- Email: [email protected]
Shifts in Hemispheric Advantage During Familiarization with Complex Visual Patterns
SHIFTS IN HEMISPHERIC .ADVANTAGE DURING FAMILIARIZATION WITH COMPLEX VISUAL PATTERNS Phyllis Kittler1, Gerald Turkewitz1 and Elkhonon Goldbergl of Psychology, Hunter College of the City University of New York; Hunter College and Graduate Center of City University of New York and Departments of Pediatrics and Psychiatry, Albert Einstein College of Medicine; 3Medical College of Pennsylvania, Eastern Pennsylvania Psychiatric Institute, Philadelphia)
(~Department 2
Investigations of hemispheric specialization have indicated the usefulness of studying changes in lateral advantage over time as well as defining overall advantage for a particular stimulus class. Work in both visual and aural modal ities has demonstrated that a visual field or ear advantage, and the implied hemispheric advantage, can shift during the course of a single experiment (Goldberg and Costa, 1981). To further explore the scope of these findings, the current study examined shifts in visual field advantage for the recognition of complex and initially unfamiliar stimuli (Japanese ideograms). A number of theories 4ave been developed to account for the nature of shifts. Two of these are directly relevant to the current study. Goldberg and Costa ( 1981) propose that the right hemisphere is better at processing novel stimuli and at intermodal integration, whereas, the left hemisphere has an advantage in uni modal processing and in processing in terms of compact, well routinized codes. Their theory predicts a crucial role for the right hemisphere in the initial stages of information processing when there is not a predefined code. It also predicts a right to left shift as a function of increased competence. Consistent with this theory is a three stage theory of hemispheric advantage ip. perceptual processing elaborated by Ross-Kossak and Turkewitz (1986). According to this theory, the initial stage of perceptual processing involves a holistic approach which is better performed by the right hemisphere. In the second stage, processing is more analytic and feature oriented and favors the left hemisphere. During stage three there is an integration of the former stages and a synthesis of analytic and holistic approaches. This stage is better performed by the right hemisphere. Thus, with in_creasing familiarity, they predict a shift in hemispheric advantage from the right to the left and back again to the right. These theories provide the conceptual base for the current experiment, the purpose of which is to examine shifts in visual field advantage with increasing competence in the recognition of complex and initially unfamiliar members of a novel stimulus class. Kanji (Japanese ideograms) were chosen as the stimuli since they were clearly both complex and novel to American subjects. The current experiment was limited to female subjects since there is some evidence that women are more likely than men to exhibit patterns of ~ting visual field advantage (Turkewitz and Ross-Kossak, 1984; Ward and Ross, 1977). Cortex, (1989) 25, 27-32
28
Phyllis Kittler, Gerald Turkewitz and Elkhonon Goldberg
The current experiment closely followed the paradigm designed by Ross and Turkewi tz (1982). A complex class of visuospatial stimuli were selected in order to examine the generalizability of the Ross and Turkewitz findings (1982, 1984) on face recognition. MATERIALS AND METHOD
Subjects
Twenty females ranging in age from 20 to 49 years with a mean age of 32.4 years were tested. All subjects qualified as right handed based upon five activities determined to be good measures of handedness. None of the subjects were familiar with kanji. All subjects had a minimum visual acuity of 20/25 as assessed with the use of a Snellen chart. Apparatus
The apparatus was a Kodak Carousel650 H projector fitted with a Uniblitz Electronic Shutter Model 225L2ADTS attached to a Hunter Timer Model 111-C. A portable rear projection screen measuring 75 em by 52 em was set at one end of a large desk. A chin rest was bolted at the opposite end of the desk directly in front of the screen. Procedure
Four kanji, photographed as slides in black on a white background and rear projected on the screen, served as stimuli. They are shown in Figure 1. The projected kanji measured Fig. I -
Kanji used as stimuli.
-1
I
Shift in hemispheric advantage during familiarization
29
approximately 2.5-3.5 em in width, and 2.5-3.0 c'm in height. A black dot with a diameter of 1.9 em was attached to the center of the screen as a target for fixation. Subjects were seated 77 em directly in front of the screen with their chins in the rest. Stimuli were projected to the right or left of the dot for 66 ms. There was a distance of 5 em from the center of the dot to the center edge of the stimuli, thus subtending a visual angle of 3.r. Following the presentation of the target slide, a slide containing all four of the kanji was presented and the subjects indicated the target figure by pointing a light pen at it. Testing was divided into 6 blocks of 24 trials. Each stimulus was presented 6 times per block, 3 times to each visual field. Within the above constraints, stimulus presentation within each block was random except that every other block was the same as the preceding one with the visual field reversed. The randomization was different for each subject. RESULTS
Subjects clearly showed increased competence over the course of the recog nition task. The percent of errors declined from 56.0% in Block 1 to 30.2% by Block 6. The mean number of errors was 29.6 in the LVF and 26.5 in the RVF. A two factor repeated measure analysis of variance showed that the decline in errors across blocks was significant (F = 21.65; d.f. = 5, 95; p <.001) while the difference between right and left visual field errors approached significance (F = 4.24; d.f. = I, 19; p <.10). One of the main purposes of the current experiment was to examine indiv idual differences in hemispheric advantage when individuals are exposed to novel stimuli. In past work significant differences have been found between subjects beginning with a LVFA and those beginning with a RVFA (Ross and Turkewitz, 1982; Turkewitz and Ross, 1983). Thus, the subjects in the current experiment were divided accordingly. When a difference of one or more errors on Block 1 was used as the criterion, nine subjects had an initial LVFA, seven had an initial RVFA and four had no advantage. The groups did not differ significantly in number of errors (initial LVFA mean= 55.2; initial RVFA mean= 60.7; t = .67; d.f. = 14). In that one of the purposes of this experiment was to look at shifts in laterality with increased competence on the recognition task, a measure of laterality (N of correct LVF - N of correct RVF X lOO) total correct was calculated for all subjects and analyzed across the six blocks. See Figure 2. To determine whether the groups identified by different directions of initial advantage differed in performance across blocks, a two by six ANOVA with direction of initial VFA (2levels) as an independent factor and block (6levels) as a repeated measure was done. Because of the ratio nature of the data they were subjected to an arc-sine transformation prior to being utilized in the ANOVA It should be noted that the results achieved without the transformation were vir tually identical to those achieved with it. The analysis indicated no significant main effects for either initial direction of VFA or for blocks but a significant interaction (F = 6.80; d.f. = 5, 70; p <.001) between them. To explicate this
Phyllis Kittler, Gerald Turkewitz and Elkhonon Goldberg
30
.........
u
0 a.. ....__,
0 - 0 all subjects (n=20) • - • initial RVFA subjects (n=7) t::.-t::. initial LVFA subjects (n=9)
30
Q)
01 0
+'
c
20
0
>
"'0 0 "'0
10
:+::
c:J
0
5
-10
0
Qj
Ul
c 0
Q)
:::::!:
-20 -30
0
2
3
4
5
6
Blocks Fig. 2 subjects.
Visualfield advantage across blocksfor all subjects, initial R VFA subjects and initial L VFA
interaction separate analysis of RVF and LVF errors were done using two factor repeated measure ANOVAs. The principal finding of these analyses was a sig nificant interaction of visual field errors over blocks for the initial LVFA group (F = 3.63; d.f. = 5, 40; p <.01) and no such interaction for the initial RVFA group (F = 1.24; d.f. = 5, 30). Separate ANOVA's of visual field advantage over blocks were done for the two groups. These resulted in a non-significant block effect for the initial RVFA group (F = 2.14; d.f. = 5, 30) and a highly significant effect for the initial LVFA group (F = 6.48; d.f. = 5, 40; p <.001). This is consistent with the finding of a significant interaction between RVF and LVF errors and blocks for this group. It should be noted that this finding of a differential block effect in the two groups makes it difficult if not impossible to sustain the otherwise plausible alternative of regression to the mean accounting for any differences between groups deter mined on a post hoc basis. To further determine whether there were systematic shifts in visual field advantage over time, trend analyses were done. For the initial LVFA group the trend analysis revealed both a significant linear trend (F = 17.7; d.f. = 1, 40; p <.001) and a significant quadratic trend (F = 5.29; d.f. 1, 40; p <.05). For the initial RVFA group, there was only a significant quadratic trend (F = 4.18; d.f. = 1, 30; p <.05).
Shift in hemispheric advantage during familiarization
31
DISCUSSION
Results from the current experiment support the observation of individual differences in visual field advantage in the perception of complex, and initially unfamiliar stimuli. In this way, they expand the Ross-Kossak and Turkewitz (1986) findings on face recognition. The results are also consistent with theories of a right to left shift in hemispheric advantages with increased competence or practice. When the subjects were divided based upon individual differences in Block 1 performance, the results from the initial LVFA group support the Goldberg and Costa ( 1981) theory of a right to left hemisphere shift with increased competence, as expressed in the significant linear trend in the visual field advantage. This is also consistent with previous work (Ross and Turkewitz, 1982; Turkewitz and Ross, 1983; Turkewitz and Ross-Kossak, 1984) in which a similar division of subjects resulted in different patterns of shifts for each group, with the initial LVFA group showing the implied right to left hemisphere shift. The initial LVFA group shifts its advantage from one hemisphere to the other (right to left), and it seems to exhibit a more consistent pattern of hemispheric advantage thah is shown by the initial RVFA group. The consistency lies in the fact that the shift itself is monotonic, i.e., the group begins with a right hemisphere advantage in Block 1, and moves steadily toward an increasingly greater left hemisphere advantage as familiarity increases, with the cross over from right hemisphere to left occurring around Block 2. Although there is a diminution of the degree of left hemisphere advantage between Blocks 3 and 4 which is evi denced by the significant quadratic trend, the linear trend supersedes this and the degree of left hemisphere advantage continues to increase. On the other hand, although the initial RVFA group shows a systematic change in magnitude of hemispheric advantage as evidenced by the significant quadratic trend, members of the group do not actually switch the direction of their hemispheric advantage (see Figure 2). Also, the direction of change is not monotonic, so that there is movement toward the left visual field and then back toward the right visual ~d
.
While supporting the theory of a right-left shift in hemispheric advantage, the current work does not seem to provide evidence for the full complement of shifts in hemispheric advantage from right to left to right, postulated by Ross-Kossak and Turkewitz (1986) even in the initial LVFA group. This may be because kanji recognition does not require the level of integration that face recognition does, or because given the unfamiliarity of kanji figures a highly integrated method of recognizing them did not develop during the limited number of exposures pre sented. In conclusion, the current work clearly increases the body of evidence which points to a right to left shift in hemispheric advantage in the processing of a novel stimulus class. It also points toward individual differences in the patterns of shifting that occur during perceptual processing of previously unfamiliar stimuli. However, based on the current experiment, it can only be said that these shifts occur in the presence of increased competence. There is no direct evidence provided on the more precise nature of these shifts.
32
Phyllis Kittler, Gerald Turkewitz and Elkhonon Goldberg ABSTRACT
Twenty females unfamiliar with kanji were given a recognition task involving tachis toscopic presentation of kanji to the right and left visual fields. Repeated exposure to these novel stimuli resulted in significantly increased competence at the task. To analyze different approaches to the task, subjects were divided into an initial left visual field advantage and an initial right visual field advantage group. Over the course of the experiment subjects in the initial left visual field advantage group shifted from a left to a right visual field advantage, showing both a linear trend and a quadratic trend, while the initial right visual field advantage group showed only a quadratic trend. The results are consistent with observations in other novel stimuli situations which have demonstrated a shift in hemispheric advantage from right to left with increased competence and which have demonstrated individual differences in the pattern of shifting hemispheric advan tage. REFERENCES
GoLDBERG, E., and CosTA, L. Hemispheric differences in the acquisition and use of descriptive systems. Brain and Language, 14: 144-173, 1981. Ross, P., and TuRKEWITZ, G. Changes in hemispheric advantage in processing information with increasing stimulus familiarization. Cortex, 18: 489-499, 1982. Ross-KossAK, P., and TURKEWITZ, G. Relationship between changes in hemispheric advantage during familiarization to faces and proficiency in facial recognition. Neuropsychologia, 22: 471-477, 1984. Ross-KossAK, P., and TURKEWITZ, G. A micro and macro developmental view of the nature of changes in complex information processing: a consideration of changes in hemispheric advan tage during familiarization. In R. Bruyer (Ed.), Neuropsychology of Facial Expression. Hillsdale, NJ: Erlbaum, 1986. TURKEWITZ, G., and Ross, P. The development of a general strategy for the processing of facial information. Cortex, 19: 179-185, 1983. TuRKEWITZ, G., and Ross-KossAK, P. Multiple modes ofright hemisphere information processing: age and sex differences in facial recognition. Developmental Psychology, 20: 95-103, 1984. WARD, T.B., and Ross, L.E. Laterality differences and practice effects under central backward masking conditions. Memory and Cognition, 5: 221-227, 1977. Gerald Turkewitz, Department of Pediatries and Psychiatry, Albert Einstein College of Medicine, Bronx, NY I 0461, U.S.A.
(~Department 2
Investigations of hemispheric specialization have indicated the usefulness of studying changes in lateral advantage over time as well as defining overall advantage for a particular stimulus class. Work in both visual and aural modal ities has demonstrated that a visual field or ear advantage, and the implied hemispheric advantage, can shift during the course of a single experiment (Goldberg and Costa, 1981). To further explore the scope of these findings, the current study examined shifts in visual field advantage for the recognition of complex and initially unfamiliar stimuli (Japanese ideograms). A number of theories 4ave been developed to account for the nature of shifts. Two of these are directly relevant to the current study. Goldberg and Costa ( 1981) propose that the right hemisphere is better at processing novel stimuli and at intermodal integration, whereas, the left hemisphere has an advantage in uni modal processing and in processing in terms of compact, well routinized codes. Their theory predicts a crucial role for the right hemisphere in the initial stages of information processing when there is not a predefined code. It also predicts a right to left shift as a function of increased competence. Consistent with this theory is a three stage theory of hemispheric advantage ip. perceptual processing elaborated by Ross-Kossak and Turkewitz (1986). According to this theory, the initial stage of perceptual processing involves a holistic approach which is better performed by the right hemisphere. In the second stage, processing is more analytic and feature oriented and favors the left hemisphere. During stage three there is an integration of the former stages and a synthesis of analytic and holistic approaches. This stage is better performed by the right hemisphere. Thus, with in_creasing familiarity, they predict a shift in hemispheric advantage from the right to the left and back again to the right. These theories provide the conceptual base for the current experiment, the purpose of which is to examine shifts in visual field advantage with increasing competence in the recognition of complex and initially unfamiliar members of a novel stimulus class. Kanji (Japanese ideograms) were chosen as the stimuli since they were clearly both complex and novel to American subjects. The current experiment was limited to female subjects since there is some evidence that women are more likely than men to exhibit patterns of ~ting visual field advantage (Turkewitz and Ross-Kossak, 1984; Ward and Ross, 1977). Cortex, (1989) 25, 27-32
28
Phyllis Kittler, Gerald Turkewitz and Elkhonon Goldberg
The current experiment closely followed the paradigm designed by Ross and Turkewi tz (1982). A complex class of visuospatial stimuli were selected in order to examine the generalizability of the Ross and Turkewitz findings (1982, 1984) on face recognition. MATERIALS AND METHOD
Subjects
Twenty females ranging in age from 20 to 49 years with a mean age of 32.4 years were tested. All subjects qualified as right handed based upon five activities determined to be good measures of handedness. None of the subjects were familiar with kanji. All subjects had a minimum visual acuity of 20/25 as assessed with the use of a Snellen chart. Apparatus
The apparatus was a Kodak Carousel650 H projector fitted with a Uniblitz Electronic Shutter Model 225L2ADTS attached to a Hunter Timer Model 111-C. A portable rear projection screen measuring 75 em by 52 em was set at one end of a large desk. A chin rest was bolted at the opposite end of the desk directly in front of the screen. Procedure
Four kanji, photographed as slides in black on a white background and rear projected on the screen, served as stimuli. They are shown in Figure 1. The projected kanji measured Fig. I -
Kanji used as stimuli.
-1
I
Shift in hemispheric advantage during familiarization
29
approximately 2.5-3.5 em in width, and 2.5-3.0 c'm in height. A black dot with a diameter of 1.9 em was attached to the center of the screen as a target for fixation. Subjects were seated 77 em directly in front of the screen with their chins in the rest. Stimuli were projected to the right or left of the dot for 66 ms. There was a distance of 5 em from the center of the dot to the center edge of the stimuli, thus subtending a visual angle of 3.r. Following the presentation of the target slide, a slide containing all four of the kanji was presented and the subjects indicated the target figure by pointing a light pen at it. Testing was divided into 6 blocks of 24 trials. Each stimulus was presented 6 times per block, 3 times to each visual field. Within the above constraints, stimulus presentation within each block was random except that every other block was the same as the preceding one with the visual field reversed. The randomization was different for each subject. RESULTS
Subjects clearly showed increased competence over the course of the recog nition task. The percent of errors declined from 56.0% in Block 1 to 30.2% by Block 6. The mean number of errors was 29.6 in the LVF and 26.5 in the RVF. A two factor repeated measure analysis of variance showed that the decline in errors across blocks was significant (F = 21.65; d.f. = 5, 95; p <.001) while the difference between right and left visual field errors approached significance (F = 4.24; d.f. = I, 19; p <.10). One of the main purposes of the current experiment was to examine indiv idual differences in hemispheric advantage when individuals are exposed to novel stimuli. In past work significant differences have been found between subjects beginning with a LVFA and those beginning with a RVFA (Ross and Turkewitz, 1982; Turkewitz and Ross, 1983). Thus, the subjects in the current experiment were divided accordingly. When a difference of one or more errors on Block 1 was used as the criterion, nine subjects had an initial LVFA, seven had an initial RVFA and four had no advantage. The groups did not differ significantly in number of errors (initial LVFA mean= 55.2; initial RVFA mean= 60.7; t = .67; d.f. = 14). In that one of the purposes of this experiment was to look at shifts in laterality with increased competence on the recognition task, a measure of laterality (N of correct LVF - N of correct RVF X lOO) total correct was calculated for all subjects and analyzed across the six blocks. See Figure 2. To determine whether the groups identified by different directions of initial advantage differed in performance across blocks, a two by six ANOVA with direction of initial VFA (2levels) as an independent factor and block (6levels) as a repeated measure was done. Because of the ratio nature of the data they were subjected to an arc-sine transformation prior to being utilized in the ANOVA It should be noted that the results achieved without the transformation were vir tually identical to those achieved with it. The analysis indicated no significant main effects for either initial direction of VFA or for blocks but a significant interaction (F = 6.80; d.f. = 5, 70; p <.001) between them. To explicate this
Phyllis Kittler, Gerald Turkewitz and Elkhonon Goldberg
30
.........
u
0 a.. ....__,
0 - 0 all subjects (n=20) • - • initial RVFA subjects (n=7) t::.-t::. initial LVFA subjects (n=9)
30
Q)
01 0
+'
c
20
0
>
"'0 0 "'0
10
:+::
c:J
0
5
-10
0
Qj
Ul
c 0
Q)
:::::!:
-20 -30
0
2
3
4
5
6
Blocks Fig. 2 subjects.
Visualfield advantage across blocksfor all subjects, initial R VFA subjects and initial L VFA
interaction separate analysis of RVF and LVF errors were done using two factor repeated measure ANOVAs. The principal finding of these analyses was a sig nificant interaction of visual field errors over blocks for the initial LVFA group (F = 3.63; d.f. = 5, 40; p <.01) and no such interaction for the initial RVFA group (F = 1.24; d.f. = 5, 30). Separate ANOVA's of visual field advantage over blocks were done for the two groups. These resulted in a non-significant block effect for the initial RVFA group (F = 2.14; d.f. = 5, 30) and a highly significant effect for the initial LVFA group (F = 6.48; d.f. = 5, 40; p <.001). This is consistent with the finding of a significant interaction between RVF and LVF errors and blocks for this group. It should be noted that this finding of a differential block effect in the two groups makes it difficult if not impossible to sustain the otherwise plausible alternative of regression to the mean accounting for any differences between groups deter mined on a post hoc basis. To further determine whether there were systematic shifts in visual field advantage over time, trend analyses were done. For the initial LVFA group the trend analysis revealed both a significant linear trend (F = 17.7; d.f. = 1, 40; p <.001) and a significant quadratic trend (F = 5.29; d.f. 1, 40; p <.05). For the initial RVFA group, there was only a significant quadratic trend (F = 4.18; d.f. = 1, 30; p <.05).
Shift in hemispheric advantage during familiarization
31
DISCUSSION
Results from the current experiment support the observation of individual differences in visual field advantage in the perception of complex, and initially unfamiliar stimuli. In this way, they expand the Ross-Kossak and Turkewitz (1986) findings on face recognition. The results are also consistent with theories of a right to left shift in hemispheric advantages with increased competence or practice. When the subjects were divided based upon individual differences in Block 1 performance, the results from the initial LVFA group support the Goldberg and Costa ( 1981) theory of a right to left hemisphere shift with increased competence, as expressed in the significant linear trend in the visual field advantage. This is also consistent with previous work (Ross and Turkewitz, 1982; Turkewitz and Ross, 1983; Turkewitz and Ross-Kossak, 1984) in which a similar division of subjects resulted in different patterns of shifts for each group, with the initial LVFA group showing the implied right to left hemisphere shift. The initial LVFA group shifts its advantage from one hemisphere to the other (right to left), and it seems to exhibit a more consistent pattern of hemispheric advantage thah is shown by the initial RVFA group. The consistency lies in the fact that the shift itself is monotonic, i.e., the group begins with a right hemisphere advantage in Block 1, and moves steadily toward an increasingly greater left hemisphere advantage as familiarity increases, with the cross over from right hemisphere to left occurring around Block 2. Although there is a diminution of the degree of left hemisphere advantage between Blocks 3 and 4 which is evi denced by the significant quadratic trend, the linear trend supersedes this and the degree of left hemisphere advantage continues to increase. On the other hand, although the initial RVFA group shows a systematic change in magnitude of hemispheric advantage as evidenced by the significant quadratic trend, members of the group do not actually switch the direction of their hemispheric advantage (see Figure 2). Also, the direction of change is not monotonic, so that there is movement toward the left visual field and then back toward the right visual ~d
.
While supporting the theory of a right-left shift in hemispheric advantage, the current work does not seem to provide evidence for the full complement of shifts in hemispheric advantage from right to left to right, postulated by Ross-Kossak and Turkewitz (1986) even in the initial LVFA group. This may be because kanji recognition does not require the level of integration that face recognition does, or because given the unfamiliarity of kanji figures a highly integrated method of recognizing them did not develop during the limited number of exposures pre sented. In conclusion, the current work clearly increases the body of evidence which points to a right to left shift in hemispheric advantage in the processing of a novel stimulus class. It also points toward individual differences in the patterns of shifting that occur during perceptual processing of previously unfamiliar stimuli. However, based on the current experiment, it can only be said that these shifts occur in the presence of increased competence. There is no direct evidence provided on the more precise nature of these shifts.
32
Phyllis Kittler, Gerald Turkewitz and Elkhonon Goldberg ABSTRACT
Twenty females unfamiliar with kanji were given a recognition task involving tachis toscopic presentation of kanji to the right and left visual fields. Repeated exposure to these novel stimuli resulted in significantly increased competence at the task. To analyze different approaches to the task, subjects were divided into an initial left visual field advantage and an initial right visual field advantage group. Over the course of the experiment subjects in the initial left visual field advantage group shifted from a left to a right visual field advantage, showing both a linear trend and a quadratic trend, while the initial right visual field advantage group showed only a quadratic trend. The results are consistent with observations in other novel stimuli situations which have demonstrated a shift in hemispheric advantage from right to left with increased competence and which have demonstrated individual differences in the pattern of shifting hemispheric advan tage. REFERENCES
GoLDBERG, E., and CosTA, L. Hemispheric differences in the acquisition and use of descriptive systems. Brain and Language, 14: 144-173, 1981. Ross, P., and TuRKEWITZ, G. Changes in hemispheric advantage in processing information with increasing stimulus familiarization. Cortex, 18: 489-499, 1982. Ross-KossAK, P., and TURKEWITZ, G. Relationship between changes in hemispheric advantage during familiarization to faces and proficiency in facial recognition. Neuropsychologia, 22: 471-477, 1984. Ross-KossAK, P., and TURKEWITZ, G. A micro and macro developmental view of the nature of changes in complex information processing: a consideration of changes in hemispheric advan tage during familiarization. In R. Bruyer (Ed.), Neuropsychology of Facial Expression. Hillsdale, NJ: Erlbaum, 1986. TURKEWITZ, G., and Ross, P. The development of a general strategy for the processing of facial information. Cortex, 19: 179-185, 1983. TuRKEWITZ, G., and Ross-KossAK, P. Multiple modes ofright hemisphere information processing: age and sex differences in facial recognition. Developmental Psychology, 20: 95-103, 1984. WARD, T.B., and Ross, L.E. Laterality differences and practice effects under central backward masking conditions. Memory and Cognition, 5: 221-227, 1977. Gerald Turkewitz, Department of Pediatries and Psychiatry, Albert Einstein College of Medicine, Bronx, NY I 0461, U.S.A.