Genetic and developmental factors in spontaneous selective attention: a study of normal twins

Psychiatry Research 71 Ž1997. 163]174

Genetic and developmental factors in spontaneous selective attention: a study of normal twins Marina Myles-Worsley U , Hilary Coon Department of Psychiatry, Uni®ersity of Utah School of Medicine, 50 North Medical Dri®e, Salt Lake City, UT 84132, USA Received 20 August 1996; revised 8 April 1997; accepted 21 April 1997

Abstract The Spontaneous Selective Attention Task ŽSSAT. is a visual word identification task designed to measure the type of selective attention that occurs spontaneously when there are multiple stimuli, all potentially relevant, and insufficient time to process each of them fully. These are conditions which are common in everyday life. SSAT performance is measured by word identification accuracy, first under a baseline divided attention condition with no predictability, then under a selective attention condition with partial predictability introduced via word repetition. Accuracy to identify novel words in the upper location which becomes partially predictable ŽP words. vs. the lower location which remains non-predictable ŽN words. can be used to calculate a baseline performance index and a PrN ratio measure of selective attention. The SSAT has been shown to identify an attentional abnormality that may be useful in the development of an attentional endophenotype for family]genetic studies of schizophrenia. This study examined age and genetic effects on SSAT performance in normal children in order to evaluate whether the SSAT has the potential to qualify as a candidate endophenotype for schizophrenia in studies of at-risk children. A total of 59 monozygotic twin pairs and 33 same-sex dizygotic twin pairs ranging from 10 to 18 years of age were tested on the SSAT, a Continuous Performance Test ŽCPT., a Span of Apprehension Test ŽSPAN. and a full-scale IQ test. Baseline performance on the SSAT, which was correlated with verbal IQ and SPAN performance, improved with age but showed no significant heritability. The PrN selectivity ratio was stable over the 10]18-year age range, was not significantly correlated with IQ, CPT, or SPAN performance, and its heritability was estimated to be 0.41. These findings suggest that the PrN selectivity ratio measured by the SSAT may be useful as a vulnerability marker in studies of children born into families segregating schizophrenia. Q 1997 Elsevier Science Ireland Ltd. Keywords: Schizophrenia; Children-at-risk; Vulnerability marker

U

Corresponding author. Tel.: q1 801 5818551; fax: q1 801 5815604; e-mail: [email protected]

0165-1781r97r$17.00 Q 1997 Elsevier Science Ireland Ltd. All rights reserved. PII S0165-1781Ž97.00042-5

164

M. Myles-Worsley, H. Coon r Psychiatry Research 71 (1997) 163]174

1. Introduction A large body of research based on family, twin, and adoption studies has convincingly shown that genetic factors play a role in the etiology of schizophrenia Žsee Gottesman et al., 1987; Kendler and Diehl, 1993, for reviews.. However, finding the genetic causes of schizophrenia using standard linkage techniques and a disease phenotype which has reduced penetrance and a nonMendelian inheritance pattern has proven to be difficult ŽKendler and Diehl, 1993.. A number of family]genetic studies have begun to examine possible ‘endophenotypes’ for schizophrenia in addition to the disease status phenotype Že.g. Grove et al., 1992; Coon et al., 1993.. The term ‘endophenotype’ ŽGottesman and Shields, 1972. refers to some endogenous sign of the schizophrenia genotype, i.e. a psychobiological dysfunction that is more closely related to the genes causing schizophrenia than the overt symptoms used to diagnose the illness. In order to function as an endophenotype in genetic linkage studies of schizophrenia, the abnormality must be specific to schizophrenia, show evidence of heritability, and be capable of identifying unaffected relatives who are gene carriers. An endophenotype that can fulfill these criteria has the potential to serve as a vulnerability marker in children born into families segregating schizophrenia who are at heightened genetic risk for the illness. Various attentional problems found in adult schizophrenic patients have been targeted as possible endophenotypes for schizophrenia. Two of the most widely studied attentional abnormalities are a deficit in sustained focused attention as measured by the Continuous Performance Test ŽCPT; e.g. Orzack and Kornetsky, 1966, 1971; Wohlberg and Kornetsky, 1973; Walker, 1981; Cornblatt et al., 1989. and deficit performance on the Span of Apprehension ŽSPAN., a visual search task Že.g. Neale et al., 1969; Neale, 1971; Asarnow et al., 1977; Asarnow and MacCrimmon, 1978, 1981; Strauss et al., 1987.. The CPT presents a long series of visual stimuli which are difficult to process, either because of visual degradation or brief exposure duration, and requires subjects to detect predesignated targets.

CPT deficits are more likely to occur in the offspring of parents with schizophrenia than in normal and psychiatric control children ŽNuechterlein, 1983; Cornblatt and Erlenmeyer-Kimling, 1985; Cornblatt et al., 1989, 1992.. A normative study by Cornblatt et al. Ž1988. has shown that the d9 signal detection component of CPT performance is heritable. Heritabilities based on family transmission patterns were estimated to be 0.39 for number stimuli and 0.49 for shape stimuli. The SPAN is a visual search task requiring subjects to indicate which of two target letters Žusually T and F. appeared in a series of briefly exposed displays containing a variable number of distractor letters. The SPAN identifies attentional deficits in children-at-risk ŽAsarnow et al., 1977. and childhood-onset schizophrenia patients ŽAsarnow et al., 1986.. Bartfai et al. Ž1991. in a study of normal twins estimated the heritability of SPAN performance to be 0.65. However, the total genetic variation in SPAN performance was the sum of 0.37 genetic variation shared with IQ and only 0.28 genetic variation unique to SPAN. We recently reported that a new laboratory attention task, the spontaneous selective attention task ŽSSAT., identifies an attentional abnormality that may qualify as an endophenotype in genetic studies of schizophrenia ŽMyles-Worsley et al., 1991a,b.. The SSAT is a visual word identification task that measures the type of selective attention that occurs spontaneously when there are multiple stimuli, all potentially relevant, and insufficient time to process them all fully, conditions which are common in everyday life. Two words are flashed on the screen at the same time, one above the other, so briefly that the average observer can identify only one of the two words. Periodically, the subject’s ability to identify either the upper or the lower word is tested, and attention is measured by the accuracy with which novel words can be identified. First, both the upper and lower stimulus locations are non-predictable: all words that appear are novel. Under these conditions, which we refer to as the divided attention segment of the task, subjects are expected to divide their attention between the two stimulus locations. Then, to elicit selective attention, a pattern of predictability is introduced into one of

M. Myles-Worsley, H. Coon r Psychiatry Research 71 (1997) 163]174

the two locations via word repetition. The same word Ž‘winter’. appears 65% of the time in the upper cell. Under these conditions, termed the selective attention segment of the task, normal subjects pay more attention to the non-predictable than to the predictable source of information. This normal pattern of selectivity in which attention is withdrawn from the predictable and directed toward the non-predictable source of information serves an important adaptive function for the individual. It ensures that attention will be drawn to change, i.e. to anything new or unexpected in the usual flow of everyday experience, so that the individual’s knowledge can be updated accordingly. Under the divided attention condition, schizophrenic patients are just as accurate as normal subjects in identifying the word stimuli, indicating no deficits in performing the task. It is performance under the selective attention condition of the SSAT that distinguishes schizophrenic patients Ž85% abnormal. from bipolar patients Ž30% abnormal. and normal controls Ž20% abnormal. ŽMyles-Worsley et al., 1991a.. Furthermore, abnormal performance under the selective attention condition occurs in 44% of the first-degree relatives of schizophrenic patients vs. only 12% of the first-degree relatives of bipolar patients ŽMyles-Worsley et al., 1991b.. These findings suggest that the attentional abnormality identified by the SSAT under the selective attention condition is specific to schizophrenia and may be inherited by the unaffected relatives of schizophrenic patients. In order to serve as a vulnerability marker for an underlying genetic predisposition to schizophrenia in the offspring of schizophrenic patients, the test used to measure the trait must be appropriate for and valid in younger as well as older children and adolescents and the trait must show significant heritability. The purpose of our study was to determine whether the SSAT can fulfill these two criteria. We studied a sample of 10]18-year-old twins which allowed us to: Ža. examine the effects of age on measures of spontaneous selective attention; and Žb. test for genetic influences on SSAT performance by comparing the performance of

165

monozygotic ŽMZ. and dizygotic ŽDZ. twins. In addition to the SSAT, the twin subjects were tested on the CPT and the SPAN so that genetic and developmental effects on attentional performance could be compared across the three tests. A full-scale IQ test was included in the testing battery so that the degree of association between IQ and the attentional measures could be assessed and estimates of the heritability of IQ could serve as a standard against which heritability estimates for the SSAT, CPT, and SPAN performance measures could be evaluated. 2. Methods 2.1. Subjects The subjects were 59 MZ twin pairs Ž26 male and 33 female. and 33 same-sex DZ twin pairs Ž17 male and 16 female.. The age of the subjects ranged from 10 to 18 years Žmean age s 14.56 years, S.D.s 2.46 years.. A total of 1437 same-sex twin pairs were identified through birth registrations on file at the Utah Bureau of Vital Statistics. Using the father’s name, a current listing in a local telephone directory was found for 345 of these twin pairs, and 204 Ž59%. of calls to these telephone numbers resulted in a personal contact with a parent. When invited to participate in a twin study of information processing, 124 parents Ž61%. expressed an interest, but appointments for screening and testing were successfully kept by only 102 twin pairs. All twin pairs were screened by one of three doctoral level clinical psychologists who administered the IQ test for professionally diagnosed psychiatric illness, behavioral problems as assessed by the Child Behavior Checklist ŽAchenbach and Edelbrock, 1983., learning disabilities, and organic impairments that could affect cognitive performance. This screening reduced the sample from 102 to 96 twin pairs. 2.2. Measures 2.2.1. Zygosity testing Zygosity was determined by a DNA probe technique estimated to be 99.999% accurate. Twin

166

M. Myles-Worsley, H. Coon r Psychiatry Research 71 (1997) 163]174

pairs were considered monozygotic only if concordant on all six of the highly polymorphic VNTR markers that were examined: YNH-24 ŽChromosome 2.; EFD-64 ŽChromosome 3.; EFD-126.3 ŽChromosome 9.; CMM-101 ŽChromosome 14.; JCZ-50 ŽChromosome 17.; and JCZ-3.1 ŽChromosome 19.. 2.2.2. IQ testing The WISC-R ŽWechsler, 1974. was administered to subjects under the age of 16.5 years, and the WAIS-R ŽWechsler, 1981. was administered to those who were 16.5 years or older at the time of testing. 2.2.3. Attention testing All three attention tests were conducted in a private cubicle that was equipped with an IBM color monitor ŽModel RGB 5153. and a microphone. The display was controlled by an IBM-XT Model 286 computer system situated in an adjoining room where an experimenter monitored the subject’s vocal responses for accuracy over stereo headphones. The subject was seated approximately 60 cm from the monitor, and the height of the monitor was adjusted to align the subject’s eyes with the center of the screen. The potential effects of task order on performance were not studied. Given that the experimental design had four grouping factors Žtwin pair, zygosity, gender, and age., the sample sizes within each group were too small to permit meaningful analysis of order effects. Therefore, we held the order of the three tasks constant across all subjects. First, a Continuous Performance Test ŽCPT. was administered. At the time thisstudy was conducted, the software to present more standard versions of the CPT such as the degraded stimulus CPT ŽNuechterlein, 1983. or the CPT, Identical Pairs Version ŽCornblatt et al., 1988. was not available to researchers in the field. The version of the CPT that we used was designed to correspond with descriptions of the degraded stimulus CPT as presented in the published literature. The CPT we used presented one of the numbers 2]9 centered within three rows of Xs for 50 ms before being backward masked by an X. The subject’s

task was to say the number ‘5’ aloud whenever it appeared. The Xs surrounding the number stimulus generated visual noise, and backward masking the stimulus limited its exposure duration, thereby increasing the difficulty of the task. After 30 practice trials, the subject was required to perform the CPT for 10 min in order to evaluate sustained focused attention. The hit rate and false alarm rate for each subject were used to calculate a d9 signal detection index of the subject’s ability to discriminate between target and non-target stimuli and the Beta index of the subject’s decision criteria ŽHochhaus, 1972.. The second test was a forced-choice Span of Apprehension ŽSPAN. which presented eight letters in randomly selected locations of an imaginary 4 = 4 matrix centered on the computer screen. Each eight-letter display appeared for 70 ms and subtended a visual angle of less than 28 vertically and horizontally. The subject’s task was to identify which one of two possible target letters ŽT or F. appeared in each display and to guess if unsure. After 10 practice trials, a series of 36 displays were presented, one every 1200 ms, and target identification accuracy was measured. The third test was the Spontaneous Selective Attention Task ŽSSAT, Myles-Worsley et al., 1991a.. Two words Žall five]seven letters and medium frequency in the language. appeared simultaneously, one above the other, in a two-cell matrix displayed continuously in the center of the screen. Each pair of words was displayed for 200 ms and then backward-masked by two rows of Xs. This cycle was repeated at the rate of one word pair every 1000 ms. The subject’s task was to identify any word followed by a question mark. The question mark probe followed a random 20% of the word pairs and was equally likely to appear in the upper cell as in the lower cell. A 3-s interval was allowed for the subject’s verbal response, and then the task was resumed. The task was performed under two conditions: divided attention and selective attention. In the divided attention condition, no word appeared more than once, the equivalent of 0% predictability. Under this divided attention condition, words that appeared in the upper cell, which later became predictable, are called P9 words, and words

M. Myles-Worsley, H. Coon r Psychiatry Research 71 (1997) 163]174

that appeared in the lower or non-predictable cell are called N9 words. The average identification accuracy for P9 and N9 words was used as the measure of the subject’s SSAT baseline accuracy. In the selective attention condition, a degree of predictability was introduced into the upper cell. The word ‘WINTER’ appeared 65% of the time and novel words appeared 35%of the time in that location. Consequently, there were two types of displays in the selective attention condition, each containing two word probe possibilities. One type of display, occurring 35% of the time, presented two novel words, and the subject could be asked to identify either the upper predictable cell word, designated a P word, or the lower non-predictable cell word, designated an N word. Identification accuracy for P words vs. N words was used to calculate each subject’s PrN selectivity ratio. The other type of display, occurring 65% of the time, presented WINTER in the upper cell combined with a novel word in the lower cell. The subject could be asked to identify either WINTER or the lower non-predictable cell word appearing in combination with WINTER. Fig. 1 shows examples of: Ža. a display sequence; Žb. a divided attention word display illustrating first a P9 word and then an N9 word being probed; and Žc. the two types of word displays presented under the selective attention condition, the first presenting the P word and N word probe possibilities and the second presenting WINTER in combination with a novel word. The task was administered in three segments: a practice segment to familiarize the subject with the demands of the task, a divided attention segment with 0% predictability, and then a selective attention segment with 65% predictability in the upper cell. Twin subjects who were unable to reach a mean word identification accuracy level of at least 0.250 in the divided attention segment were eliminated Žfour out of 92 twin pairs.. 2.3. Statistical analyses 2.3.1. Age effects SSAT baseline accuracy and the PrN selectivity ratio, the d9 and Beta measures of CPT performance, and SPAN accuracy were evaluated for

167

age effects by entering each dependent variable into an analysis of covariance with sex and zygosity as grouping factors and age as a covariate. 2.3.2. Genetic effects For each measure, intraclass correlations were computed to estimate degree of similarity between MZ and DZ twins. Structural modeling techniques ŽPlomin et al., 1990. were then used to estimate the genetic and environmental components underlying the individual differences based on the observed covariances computed separately for MZ and DZ twin pairs. The standard LISREL model, presented schematically in Fig. 2, assumes that the phenotype ŽP. is completely determined by additive genetic effects ŽA., environmental influences that are common to both twins ŽC., and environmental influences that are not shared by the twins ŽE.. Within twin pairs, correlations between the latent, unobserved variables ŽA, C, E. are fixed as follows. For MZ pairs who have identical genotypes, the correlation between the additive genetic value of the twins is fixed at 1.0. For DZ pairs who share only half of their genes on average, this correlation is fixed at 0.5. Correlations between the common environments are fixed at 1.0 for both MZ and DZ pairs, as these aspects of the environment are the same within a pair. The correlation is set to 0 for the unique environmental influences ŽE. because these effects are unshared and therefore do not contribute to twin pair similarity. We quantify the variation observed in the phenotype ŽP. as partially due to genetic effects Žh., the square root of the heritability of the phenotype, partially due to the effects of the shared environment Žc., and partially due to the environment unique to a twin Že.. The program LISREL ŽJoreskog and Sorbom, 1988. was used to estimate the effects of the latent variables on the measured phenotype P by estimating standardized regression coefficients Že.g. see Heath et al., 1989.. 3. Results 3.1. Descripti®e statistics Means and standard deviations for IQ scores,

168

M. Myles-Worsley, H. Coon r Psychiatry Research 71 (1997) 163]174

Fig. 1. Schematic illustration of the spontaneous selective attention task ŽSSAT..

the two SSAT measures, the d9 and Beta measures of CPT performance, and SPAN accuracy

are presented in Table 1. Mean values for IQ were within the range of normal values for the

M. Myles-Worsley, H. Coon r Psychiatry Research 71 (1997) 163]174 Table 1 Means and standard deviations for IQ, the SSAT measures, CPT d9 and Beta, and SPAN accuracy

IQ Verbal Performance Total

Mean

S.D.

109.1 109.0 109.9

13.1 12.3 12.0

SSAT Baseline accuracy PrN ratio

0.511 1.225

0.166 0.925

CPT d9 Beta

3.628 2.346

1.066 2.528

SPAN accuracy

0.796

0.109

general population, indicating that the level of intelligence in the twin sample was representative of 10]18-year-olds. Mean baseline word identification accuracy under the divided attention condition was 0.511" 0.166, a level of accuracy that closely corresponds

169

with the mean baseline identification accuracy of 0.496 reported for adult normal controls ŽMylesWorsley et al., 1991a.. Thus, on average, normal adult as well as child subjects could accurately identify only one of the two novel words presented simultaneously for 200 ms. This result indicates that the level of difficulty of the SSAT is appropriate for children as young as 10 years of age: there were no apparent floor effects in the age range studied. SSAT performance is depicted graphically in Fig. 3. As found in a sample of normal adults ŽMyles-Worsley et al., 1991b., these young subjects showed an attentional preference for the upper location under the divided attention condition which was eliminated under the selective attention condition when predictability was introduced into the upper word stimulus location. Identification accuracy for upper cell words ŽP9 and P words. dropped from 0.63 to 0.47 Ž t s 6.87, d.f.s 366, P- 0.001, two-tailed. while identification accuracy for lower cell words ŽN9 and N words. rose from 0.39 to 0.48 Ž t s 4.02, d.f.s 366, P- 0.001, two-tailed.. Thus, under the selective attention condition with partial predictability in

Fig. 2. Standard LISREL model assuming that the phenotype ŽP. is completely determined by additive genetic effects ŽA., environmental influences common to both twins ŽC., and unshared environmental influences ŽE..

170

M. Myles-Worsley, H. Coon r Psychiatry Research 71 (1997) 163]174

Fig. 3. Mean identification accuracy for novel words appearing in the upper predictable cell ŽP9 and P. vs. the lower non-predictable cell ŽN9 and N. under the divided attention and the selective attention conditions.

the upper word location, the normal child subjects, like their adult counterparts, adjusted their attentional allocation to match the new task de-

mands by allocating more attention to N words and less attention to P words. Intercorrelations among the various measures are shown in Table 2. There was a significant correlation between baseline accuracy on the SSAT and verbal IQ ŽPearson r s 0.239, P- 0.01., suggesting that the word identification required by the SSAT depends on the individual’s level of verbal intelligence. Baseline accuracy was also correlated with SPAN accuracy ŽPearson r s 0.256, P- 0.001., indicating that both word identification and the partial report span of apprehension may tap a common component of attentional processing, possibly speed of stimulus identification. However, the PrN selectivity ratio was uncorrelated with IQ, CPT performance, and SPAN accuracy, suggesting that this measure is indexing a particular aspect of attentional processing that is not assessed by the other, more widely used, tests. However, it is important to note that the particular versions of the CPT and SPAN used in the present study differ in some attributes from versions used in other studies, and these differences may contribute to the absence of overlap in the attentional tests. 3.2. Age effects There were reliable developmental trends in

Table 2 Intercorrelations among the SSAT measures, CPT d9, and Beta SPAN accuracy, and IQ scores Pearson r values SSAT Baseline accuracy

CPT PrN ratio

SSAT Baseline accuracy PrN ratio

1.000 0.083

1.000

CPT d9 Beta

0.184 0.001

0.014 0.100

SPAN accuracy IQ Verbal Performance Total U

P- 0.01 P- 0.001

UU

UU

0.256

0.239U 0.162 0.247U

SPAN

d9

Beta

1.000 y0.093 UU

accuracy

1.000

0.109

0.278

y0.042

1.000

y0.145 y0.111 y0.160

0.064 0.102 0.101

y0.124 y0.143 y0.162

y0.072 0.052 y0.028

M. Myles-Worsley, H. Coon r Psychiatry Research 71 (1997) 163]174

SSAT baseline accuracy, CPT d9, and SPAN accuracy over the 10]18-year age range. The analysis of covariance of the SSAT measures revealed that the effect of age as a covariate was significant for baseline accuracy, F1,179 s 18.32, P- 0.001, but not for the PrN selectivity ratio. Age was also significant as a covariate in the case of the CPT d9 measure, F1,179 s 6.49, P- 0.01, and SPAN accuracy, F1,179 s 15.31, P- 0.001. Sex and zygosity were non-significant across all attentional measures. Baseline accuracy in the divided attentioncondition increased linearly with age ŽPearson r s 0.300, P- 0.001., but the PrN selectivity ratio was stable over the age range studied ŽPearson r s y0.004, n.s.. 3.3. Genetic effects Intraclass correlations for MZ twins and DZ twins were calculated for IQ and each of the attentional measures as presented in Table 3. LISREL analysis was not run on the CPT Beta and SPAN accuracy measures because the standard model could not be fit to the data given the unexpected DZ intraclass correlations. The LISREL model described above was fitted to the variance]covariance matrices for the MZ

171

Table 3 Intraclass correlations for MZ and DZ twins Intraclasscorrelation MZ Twins

DZ Twins

Total IQ

0.857

0.485

SSAT Baseline accuracy PrN ratio

0.505 0.437

0.200 0.012

CPT d9 Beta

0.255 0.366

0.079 y0.135

SPAN accuracy

0.187

0.307

and DZ twin pairs. Table 4 gives the parameter estimates computed when the model was fit to the IQ data, the SSAT baseline accuracy and PrN ratio measures, and CPT performance as measured by d9. For the PrN ratio measure and CPT d9, none of the observed variance in the data was attributed to shared environmental influences Žc.. The observed correlation for MZ twins is relatively high but the correlation for DZ twins is lower than expected. Additive genetic influences

Table 4 LISREL estimates and model fitting statistics for IQ, SSAT baseline accuracy, SSAT PrN ratio, and CPT d9 Total

SSAT

IQ

Baseline accuracy

Parameter h Ž h2 .a c e

0.828" 0.181 0.686 0.717" 0.224 y0.491" 0.045

0.113" 0.045 0.013 0.015" 0.297 0.119" 0.011

Chi-square Ž1. Full model Ž2. h set to 0

3.51 11.71UU

2.41 3.86

7.07U 10.89U

1.84 2.31

Difference between 1 and 2

8.20UU

1.45

3.82U

0.47

a

2

CPT d9 PrN ratio

0.637" 0.092 0.406 0 0.771" 0.062

The parameter h Žheritability . is not estimated directly; therefore, a standard error cannot be computed. P- 0.05 UU P- 0.01 U

y0.532" 0.134 0.283 0 0.931" 0.078

172

M. Myles-Worsley, H. Coon r Psychiatry Research 71 (1997) 163]174

would suggest some positive correlation among DZ pairs who share approximately half their genes. To take into account the low DZ intraclass correlation for the PrN ratio, we tested a dominant genetic model substituting dominant genetic effects ŽD. for common environmental effects ŽC. and reducing the correlation on this factor from 1.0 for MZ twins to 0.25 for DZ twins. The resulting change in the estimate of the heritability of the PrN ratio was negligible. The significance of the effects of heredity can be tested by fixing the parameter h to 0 and noting how the fit of the model worsens. This test evaluates if the similarity observed among twin pairs could be explained without genetic influences Žsee Neale et al., 1989, for details on hypothesis testing using LISREL.. This test shows that genetic effects Žh. are necessary to explain the similarity observed among twin pairs for IQ and the SSAT PrN ratio, as indicated by the significant change in chi-square between the full model and the model without h. In summary, the LISREL analysis estimates heritability of the PrN ratio to be 0.41 and heritability of IQ to be 0.69. SSAT baseline accuracy and CPT d9 showed no significant heritability. Measurement error for the particular versions of the CPT and SPAN used in our study may have been too large to detect genetic effects. Although our twin sample is relatively small, the result for the heritability of IQ is within the range of estimates that have been reported in the world literature. Heritability estimates from the larger, more recent studies cluster around 0.50, indicating that about half of the variation observed in the IQ scores is due to genetic influences ŽBouchard and McGue, 1981.. 4. Discussion This study evaluated age and genetic influences on spontaneous selective attention performance using a new laboratory task, the SSAT, and compared the results to those for two other commonly used attention tests, the CPT and SPAN. The overriding aim of the study was to determine whether the SSAT provides a valid measure of selective attention in normal children and adolescents 10]18 years of age and shows sufficient

heritability to warrant consideration as a vulnerability marker for schizophrenia in studies of children born into families segregating the illness. The results confirmed that the SSAT PrN ratio which is used to measure selective attention has a significant genetic component. These heritable effects accounted for an estimated 41% of the observed variance in the PrN selectivity ratio. In addition, the PrN selectivity ratio was not significantly correlated with IQ, suggesting that genetic effects on selective attention are independent of genetic effects on IQ. With respect to age effects, baseline performance on the SSAT did improve with age, but there were no floor effects in the youngest age group. In fact, most of the children were able to identify words just as accurately as normal adult subjects who have been found to average 49.4% correct at this exposure duration ŽMyles-Worsley et al., 1991a.. Furthermore, the PrN ratio used to identify abnormal selective attention was stable over the 10]18-year age range we studied, indicating that the SSAT is suitable for testing children in fourth or fifth grade and beyond. The generalizability of these findings is, however, subject to the limitations inherent in using a crosssectional design to evaluate developmental effects. The absence of significant intercorrelations between the PrN selectivity ratio and the CPT and SPAN measures suggests that the SSAT assesses a distinctly different aspect of attention from those measured by the other two attentional tests. The CPT measures the type of sustained focused attention typically associated with our ability toconcentrate. The SPAN measures our ability to efficiently scan a complex visual environment and find what we are looking for. The SSAT evaluates our ability to selectively attend to what is non-predictable, what is new or unexpected, in the flow of information we encounter in our daily lives. Impairment in this selective component of attention may be related to an inability to filter out irrelevant information so that the individual is easily distracted and feels overloaded by too many stimuli at the same time. These are the types of attentional problems that many schizophrenic patients say they experience

M. Myles-Worsley, H. Coon r Psychiatry Research 71 (1997) 163]174

ŽMcGhie and Chapman, 1961; Freedman and Chapman, 1973.. Therefore, as a measure of the selective component of attention, the SSAT may be a useful addition to testing batteries designed to evaluate information-processing impairments in children who are at heightened genetic risk for schizophrenia. Acknowledgements This research was supported by a grant from the National Institute of Mental Health ŽMH43901.. References Achenbach, T.M., Edelbrock, C., 1983. Manual for the Child Behavior Checklist and Revised Child Behavior Profile. University of Vermont Department of Psychiatry, Burlington. Asarnow, R.F., MacCrimmon, D.J., 1978. Residual performance deficit in clinically remitted schizophrenics: a marker of schizophrenia? J. Abnorm. Psychol. 87, 597]608. Asarnow, R.F., MacCrimmon, D.J., 1981. Span of apprehension deficits during postpsychotic stages of schizophrenia. Arch. Gen. Psychiatry 38, 1006]1011. Asarnow, R.F., Steffy, R.A., MacCrimmon, D.J., Cleghorn, J.M., 1977. An assessment of foster children at risk for schizophrenia. J. Abnorm. Psychol. 86, 267]275. Asarnow, R.F., Sherman, T., Strandburg, R., 1986. The search for the psychobiological substrate of childhood onset schizophrenia. J. Am. Acad. Child Adolescent Psychiatry 26, 601]604. Bartfai, A., Pederson, N., Asarnow, R.F., Schalling, D., 1991. Genetic factors for the span of apprehension test: a study of normal twins. Psychiatry Res. 38, 115]124. Bouchard, T.J., McGue, M., 1981. Familial studies of intelligence: a review. Science 212, 1055]1059. Coon, H., Plaetke, R., Holik, J., Hoff, M., Myles-Worsley, M., Waldo, M., Freedman, R., Byerley, W., 1993. Use of a neurophysiological trait in linkage analysis of schizophrenia. Biol. Psychiatry 34, 277]289. Cornblatt, B.A., Erlenmeyer-Kimling, L., 1985. Global attentional deviance as a marker of risk for schizophrenia: specificity and predictive validity. J. Abnorm. Psychol. 94, 470]486. Cornblatt, B.A., Winters, L., Erlenmeyer-Kimling, L., 1989. Attentional markers of schizophrenia: evidence from the New York High Risk Study. In: Schulz, S.C., Tamminga, C.A. ŽEds.., Schizophrenia: Scientific Progress. Oxford University Press, New York, NY, pp. 83]92. Cornblatt, B.A., Lenzenweger, M.R., Dworkin, R.H., Erlenmeyer-Kimling, L., 1992. Childhood attentional dysfunctions predict social deficits in unaffected adults at risk for schizophrenia. Br. J. Psychiatry 161, 59]64.

173

Cornblatt, B.A., Risch, N., Faris, G., Friedman, D., Erlenmeyer-Kimling, L., 1988. The continuous performance test, identical pairs version ŽCPT-IP.: I New findings about sustained attention in normal families. Psychiatry Res. 26, 223]238. Freedman, B.J., Chapman, L.J., 1973. Early subjective experience in schizophrenic episodes. J. Abnorm. Psychol. 82, 46]54. Gottesman, I., Shields, J., 1972. Schizophrenia and Genetics: A Twin Study Vantage Point. Academic Press, New York, NY. Gottesman, I., McGuffin, P., Farmer, A., 1987. Clinical genetics as clues to the ‘real’ genetics of schizophrenia: a decade of modest gains while playing for time. Schizophr. Bull. 13, 23]47. Grove, W.M., Clementz, B.A., Iacono, W.G., Katsanis, J., 1992. Smooth pursuit ocular motor dysfunction in schizophrenia: evidence for a major gene. Am. J. Psychiatry 149, 1362]1368. Heath, A.C., Neale, M.C., Hewitt, J.K., Eaves, L.J., Fulker, D.W., 1989. Testing structural equation models for twin data using LISREL. Behav. Genet. 19, 9]35. Hochhaus, L.A., 1972. A table for the calculation of d9 and Beta. Psychol. Bull. 77, 375]376. Joreskog, K.G., Sorbom, D., 1988. LISREL VII: A Guide to the Program and Applications. SPSS, Chicago. Kendler, K.S., Diehl, S.R., 1993. The genetics of schizophrenia: a current genetic]epidemiologic perspective. Schizophr. Bull. 19, 261]285. McGhie, A., Chapman, J., 1961. Disorders of attention and perception in early schizophrenia. Br. J. Med. Psychol. 34, 103]116. Myles-Worsley, M., Johnston, W.A., Wender, P., 1991a. Spontaneous selective attention in schizophrenia. Psychiatry Res. 39, 167]179. Myles-Worsley, M., Emmerson, R., Plaetke, R., Wender, P., Byerley, W., 1991b. Investigations of abnormal selective attention as a possible vulnerability marker for schizophrenia. Schizophr. Res. 4, 389]390. Neale, J.M., 1971. Perceptual span in schizophrenia. J. Abnorm. Psychol. 74, 196]204. Neale, J.M., McIntyre, C.W., Fox, R., Cromwell, R.L., 1969. Span of apprehension in acute schizophrenics. J. Abnorm. Psychol. 74, 593]596. Neale, M.C., Heath, A.C., Hewitt, J.K., Eaves, L.J., Fulker, D.W., 1989. Fitting genetic models with LISREL: hypothesis testing. Behav. Genet. 19, 37]49. Nuechterlein, K., 1983. Signal detection in vigilance tasks and behavioral attributes among offspring of schizophrenic mothers and among hyperactive children. J. Abnorm. Psychol. 92, 4]24. Orzack, M.H., Kornetsky, C., 1966. Attention dysfunction in chronic schizophrenia. Arch. Gen. Psychiatry 14, 323]326. Orzack, M.H., Kornetsky, C., 1971. Environmental and familial predictors of attention behavior in chronic schizophrenics. J. Psychiatr. Res. 9, 21]29.

174

M. Myles-Worsley, H. Coon r Psychiatry Research 71 (1997) 163]174

Plomin, R., Defries, J.C., McClearn, G.E., 1990. Behavioral Genetics: A Primer. Freeman, New York. Strauss, M.E., Prescott, C.A., Gutterman, D.F., Tune, L.E., 1987. Span of apprehension deficits in schizophrenia and mania. Schizophr. Bull. 13, 699]704. Walker, E., 1981. Attentional and neuromotor functions of schizophrenics, schizoaffectives and patients with other affective disorders. Arch. Gen. Psychiatry 38, 1355]1358.

Wechsler, D., 1974. Wechsler Intelligence Scale for ChildrenRevised ŽWISC-R.. Harcourt Brace Jovanovich, San Antonio, TX. Wechsler, D., 1981. Wechsler Adult Intelligence Scale-Revised ŽWAIS-R.. Psychological Corp., New York. Wohlberg, G.W., Kornetsky, C., 1973. Sustained attention in remitted schizophrenics. Arch. Gen. Psychiatry 28, 533]537.