male, 2 = female 1 = no, 2 = yes high = LVF bias high - LVF bias high - LVF bias high - RVF bias 1 = unequal, 2 = equal 1 strong, 2 = weak 1 = no, 2 - yes 1 = no, 2 = yes 1 = no, 2 = yes persons who were classified "weak" on both the dexterity test and the handedness questionnaire (N - 11) persons classified "weak" on the dexterity test who also 14 15 16 Hand group 1 31 50 36 Writing hand 2 21 -12 19 Sex 3 8 -9 -l6 Fam. 1.h. 4 -3 43 43 FR 5 -4 -40 -19 BD 6 12 -17 12 0A 7 3 -28 —11 Half—face 8 17 12 11 Face RT 9 10 -8 -1 PI RT 10 25 27 8 NI RT 11 l -l -7 Dexterity 12 63 70 13 Quest. 13 65 14 72 Dex. & Quest. l4 -- 39 39 Dex. & fam. 1.h. 15 39 -- 32 Fam. 1.h. & Quest. 16 39 32 -- * #14: #15: had left-handedness in the family (N - 13) #16: persons classified "weak" on the handedness questionnaire who also had left-handedness in the family (N = 12) 41 Variable #5, facial recognition, showed the strongest overall relationship with the handedness variables: FR x handedness group: r = -.24. This indicates that persons in the WLH and WRH groups perform more poorly at facial recognition. FR x sex: r - .32. This indicates that females were better at facial recognition. FR x familial left-handedness: r = -.47. This shows that that the presence of left—handedness in the subject's family is strongly associated with a decrement in facial-recognition performance. FR x Object Assembly: r - .26. This suggests either that Object Assembly and the facial recognition test tap somewhat the same abilities, or that they are both correlated with a third undefined factor. FR x the combination of equal dexterity and familial left- handedness: r - -.40. Reaction-time Asymmetries On the whole, reaction time to faces shows lower correlations than any other variable with the handedness variables. Under the PI condition, a LVF reaction-time bias for the letter-pairs correlated .25 and .27 with two combined "weak" variables: #14 (equal dexterity and "weak" questionnaire) and #15 (equal dexterity and familial left-hand- edness. Thus the expected left-field superiority for the PI condition appeared most often in those persons who had other signs of being lg§g_ lateralized. For the NI condition, the highest correlations were with "Mil". FY-h‘é'L 42 writing hand (-.24) showing that right-handers more often show the expected right-field superiority, and with the PI condition (-.23) showing that persons with a left-field bias for the PI discrimination are less likely to show a right-field bias for the NI condition. 0f the two spatial tests, Block Design and‘Object Assembly, only Object Assembly seemed related to the variables of handedness, with a -.28 correlation with the combined variable #15, familial left-handedness and equal dexterity. The two spatial tests correlated .42 with each other. Inter—correlations of variables of handedness To clarify the relationships among the handedness variables, Table 14 contains only the variables directly concerned with handedness: Table 14. Correlations of handedness variables. writing Familial Equal "Weak" Hand 1.h. Dex. Quest. writing hand -- -9 -13 30 l-right, 2-left Familial left- handedness -9 -- 2 -6 1-no, 2-yes Equal ' Dexterity -l3 2 -- 26 l-unequal, 2-equal Handedness Questionnaire 30 -6 26 -- l-strong, 2-weak Since familial left-handedness had such a strong relationship with facial recognition, the average absolute differences in reaction times between visual fields were calculated and are shown in Table 15. 43 Table 15. Average absolute VF differences in RT for subjects divided on the basis of familial left-handedness (in milliseconds). Letters Faces PI NI (s.e.m.) (s.e.m.) Without familial left-handedness (N 3 27) 27.1 28.2 52.7 (4.3) (5.8) With familial left—handedness (N - 37) 25.0 30.6 35.9 (4.4) (5.1) A 2 x 2 analysis of variance performed on the two letter conditions only confirmed that the mean difference score found in persons without familial left-handedness (under N1 condition) was significantly greater than the other means (Fl,62 - 5.8, p‘ .03). This direct measure of degree of hemisphere difference revealed a more pronounced asymmetry for the first group, as predicted. 44 DISCUSSION Reaction-time testing With subjects divided into the four initial handedness groups, the results of the two letter-discrimination tasks showed no signifi- cant signs of the expected relationships. For the NI condition the largest right-field bias (22 msec.) was found in the SRH group, as expected, but it could not be rejected as due to chance. The PI con— dition produced no visual-field differences greater than 10 msec.; this lack of differences agrees with the results of Ledlow et a1. (1972) and with Gazzaniga (1970) but not with Geffen et al.(l972), who reported a definite right-hemisphere superiority for the PI condition. The analysis of PI-NI differences within hemispheres did not discriminate among the handedness groups, nor did examination of the size of the absolute differences between the hemispheres. The absolute— difference analysis was useful in examination of the familial handedness variable, and will be discussed later. The only other real difference shown by the letter reaction-time testing was between the PI and NI conditions disregarding visual-field: PI discriminations were on the average 118 msec. faster. Posner & Mitchell (1967) found a corresponding difference of 70 msec. and Geffen et a1. (1972) reported a difference of 110 msec. . One conclusion from the letter reaction—time testing is that a larger number of observations is necessary for accurately assessing reaction-time biases, in view of the great response variability. Although it was important to have a large number of subjects in the study for the other measures, the size of the sample plus the necessary hand 45 recording of the data made more extensive reaction-time testing im- practical. Each subject had to be trained in the letter-discrimination and the face discrimination tasks, with an adequate amount of practice, and then make nearly 200 recorded discriminations, all in one session. For the letter discrimination, since only the "same" responses were analyzable, the 64 response times per subject were divided into PI—NI conditions and also into the two visual-field conditions. Reaction times to faces Because the face-discrimination task had no PI-NI distinction, more observations per subject were available. The significantly faster LVF reaction time, averaging 13 msec., supports the earlier findings of Rizzolatti et a1. (15 msec.), of Geffen et a1. (25 msec.) and of Mosco- vitch & Catlin (5 msec.). The surprising result is that all four handed— ness groups had the same bias, of approximately the same magnitude. If this right—hemisphere superiority is truly due to right-hemisphere specialization for face processing, then the only logical conclusion is that localization of this function is unaffected by handedness. This conclusion is reinforCed by the consistently low correlations between face reaction times and all other handedness variables (Table 12); none is higher than .09. Since there is abundant evidence that handedness is associated with shifts in speech localization, theorists in this area have often assumed that when speech dominates in the right hemis- phere, the left hemisphere would naturally become dominant for non- verbal functions. Yet this assumption has never been rigorously tested or even seriously questioned, and it should be, judging from the sparse evidence available so far. 46 H. L. Dee's dichotic—listening study (1971) found a large difference in apparent cerebral lateralization between weakly and strongly lateralized subjects on a verbal dichotic-listening task. But for these same subjects, strength and side of handedness did not affect performance on a non-verbal dichotic-listening test composed of melodies. Melody recognition has been fairly well verified as a right-hemisphere function (Kimura, 1964; Milner, 1962; Shankweiler, 1966), yet in Dee's sample all handedness groups had a right-hemis- phere superiority for discriminating melodies, regardless of their performance on the verbal dichotic test. Another bit of evidence along this line was provided by Levy, Trevarthen and Sperry (1972) in their chimeric-face study with split- brain patients. They located one split—brain patient with language definitely in the right hemisphere, and his hemispheric bias for non- verbal perception resembled that of split-brain persons having speech in the left hemisphere: this subject's perception of faces and recog- nition of objects both showed a LVF bias, indicating right-hemisphere specialization for non-verbal as well as for verbal function. If right—hemisphere functions eventually prove to be less "movable," this would give new weight to the idea that non-verbal per- ception is impaired when speech function is not confined to the left hemisphere. If the two functions simply changed places, no interference would logically be expected, but if facial recognition and other visuo- spatial abilities are firmly embedded in the right hemisphere, then the presence of speech in that same hemisphere could interfere with the established neural organization. 47 There is some conflicting evidence, however; the matter was examined recently by McGloning & Davidson (1973), who correlated locali— zation of speech function (as determined by a dichotic listening test) with ability to estimate the number of dots in the left visual field-- supposedly a right-hemisphere function. The results did not show a decrement in LVF dot-enumeration ability in subjects who showed a left- ear superiority for digits. Also, Curry (1967) reported from a dichotic listening test that there was a "tendency," though non-significant, for the left hemisphere to be superior for non-verbal environmental sound discrimination (normally the right hemisphere excels) when the right hemisphere was superior for speech discrimination. The faster left-field reaction times to faces in the present study may, of course, have nothing to do with right—hemisphere speciali- zation, but no other explanation is immediately obvious. If some un- evenness of lighting existed in the tachistoscope, it should have affected the letter-discrimination task in the same way. If a left-to-right post-exposural scanning effect existed, as White (1971) has suggested to explain visual asymmetries for letter detection, then again there is no obvious reason why letter reaction times would not be similarly af- fected. So from the entire reaction—time testing this left-field bias is the most significant and the most interesting finding, precisely because it is unrelated to handedness. Multivariate analyses The initial division of subjects into the four handedness groups revealed no significant differences on any of the dependent variables. It should be noted that for all three of the right-hemisphere tests 48 (Block Design, Object Assembly, and facial recognition) the mean for the two "weak" groups combined was lower than the mean for the two "strong" groups combined; but this is meager support for the prediction and can only be called a non-significant trend. Facial recognition is the only variable tested before in this manner; the only difference from the details of the earlier study (Gilbert, 1973) was in the less restrictive dividing point between "weak" and "strong" on the handedness questionnaire: 30 in the present study and 40 in the previous one. The change admitted a few more strongly lateralized left-handers into the "weak" category. These results are not in strong agreement with Levy's 1969 study. There are three important differences to consider: 1) The present sample size is three times as large as Levy's sample. 2) Levy's sample was made up of Cal Tech graduate students, while the present subjects were college freshmen and sophomores. 3) Levy used the entire WAIS Performance scale for assessment of non-verbal ability, while the present study used only Object Assembly and Block Design subtests. Of the several Performance scale subtests, these two have the strongest likelihood of tapping right—hemisphere abilities exclusively, judging from the performance decrements in persons with right-hemisphere damage. However, Levy does not report which subtests contributed the most to the left-handed deficit; part of it could be due to the Digit Symbol sub- test for reasons having nothing to do with hemisphere dominance. Bonier & Henley (1961) found that performance on this test is impaired when the writing hand obscures the symbols which have just been written, which is what happens with most left-handers. In their study, rearranging the response sheet layout erased the difference between left-handers and 49 right-handers. Since the handedness-group subject differentiation based on dexterity and questionnaire performance yielded few strong dependent- variable differences, the subjects were re-ordered in other ways to examine the separate components of handedness. Dividing subjects into "weak" and "strong" on the basis of the dexterity test alone produced no significant differences. As with the handedness-group classification, the three right-hemisphere tests all produced lower means for the less lateralized group than for the strongly lateralized group, but the dif- ferences did not reach the necessary significance level. The same in- significant results occurred for the ordering of subjects on the basis of the handedness questionnaire and of the subject's writing hand. So the failure of the more general handedness-group classification to dis- criminate performance on any of these variables was not due to one handedness component masking another; none of the three components showed much relationship with the perceptual skills tested. Dividing subjects on the basis of sex did reveal a female su- periority for facial recognition. This result has appeared before (Howells, 1938; Witryol & Kaess, 1957; Goldstein & Chance, 1970; Cross, Cross and Daly, 1971) and has often been attributed to greater social interests on the part of the females. One more thorough recent study (Kent, 1972) showed that although females recognize females better, males recognize males better. In the present study, however, this interaction was not examined since females had an overall superiority. Familial left-handedness A strong right-hemisphere deficit appeared only when a variable more remotely connected with the subject's handedness was examined. 50 When subjects were divided on the basis of familial left-handedness, a large difference in facial recognition scores appeared. Oddly enough, the handedness of the subjects themselves did not seem to matter; the correlation between writing hand and facial recognition score was .17, with left-handers having slightly higher scores. The association of familial left—handedness with decreased cerebral lateralization has considerable experimental support. Hecaen & Sauguet's previously cited survey of components of aphasia following brain damage showed clearly that familial left-handedness is associated with cerebral ambilaterality: in "familial" left—handers, disturbances of oral language and of reading occurred equally often following either right or left-hemisphere damage. Left-handers with no left-handedness in the family showed almost no disturbances of language following right-hemisphere damage. The study did not include right-handers, un- fortunately. Two other studies are relevant to this issue. Hines & Satz (1971), using a letter-detection test, noticed that right-handers with familial left-handedness had much less visual-field asymmetry. Zurif & Bryden (1969) used both auditory and visual testing to examine the perceptual asymmetries of a group of left-handers and right-handers. Right-handers with familial left-handedness were not included, but the left-handers were divided into familial and non-familial groups. Zurif & Bryden's strongest finding was the diminished asymmetry of the familial left-handed group relative to the other two groups on both the dichotic listening and the tachistosc0pe letter-detection tasks, while non-famil— ial left-handers performed like the non-familial right-handers; both 51 groups showed right—ear and right visual field dominance for verbal stimuli. These results, taken together, suggest a genetic component to lateralization, especially considering that the handedness of the subject himself does not seem more important than the presence of left- handedness in the family. Annett (1964) has proposed such a model which includes explanations for ambilaterality, strong vs. weak left- handedness, and reversal of hemisphere function with dominant and re- cessive alleles. The model is discussed further in Satz, Fennel and Jones (1969); the predictions do not lead to the results reported here, however, and Miller (1971) also found that his results which supported Levy's "interference" theory were contrary to Annett's predictions. The present study offers more support for the hypothesis that familial left-handedness is associated with decreased lateralization: from the analysis of absolute reaction time differences in the NI letter-discrimination task, persons without familial left-handedness had significantly greater differences between their two visual-field scores. This measure is assumed to reflect greater lateralization of speech function, and the direct relationship neatly supports the con- nection between familial left-handedness and facial recognition ability, adding to the likelihood that lateralization, rather than some other unspecifiable factor, is the variable responsible for the drop in facial-recognition ability among the familial left-handedness group. Half-face bias Throughout these subject-classification manipulations, the half-face test scores remained firmly centered near the mean of 15, indicating no bias for one side of the face over the other. Considering 52 the consistent 60% bias obtained in previous samples (60% would be a score of 18 on this test) the lack of bias this time is very likely due to the face that subjects were not allowed to compare the compos— ites to the whole-face portrait, but only to their memory of it. The explanation for the half-face bias has been as follows: Subjects are somewhat subliminally aware of lateral asymmetries in the whole face, and because of the more direct visual transmission between the left visual field and right hemisphere, the components of the left side of the face