REACTION TIME VARIABILITY IN EPILEPTIC AND BRAIN-DAMAGED PATIENTS Peter Bruhn and Oscar A. Parsons (Rigshospitalet, Copenhagen, and University of Oklahoma Health Sciences Center)
In a previous study the frequent occurrence of delayed reaction times (RTs), i.e., greater variability, in patients with epilepsy of subcortical or centrencephalic origin was demonstrated (Bruhn, 1970). Several questions, however, remained to be answered. First, the group of subjects (Ss) in the previous study was rather small (N = 15) and selected on very specific EEG criteria (generalized paroxysms of high amplitude, spikes and slow waves). Whether the appearance of delayed RTs are specific to this group of epilep tics in particular, or if similar RT abnormalities are also seen in epileptics with other types of EEG abnormalities and/or seizures, is the primary ques tion to be resolved here. Secondary questions concern the role of frequency of seizures, age of onset, duration of illness and etiology of disorder in RT variability. The second major question is concerned with RT performance in brain damaged but non-epileptic patients. It is known that such patients are more variable (see Bruhn and Parsons, 1971) in their RT performance than matched controls; however, the general slowing of RT latencies over the whole distribution is typically assumed to be the main characteristics of their RT performance. It is possible that RT variability is a phenomenon assodated with "cerebral dysfunction" in general as well as with the occurrence of clinical seizures or EEG abnormality. A third question is whether the continuous reaction time test can provide a reliable and rapid screening test among epileptics, brain-damaged and non-brain-damaged individuals. Bruhn (1970) found almost complete· dis crimination between his epileptics and controls but the Ns were small and the controls were healthy normals rather than other brain-damaged or non brain-damaged patients.
Cortex (1977) 13, 373-384.
374
P. Bruhn and 0. A. Parsons
11ATERIAL AND 11ETHOD
Subjects
Patients from Rigshospitalet of Copenhagen (in and outpatients from Depart ments of Neurology and Neurosurgery) were examined. Generally, patients with dubious diagnoses were not used. Psychiatric and general medical problems also led to exclusion from all groups. Finally, only patients with the necessary visual and motor capabilities for participating in the experiment were included. 113 Ss who met the patient selection criteria were then divided into three groups: Controls (C) (N = 25), Epileptics (Ep) (N = 63 ), Brain-damaged (BD) without epilepsy (N = 25). The C group consisted of Ss who, based on history and clinical neurological examination, did not show any signs of CNS dysfunction {structural damage or seizures of any kind). Also patients with previous head trauma causing loss of consciousness were excluded. All Ss in this C group had a diagnosis of peripheral neurological disorders. There were 11 males and 14 females in this group, the median age being 48.0 years {range 33-61 ). EEG's were not available in this group. The Ep group consisted of Ss who all at the time of the investigation had the diagnosis of epilepsy based solely on the presence of manifest (clinical) epileptic seizures of any kind. No restrictions as to origin, etiology, presence of EEG abnormality, types and frequency of seizures, etc., were put on this group. Also the Ss at the time of the investigation were varied with respect to drug treatment, a factor very difficult to control in studies like this. There were 31 males and 32 females in this group. The median age being 29.0 years {range 11-63 years). Standard EEG records were available on all Ss in this group. The EEG ex amination usually was given within one month of the RT experiment. All Ss in the BD group had a diagnosis of brain damage based on history, clinical neurolgical examination and one or more of the following radiological evaluations: arteriography, brain scan and pneumoencephalography. The group was heterogeneous with respect to extent, location and etiology, however, predo· minately diffuse lesions were present. Excluded from this group were brain-damaged Ss that manifested any sign of clinical seizures: the mere suspicion of epilepsy would automatically exclude a patient from this group. There were 16 males and 9 females in the group. The median age was 43.0 years (range 22-64 years). Standard EEG records were available on all Ss. ' The groups did not differ significantly in proportion of males and females (chi" = .89, p = n.s.) but did differ in age. The Ep group was significantly younger than C and BD (chi'= 25.13; p < .001); the latter two groups did not differ. Apparatus and Procedure
A vanatlon of the simple visual, continuous RT paradigm used by Bruhn (1970) was employed in this experiment. The visual stimulus consisted of three red diode lamps, each with an intensity of about 1 mcd, spaced 17 mm apart to form the shape of a triangle, and placed about 75 em in front of the S. Stimuli were delivered by a programmed magnetic tape at a rate of 15 per min. with randomized intertrial intervals of 2-6 sec. A pulse from the tape would simulta
Reaction time variability
375
neously turn on the stimulus light and activate an electronic digita~ timer (Type LEC 1004). The Ss task was to terminate the stimuli as fast as possible by depressing a thumb-key (a micro-switch activated by a force of about 60 gr.) held in the preferred hand. A depression of the key would switch off the light and stop the digital clock, showing the RT latencies (from stimulus onset to depression of the response key) in 1/100 of a sec. Through the experiment the Ss were seated in a comfortable chair in a dimly lit room. Earphones provided a low level white noise to mask distracting sounds from outside the room. A warm-up period of 15 stimuli was followed by a taped instruction, urging the S to respond as fast as possible to each stimulus, where after 100 consecutive RTs were measured. As no warnings were given, the alertness of the S had to be maintained throughout a rather monotonous task lasting about 61;2 min. RESULTS
A first step in the data analysis was to find a relevant, numerical expres sion of the RT variability. RT distributions are typically skewed (our data are no exception) and can be transformed by several techniques to a more normal distribution to which parametric statistics can be applied. We wished, however, to keep our data as close to the raw scores as possible to preserve clinical utility. Therefore, we employed medians of the .50 per centile, the .10 percentile and the .90 percentile. These measures have the advantage of providing a test of whether groups differ in their fastest responses or their slowest responses, in addition to their median scores. The intra-subject variability can be expressed numerically as the .90 - .10 percentile difference. In Table I the basic data for the groups are presented. To facilitate reading Table I consider the .10 percentile (fast response) row. The median of the distribution of C scores is 220 with an interquartile range of 201 TABLE I
Group Medians (and Quartiles) of the .10, .50, .90 Percentiles of the Reaction Times for Individual Subjects in the Three Groups. Also, the .90-.10 Percentile Difference and the Median of the Individual Means are Given ( msecs.) Controls (N = 25)
Epileptics (N = 63)
Brain-damaged (N = 25)
Median ( quartiles)
Median (quartiles)
Median ( quartiles)
.90 .90-.10
220 (201-231) 250 (230-270) 310 (270-330) 90 (65-100)
240 (220-270) 285 (260-330) 370 (320-449) 130 (100-199)
Mean
257 (234-277)
296 (270-344)
Percentile .10
.50
240 (220-250)
305 (275-325) 409 ( 344-500) 150 (124-255) 321 (283-357)
376
P. Bruhn and 0. A. Parsons
to 231 msec. compared with the Ep median of 240, interquartile range of 220-250 msec., and the BD median of 240, interquartile range of 220 250 msec. Other rows are read similarly. Mann-Whitney tests applied to group comparison of medians indicate that significant differences were obtained in all comparison,s involving the C group; the latter had significantly faster median RTs than Ep or BD and significantly lower variability (.90 - .10). Ep and BD did not differ significantly in any comparison (Table II). TABLE II
Mann-Whitney U-tests of Differences between Groups
Percentile
C vs. Ep
Ep vs. BD
C vs. BD
.10 .50 .90 .90-.10 Mean
P< .01 p < .001 p < .001 p < .001 p < .001
n.s. n.s. n.s . n.s. n.s.
P< .01 p < .001 p < .001 p < .001 p < .001
Presence and severity of BEG abnormality
EEG abnormalities were graded into several categories: ( 1) not abnor mal, (2) slightly abnormal, (3) slight to moderate, (4) moderate, (5) mo derate to severe and (6) severely abnormal. (For criteria used for this classification see Buchthal and Lennox, 1953.) To reach acceptably large subgroups, categories 3, 4 and 5 were treated as one, giving a total of four subgroups representing increasing severity of abnormality. The first comparison were between patients with abnormal vs. normal EEGs in the Ep group. 38 of 63 Ss in the Ep group were designated sligthly to severely abnormal; the median RT variability scores (.90 - .10) were 130 msec. and 134.5 msec. for normal and abnormal Ss, respectively. Mann-Whit ney U-tests clearly indicate no difference in these groups. Within the abnormal EEG groups no significant differences were obtain~d among the different ca tegories. Similarly in the BD group the 10 Ss with abnormal BEGs had the sam_e median RT variability (150 msec.) as the 15 normal EEG Ss (150 msec.). Kind of BEG abnormality'
Using the standard records, EEG's were scored as to whether the abnormal discharges were diffusedly represented or focal and whether paroxysmal activity 1 The EEG classification was suggested by Dr. 0. Thage, who also assisted in the evaluation of the single records.
Reaction time variability
377
(spikes and sharp waves) or slow wave (less than 6 per sec.) activity was pre sent. Theoretically, in a single S all four types of EEG abnormality may be present (paroxysmal diffuse, paroxysmal focal, slow wave activity diffuse, slow wave activity focal). Comparing Ss having one of the abnormalities mentioned above with those Ss not having the same abnormality did not indicate any significant differences. Although the sample was not analyzed by combinations of EEG abnorma lities, the fact that presence vs. absence of a given EEG abnormality in itself was not differentiating, makes it very unlikely that any kind of EEG discharges should be specifically related to RT variability. Thus kind of EEG abnor mality was not associated with RT variability.
Type of seizures Classifying i:he Ep sample by seizure type, the following categories were used: (1) grand mal(= major motor); (2) petit mal; (3) psychomotor; (4) focal ( = minor motor) and (5) other kinds of seizures. This classification was based on clinical description of the seizures in accordance with directions used traditionally in clinical neurology. Obviously all kinds of seizures inay be present in one S. (Petit mal was only present in 6 Ss and never as the only type of seizure). A first analysis was to score each S in the Ep group at to the presence vs. absence of a specific seizure type (Table III). The group of Ss having grand TABLE III
Relationship between Type of Seizures and Median RT Variability ( msec.) Absent
Type of seizures Grand mal Petit mal Psychomotor Focal Other 1
*
(N (N (N (N (N
= 19) = 57) = 31) = 44) = 61)
119.0 139.0 120.0 130.0 130.0
Present' (N (N (N (N (N
= 44) = 6} = 32) = 19) = 2)
139.5* 100.0 134.5 129.0
Mann-Whitney Test used to test for differences. < .05.
p
mal seizures (N = 44) vs. those not having this type of seizures (N = 19) had median RT variability (.90 - .10) scores of 139.5 msec. and 119.5 msec. respectively. This difference was significant (Mann-Whitney Test p < .05). No other differences were significant. Secondly, an analysis was performed based on subgroups of Ep having only one kind of seizures, i.e., grand mal only (mean = 154 msec.), psychomotor only (mean = 80 msec.), focal only
P. Bruhn and 0. A. Parsons
378
(mean = 114.5 msec.). Although the "psychomotor seizures only" group had a median variability of only 80 msec. as compared to 154 msec. in the "grand mal only" group, no significant difference was reached (Kruskal-Wallis Test) because of a considerable overlap between subgroups.
Frequency of seizures Categorizing the Ss according to ftequency of seizures within the prior year, Ep Ss with less than 4 seizures a year (N = 15), with 4-12 seizures a year (N = 9), and with more than 12 seizures a year (N = 39) were com pared (see Table IV) as to the variability score. For comparative purposes TABLE IV
Frequency of Seizures in Relation to RT Variability ( msec.)
Median RT variability
Frequency of seizures Epileptics
(4 a year) ( 4-12 a year} (> 12 a year)
Controls (no seizures) Brain-damaged (no seizures) 1
(N = 15) (N = 9) (N = 39)
90.0 229.0 139.0
= 25) = 25)
90.0 150.0
(N (N
Differences'
p
= .001
p
< .001
Kruskal-Wallis Test.
C and BD medians also are presented. Ep Ss with less than 4 seizures a year are no different from the controls. A finding more difficult to explain is that the 4-12 seizures a year group shows highly elevated RT variability scores compared to the even more frequently seizuring group. One might speculate whether medication is insufficient in the medium group as 4-12 seizures a year might not alarm the pati~nt or the doctor too much. (The three subgroups were not different in respect to age). It is interesting to note that epileptics with less than 4 seizures have a median RT variability score identical to that of the C group, while those who have greater than 12 seizures per year are close to the BD median score.
Age at onset Other studies have tried to relate age at onset of epilepsy to mental de terioration (Dikmen, Matthews and Harley, 1975). In these studies the age at onset categories were: (a) 5 years or below; (b) 6-16 years and (c) above 16 years at the first seizure. The Ep group was divided using the same age inter
Reaction time variability
379
vals. Group A consisted of 15 Ss (Md RT variability 130 msec.); Group B of 16 Ss (Md 139.5 msec.) and Group C of 32 Ss (Md 129.5 msec.). No significant difference between Ss with early, medium and late onset of the disease was obtained by Kruskal-Wallis Test. (Groups A, B and C were not significantly different in respect to age.) Duration of illness
In all Ep Ss the time since the first epileptic seizure was noted, and a Spearman rank order correlation between duration of illness (in years) and RT variability was computed. In the total Ep sample (N = 36) the Spearman correlation coefficient (rho) was .14. In subgroups of Eps based on etiology: etiology unknown (N = 36), rho = .24; posttraumatic (N = 13), rho = - .11 ; epilepsy secondary to other brain damage (N = 14) rho= .15. No correla tions were significant. Etiology of the epileptic disorder
In order to find out if RT variability was related to the etiology of the disease the Ep group was divided into 3 subgroups: (1) those having epileptic seizures of unknown etiology, i.e., the idiopathic or cryptogenetic group (N = 36), (Md RT variability = 124.5 msec.); (2) those having their seizures related to head trauma, i.e., the posttraumatic group (N = 13 ), (Md = 160 msec.) and finally ( 3) those having seizures as symptoms of known brain damage or disease, i.e., degenerative, vascular, neoplastic. etc. (N = 14), (Md = 129.5 msec.). White the median RT variability was somewhat higher in the posttraumatic group than in the other two groups, the difference was not significant (Kruskal-Wallis Test). (Again, no significant age differences between the subgroups were seen.) Correlation of RT variability and age
Correlating RT variability with age, a positive correlation of rho = .34, p < .01 was found in the Ep group, the older Ep Sc; are more variable than the younger ones. In the C as well as in the Bi[) group, the correlations were close to zero (- .08 and .10 respectively). EP vs. BD comparisons
The second major question was to determine if Ep and BD Ss are different with respect to RT variability and general RT slowing. As seen in Table I and Table II, no significant differences between Ep and BD were found in median RTs or variability.
P. Bruhn and 0. A. Parsons
380
Identification of brain-damaged and epileptics
The final question was whether this rapid technique of measuring RTs could have screening value. In Table V the relevant data are given. The TABLE V
Identification of Controls, Brain-Damaged and Epileptics Using Optimal Cutting Score on Each Measure
Groups
Control (N = 25) Brain-damaged (N = 25) Epileptics (N = 63)
.90- .10 percent. (115 msec.)
.90 percentile (330 msec.)
.50 percentile (288 msec.)
.10 percentile (122 msec.)
Below
Above
Below
Above
Below
Above
Below
Above
92%
8%
80%
20%
80%
20%
80%
20%
16%
84%
20%
80%
31%
69%
28%
72%
30%
70%
29%
71%
43%
57%
38%
62%
best overall discrimination, using optimal cut-off scores, is given .10 measure of variability which identified correctly 92% of 84% of the BD and 70% of the Epileptic patients. The two could not be distinguished from each other on the .90 - .10 any of the others in Table V.
by the .90 the Controls, latter groups measures vs.
DISCUSSION
The present investigation was devoted primarily to the study of RT variability in epileptics and non-epileptic BD Ss. RT variability as the depen dent variable was selected to represent an instance of intra-individual response variability due to the o~inion of the authors, that the mean RT scores traditionally used in most studies conceal very important information on the underlying disturbances of the CNS. Fiske and Rice (1955) define "pure intraindividual variability... as the difference between the two responses of an individual at two points in time under the following conditions: (a) the individual is exposed each time to the same stimulus or to objectively indistinguishable stimuli; (b) the total situation in which the responses are made is the same on both occasions." Obviously, the continuous RT task fulfills this model in a nearly perfect way. The first question concerned RT variability as a function of a number of variables known to be important in epilepsy.
Reaction time variability
381
Is RT variability related to the presence, severity and type of EEG ab normality? The present study, contrary to our expectations, has demonstrated no differences between Ep with and without EEG abnormalities (as was the case with BD Ss with and without EEG disturbances). Nor were there any indications of an increase in variability in the more severely abnormal EEG groups. One explanation to account for these results in comparison with other studies (Browne, Penny, Porter and Dreifuss, 1974; Friedman and Taub, 1969; Porter, Penry and Dreifuss, 1973; Kooi and Hovey, 1957; Mirsky and Van Buren, 1965; Tizard and Margerison, 1963; Woodruff, 1974) is that we used the standard iEEG examination, recorded when the S was relaxed, with eyes closed etc. This obviously is a recording situation very different from the aroused state of the individual participating in a visual RT experiment. Also, only surface-detectable abnormalities picked up from the scalp were recorded. Thus it is possible that patterns of electro physiological abnormalities of a more subtle nature were present. Finally, as the EEG records used here were read by the eye, and not computer-analyzed, only rather prominent disturbances are detected. The observation that many Ss with a definite abnormality of brain function, epilepsy, have a normal EEG pattern (N = 25 or 39%) leads one to conclude that the EEG investigation results in many "false negatives." As stated by Rodin ( 1968 ), " ... a normal EEG in an epileptic does not automatically guarantee normal intellect and/or personality" (p. 155). It may be added: neither does it guarantee stable RT performance. Our results suggest that a continuous RT procedure deserves the status of a supplementary clinical method for diagnostic purposes. Is RT variability related to type of seizures? The generalized conclusion from previous studies seems to be that temporal lobe or psychomotor epi leptics have their major problems in respect to learning and memory (Quad fasel and Pruyser, 1954 ), while centrencephalics (epileptics with grand mal seizures without aura and with generalized paroxysms in their EEGs) have maximal "attentional" problems (Mirsky, Primae, Marsan, Rosvold and Ste vens, 1960). It is noteworthy that, in the present study, patients with grand mal seizures also were more variable. The grand mal seizure is one characterized by loss of consciousness, indicating that the basal arousal systems of the brain are temporarily affected. Psychomotor seizures, quite different in their appearance as well as in their pathophysiological mechanisms, are not characterized by loss of consciousness (although conosciusness is certainly altered). The differential performance found in previous as well as the present study do indicate that interictal performance of these groups are characterized by the same kind of disturbances - although of less intensity - as will be seen during actual seizures.
382
P. Bruhn and 0. A. Parsons
Etiology and age of onset of the epileptic disorder
In neurology Ep Ss will usually be placed in one of three etiological classes: ( 1) the idiophatic (or cryptogenetic) group where no exogeneus etiology is known; (2) a posttraumatic group, where the first seizures appear time related to concussion or head trauma and (3) a symptomatic group where seizures are looked upon as symptoms of other kinds of verified brain damage or disease. From a neurological point of view, the idiopathic group will usually not be considered "brain-damaged." Although this type of Eps certainly demonstrate an abnormal tendency for ·seizures, there is usually no structural evidence to suspect that they show signs of interictal cerebral dysfunction. It is therefore surprising to notice that the group with idiopathic epilepspy are no different from Ep Ss with known brain damage as well as the BD Ss without epilepsy. The RT experiment certainly indicates that their interictal RT pedormances (and probably their brain functions) are impaired to a considerable degree. This is also the conclusion from studies by Kl0ve and Matthews ( 197 4 ), who found that idiopathic Eps were significantly impaired, compared to non-epileptic controls in more complex neuropsycho logical functions. Age of onset was not found to be related to RT variability. As our Ep patients were primarily outpatients and thus able to adapt outside of insti tutions, we may have a biased sample. However, our findings do suggest that generalizations about age of onset and severity of impairment (Kl0ve and Matthews, 197 4) must be limited. The second major question investigated RT pedormance of epileptics and non-epileptic BD Ss. It was suspected on the basis of previous research findings (Bruhn, 1970; Bruhn and Parsons, 1971) that "lapses in pedormance seem to be related to paroxysmal electrobiological disturbances in cortical activation, whereas a general delay seems to be related to structural damage of nervous tissue." This hypothesis was derived on the basis of different studies using different patients as well as different experimental set-ups. This statement was not supportetd by the direct empirical test in the present study. The BD RT distributions were as skewed as those of the Ep and prolonged RTs were found to the same amount in both groups, indicating that RT "lapses" were not phenomena specific to Ep. A nonsignificant difference between the fast end of the individual RT distributions for Ep and BD groups suggests that general slowness is characteristic of both groups to an equal degree. Considering the age differences between the two groups (Ep younger than DB), the deleterious effect of epilepsy on RT performance is rather marked, but caution must be used in interpreting these findings in that drug intake across groups was not equated. In a similar vein it is of interest to note that only in the Ep group was age related to RT performance.
Reaction time variability
383
Whether this reflects an increasing vulnerability of the aging brain to seizures or reflects a complex interaction of anti-convulsant drugs and age remain to be explored in the future. The third question concerned the discriminating power of the RT test. Although we are not able to define those disturbed neurofunctions shared by the two groups, the behavioral similarities observed in RT variability do suggest a very basic disturbance of the GNS functions in both groups. Ep and BD Ss were separated from controls by this 61;2 minute procedure at a level comparable to more sophisticated and time consuming neuro psychological methods. This reliable discrimination of cerebral dysfunction should be of value to the clinical neuropsychologist, but further research in the area is obviously needed. From a neurophysiological approach, one should focus on subtle fluctuations in neural processes underlying intra-individual response variability. Within a behavioral context, the consequences or rela tionships of fluctuating RT latencies on more complex information processing and behavior still has to be determined.
SUMMARY
Median reaction times and intra-individual variability were studied in epi leptic (N = 63 ), brain-damaged (non-epileptic) (N = 25) and control patients (N = 25) using a six and one half minute visual, continuous reaction time task. Epileptic and brain-damaged groups were significantly slower than control patients on median reaction times at the lOth, 50th and 90th percentiles and on the differences between the lOth and 90th percentiles. Thus both general slowing and greater intra-individual variability were found in the epileptic and brain damaged patients. Reaction times were not related to presence, type and severity of EEG abnormality or to age of onset of epilepsy. Grand mal patients did have significantly greater variability than other types of seizure patients. Epileptic and brain-damaged patients did not differ significantly on any reaction time variables. Both groups were discriminated significantly from the controls on all reaction time measures, especially on the intra-individual variability measure. REFERENCES BROWNE, T. R., PENRY,]. K., PoRTER, R.]., and DREIFUSS, F. E. (1974) Responsiveness before, during and after spike-wave paroxysms, "Neurology," 24, 659-665. BRUHN, P. (1970) Disturbances of vigilance in subcortical epilepsy, "Acta Neurologica Scandi navica," 46, 442-454. - , and PARSONS, 0. A. (1971) Continuous reaction time in brain damage, "Cortex," 7, 278-291. BucHTHAL, F., and LENNOX, M. (1953) The EEG effect of Metrazol and photic stimulation in 682 subjects, "EEG Clin. Neurophysiol.," 5, 545-558. DIKMEN, S., MATTHEWS, C. G., and HARLEY, ]. P. (1975) The effect of early versus late onset of major motor epilepsy upon cognitive-intellectual performance, "Epilepsia," 16, 73-81. FISKE, D. W., and RICE, L. (1955) Intra-individual response variability, "Psycho!. Bull.," 52, 217-251. FRIEDMAN, H., and TAUB, H. A. (1969) The transcephalic DC potential and reaction time performance, "Psychophysiology," 5, 504-509.
384
P. Bruhn and 0. A. Parsons
KLjljVE, H., and MATTHEWS, C. G. (1974) Neuropsychological studies of patients with epilepsy, in Clinical neuropsychology: Current status and applications, ed. by R. Reitan and L. A. Davidson, Winston and Wiley, Washington, D.C. Kom, K. A., and HovEY, H. B. (1957) Alterations in mental function and paroxysmal cerebral activity, "A.M.A. Archives of Neurology and Psychiatry," 78, 264-271. MIRSKY, A. F., PRIMAe, D. W., MARSAN, C. A., RosvoLD, H. E., and STEVENS,]. R. (1960) A comparison of the psychological test performance of patients with focal and nonfocal epilepsy, "Experimental Neurology," 2, 75-89. -, and VAN BuREN, ]. M. ( 1973) On the nature of the "absence" in centrencephalic epilepsy: a study of some behavioral, electroencephalographic and autonomic factors, "Electroenceph. Clin. Neurophysiol.," 34, 239-245. PoRTER, R. ]., PENRY,]. K., and DREIFUss, F. E. (1973) Responsiveness at the onset of spikewave bursts, "Electroenceph. Clin. Neurophysiol.," 34, 239-245. QuADFASEL, A. F., and PRuYSER, P. W. (1954) Cognitive deficit in patients with psychomotor epilepsy, "Epilepsia," 3, 80-90. RoDIN, E. A. (1968) The prognosis of patients with epilepsy, Thomas, Springfield, Illinois. TIZARD, B., and MARGERISON, J. H. (1963) The relationship between generalized paroxysmal BEG discharges and various test situations in two epileptic patients, "]. Neural. Neurosurg. Psychiat.," 26, 308-313. WooDRUFF, M. L. (1974) Subconvulsive epileptiform discharge and behavioral impairment, "Behavioral Biology," 11, 431458.
Peter Bruhn, Chief psychologist, Department of Neurology and Neurosurgery, Rigshospitalet, 9 Blegdamsvej, DK-2100 Copenhagen 0, Denmark. Oscar A. Parsons, Ph.D., Professor and Vice Chairman, The University of Oklahoma Health Sciences Center, Department of Psychiatry and Behavioral Sciences, Post Office Box 26901, Oklahoma Ciry, Oklahoma 73190, U.S.A
In a previous study the frequent occurrence of delayed reaction times (RTs), i.e., greater variability, in patients with epilepsy of subcortical or centrencephalic origin was demonstrated (Bruhn, 1970). Several questions, however, remained to be answered. First, the group of subjects (Ss) in the previous study was rather small (N = 15) and selected on very specific EEG criteria (generalized paroxysms of high amplitude, spikes and slow waves). Whether the appearance of delayed RTs are specific to this group of epilep tics in particular, or if similar RT abnormalities are also seen in epileptics with other types of EEG abnormalities and/or seizures, is the primary ques tion to be resolved here. Secondary questions concern the role of frequency of seizures, age of onset, duration of illness and etiology of disorder in RT variability. The second major question is concerned with RT performance in brain damaged but non-epileptic patients. It is known that such patients are more variable (see Bruhn and Parsons, 1971) in their RT performance than matched controls; however, the general slowing of RT latencies over the whole distribution is typically assumed to be the main characteristics of their RT performance. It is possible that RT variability is a phenomenon assodated with "cerebral dysfunction" in general as well as with the occurrence of clinical seizures or EEG abnormality. A third question is whether the continuous reaction time test can provide a reliable and rapid screening test among epileptics, brain-damaged and non-brain-damaged individuals. Bruhn (1970) found almost complete· dis crimination between his epileptics and controls but the Ns were small and the controls were healthy normals rather than other brain-damaged or non brain-damaged patients.
Cortex (1977) 13, 373-384.
374
P. Bruhn and 0. A. Parsons
11ATERIAL AND 11ETHOD
Subjects
Patients from Rigshospitalet of Copenhagen (in and outpatients from Depart ments of Neurology and Neurosurgery) were examined. Generally, patients with dubious diagnoses were not used. Psychiatric and general medical problems also led to exclusion from all groups. Finally, only patients with the necessary visual and motor capabilities for participating in the experiment were included. 113 Ss who met the patient selection criteria were then divided into three groups: Controls (C) (N = 25), Epileptics (Ep) (N = 63 ), Brain-damaged (BD) without epilepsy (N = 25). The C group consisted of Ss who, based on history and clinical neurological examination, did not show any signs of CNS dysfunction {structural damage or seizures of any kind). Also patients with previous head trauma causing loss of consciousness were excluded. All Ss in this C group had a diagnosis of peripheral neurological disorders. There were 11 males and 14 females in this group, the median age being 48.0 years {range 33-61 ). EEG's were not available in this group. The Ep group consisted of Ss who all at the time of the investigation had the diagnosis of epilepsy based solely on the presence of manifest (clinical) epileptic seizures of any kind. No restrictions as to origin, etiology, presence of EEG abnormality, types and frequency of seizures, etc., were put on this group. Also the Ss at the time of the investigation were varied with respect to drug treatment, a factor very difficult to control in studies like this. There were 31 males and 32 females in this group. The median age being 29.0 years {range 11-63 years). Standard EEG records were available on all Ss in this group. The EEG ex amination usually was given within one month of the RT experiment. All Ss in the BD group had a diagnosis of brain damage based on history, clinical neurolgical examination and one or more of the following radiological evaluations: arteriography, brain scan and pneumoencephalography. The group was heterogeneous with respect to extent, location and etiology, however, predo· minately diffuse lesions were present. Excluded from this group were brain-damaged Ss that manifested any sign of clinical seizures: the mere suspicion of epilepsy would automatically exclude a patient from this group. There were 16 males and 9 females in the group. The median age was 43.0 years (range 22-64 years). Standard EEG records were available on all Ss. ' The groups did not differ significantly in proportion of males and females (chi" = .89, p = n.s.) but did differ in age. The Ep group was significantly younger than C and BD (chi'= 25.13; p < .001); the latter two groups did not differ. Apparatus and Procedure
A vanatlon of the simple visual, continuous RT paradigm used by Bruhn (1970) was employed in this experiment. The visual stimulus consisted of three red diode lamps, each with an intensity of about 1 mcd, spaced 17 mm apart to form the shape of a triangle, and placed about 75 em in front of the S. Stimuli were delivered by a programmed magnetic tape at a rate of 15 per min. with randomized intertrial intervals of 2-6 sec. A pulse from the tape would simulta
Reaction time variability
375
neously turn on the stimulus light and activate an electronic digita~ timer (Type LEC 1004). The Ss task was to terminate the stimuli as fast as possible by depressing a thumb-key (a micro-switch activated by a force of about 60 gr.) held in the preferred hand. A depression of the key would switch off the light and stop the digital clock, showing the RT latencies (from stimulus onset to depression of the response key) in 1/100 of a sec. Through the experiment the Ss were seated in a comfortable chair in a dimly lit room. Earphones provided a low level white noise to mask distracting sounds from outside the room. A warm-up period of 15 stimuli was followed by a taped instruction, urging the S to respond as fast as possible to each stimulus, where after 100 consecutive RTs were measured. As no warnings were given, the alertness of the S had to be maintained throughout a rather monotonous task lasting about 61;2 min. RESULTS
A first step in the data analysis was to find a relevant, numerical expres sion of the RT variability. RT distributions are typically skewed (our data are no exception) and can be transformed by several techniques to a more normal distribution to which parametric statistics can be applied. We wished, however, to keep our data as close to the raw scores as possible to preserve clinical utility. Therefore, we employed medians of the .50 per centile, the .10 percentile and the .90 percentile. These measures have the advantage of providing a test of whether groups differ in their fastest responses or their slowest responses, in addition to their median scores. The intra-subject variability can be expressed numerically as the .90 - .10 percentile difference. In Table I the basic data for the groups are presented. To facilitate reading Table I consider the .10 percentile (fast response) row. The median of the distribution of C scores is 220 with an interquartile range of 201 TABLE I
Group Medians (and Quartiles) of the .10, .50, .90 Percentiles of the Reaction Times for Individual Subjects in the Three Groups. Also, the .90-.10 Percentile Difference and the Median of the Individual Means are Given ( msecs.) Controls (N = 25)
Epileptics (N = 63)
Brain-damaged (N = 25)
Median ( quartiles)
Median (quartiles)
Median ( quartiles)
.90 .90-.10
220 (201-231) 250 (230-270) 310 (270-330) 90 (65-100)
240 (220-270) 285 (260-330) 370 (320-449) 130 (100-199)
Mean
257 (234-277)
296 (270-344)
Percentile .10
.50
240 (220-250)
305 (275-325) 409 ( 344-500) 150 (124-255) 321 (283-357)
376
P. Bruhn and 0. A. Parsons
to 231 msec. compared with the Ep median of 240, interquartile range of 220-250 msec., and the BD median of 240, interquartile range of 220 250 msec. Other rows are read similarly. Mann-Whitney tests applied to group comparison of medians indicate that significant differences were obtained in all comparison,s involving the C group; the latter had significantly faster median RTs than Ep or BD and significantly lower variability (.90 - .10). Ep and BD did not differ significantly in any comparison (Table II). TABLE II
Mann-Whitney U-tests of Differences between Groups
Percentile
C vs. Ep
Ep vs. BD
C vs. BD
.10 .50 .90 .90-.10 Mean
P< .01 p < .001 p < .001 p < .001 p < .001
n.s. n.s. n.s . n.s. n.s.
P< .01 p < .001 p < .001 p < .001 p < .001
Presence and severity of BEG abnormality
EEG abnormalities were graded into several categories: ( 1) not abnor mal, (2) slightly abnormal, (3) slight to moderate, (4) moderate, (5) mo derate to severe and (6) severely abnormal. (For criteria used for this classification see Buchthal and Lennox, 1953.) To reach acceptably large subgroups, categories 3, 4 and 5 were treated as one, giving a total of four subgroups representing increasing severity of abnormality. The first comparison were between patients with abnormal vs. normal EEGs in the Ep group. 38 of 63 Ss in the Ep group were designated sligthly to severely abnormal; the median RT variability scores (.90 - .10) were 130 msec. and 134.5 msec. for normal and abnormal Ss, respectively. Mann-Whit ney U-tests clearly indicate no difference in these groups. Within the abnormal EEG groups no significant differences were obtain~d among the different ca tegories. Similarly in the BD group the 10 Ss with abnormal BEGs had the sam_e median RT variability (150 msec.) as the 15 normal EEG Ss (150 msec.). Kind of BEG abnormality'
Using the standard records, EEG's were scored as to whether the abnormal discharges were diffusedly represented or focal and whether paroxysmal activity 1 The EEG classification was suggested by Dr. 0. Thage, who also assisted in the evaluation of the single records.
Reaction time variability
377
(spikes and sharp waves) or slow wave (less than 6 per sec.) activity was pre sent. Theoretically, in a single S all four types of EEG abnormality may be present (paroxysmal diffuse, paroxysmal focal, slow wave activity diffuse, slow wave activity focal). Comparing Ss having one of the abnormalities mentioned above with those Ss not having the same abnormality did not indicate any significant differences. Although the sample was not analyzed by combinations of EEG abnorma lities, the fact that presence vs. absence of a given EEG abnormality in itself was not differentiating, makes it very unlikely that any kind of EEG discharges should be specifically related to RT variability. Thus kind of EEG abnor mality was not associated with RT variability.
Type of seizures Classifying i:he Ep sample by seizure type, the following categories were used: (1) grand mal(= major motor); (2) petit mal; (3) psychomotor; (4) focal ( = minor motor) and (5) other kinds of seizures. This classification was based on clinical description of the seizures in accordance with directions used traditionally in clinical neurology. Obviously all kinds of seizures inay be present in one S. (Petit mal was only present in 6 Ss and never as the only type of seizure). A first analysis was to score each S in the Ep group at to the presence vs. absence of a specific seizure type (Table III). The group of Ss having grand TABLE III
Relationship between Type of Seizures and Median RT Variability ( msec.) Absent
Type of seizures Grand mal Petit mal Psychomotor Focal Other 1
*
(N (N (N (N (N
= 19) = 57) = 31) = 44) = 61)
119.0 139.0 120.0 130.0 130.0
Present' (N (N (N (N (N
= 44) = 6} = 32) = 19) = 2)
139.5* 100.0 134.5 129.0
Mann-Whitney Test used to test for differences. < .05.
p
mal seizures (N = 44) vs. those not having this type of seizures (N = 19) had median RT variability (.90 - .10) scores of 139.5 msec. and 119.5 msec. respectively. This difference was significant (Mann-Whitney Test p < .05). No other differences were significant. Secondly, an analysis was performed based on subgroups of Ep having only one kind of seizures, i.e., grand mal only (mean = 154 msec.), psychomotor only (mean = 80 msec.), focal only
P. Bruhn and 0. A. Parsons
378
(mean = 114.5 msec.). Although the "psychomotor seizures only" group had a median variability of only 80 msec. as compared to 154 msec. in the "grand mal only" group, no significant difference was reached (Kruskal-Wallis Test) because of a considerable overlap between subgroups.
Frequency of seizures Categorizing the Ss according to ftequency of seizures within the prior year, Ep Ss with less than 4 seizures a year (N = 15), with 4-12 seizures a year (N = 9), and with more than 12 seizures a year (N = 39) were com pared (see Table IV) as to the variability score. For comparative purposes TABLE IV
Frequency of Seizures in Relation to RT Variability ( msec.)
Median RT variability
Frequency of seizures Epileptics
(4 a year) ( 4-12 a year} (> 12 a year)
Controls (no seizures) Brain-damaged (no seizures) 1
(N = 15) (N = 9) (N = 39)
90.0 229.0 139.0
= 25) = 25)
90.0 150.0
(N (N
Differences'
p
= .001
p
< .001
Kruskal-Wallis Test.
C and BD medians also are presented. Ep Ss with less than 4 seizures a year are no different from the controls. A finding more difficult to explain is that the 4-12 seizures a year group shows highly elevated RT variability scores compared to the even more frequently seizuring group. One might speculate whether medication is insufficient in the medium group as 4-12 seizures a year might not alarm the pati~nt or the doctor too much. (The three subgroups were not different in respect to age). It is interesting to note that epileptics with less than 4 seizures have a median RT variability score identical to that of the C group, while those who have greater than 12 seizures per year are close to the BD median score.
Age at onset Other studies have tried to relate age at onset of epilepsy to mental de terioration (Dikmen, Matthews and Harley, 1975). In these studies the age at onset categories were: (a) 5 years or below; (b) 6-16 years and (c) above 16 years at the first seizure. The Ep group was divided using the same age inter
Reaction time variability
379
vals. Group A consisted of 15 Ss (Md RT variability 130 msec.); Group B of 16 Ss (Md 139.5 msec.) and Group C of 32 Ss (Md 129.5 msec.). No significant difference between Ss with early, medium and late onset of the disease was obtained by Kruskal-Wallis Test. (Groups A, B and C were not significantly different in respect to age.) Duration of illness
In all Ep Ss the time since the first epileptic seizure was noted, and a Spearman rank order correlation between duration of illness (in years) and RT variability was computed. In the total Ep sample (N = 36) the Spearman correlation coefficient (rho) was .14. In subgroups of Eps based on etiology: etiology unknown (N = 36), rho = .24; posttraumatic (N = 13), rho = - .11 ; epilepsy secondary to other brain damage (N = 14) rho= .15. No correla tions were significant. Etiology of the epileptic disorder
In order to find out if RT variability was related to the etiology of the disease the Ep group was divided into 3 subgroups: (1) those having epileptic seizures of unknown etiology, i.e., the idiopathic or cryptogenetic group (N = 36), (Md RT variability = 124.5 msec.); (2) those having their seizures related to head trauma, i.e., the posttraumatic group (N = 13 ), (Md = 160 msec.) and finally ( 3) those having seizures as symptoms of known brain damage or disease, i.e., degenerative, vascular, neoplastic. etc. (N = 14), (Md = 129.5 msec.). White the median RT variability was somewhat higher in the posttraumatic group than in the other two groups, the difference was not significant (Kruskal-Wallis Test). (Again, no significant age differences between the subgroups were seen.) Correlation of RT variability and age
Correlating RT variability with age, a positive correlation of rho = .34, p < .01 was found in the Ep group, the older Ep Sc; are more variable than the younger ones. In the C as well as in the Bi[) group, the correlations were close to zero (- .08 and .10 respectively). EP vs. BD comparisons
The second major question was to determine if Ep and BD Ss are different with respect to RT variability and general RT slowing. As seen in Table I and Table II, no significant differences between Ep and BD were found in median RTs or variability.
P. Bruhn and 0. A. Parsons
380
Identification of brain-damaged and epileptics
The final question was whether this rapid technique of measuring RTs could have screening value. In Table V the relevant data are given. The TABLE V
Identification of Controls, Brain-Damaged and Epileptics Using Optimal Cutting Score on Each Measure
Groups
Control (N = 25) Brain-damaged (N = 25) Epileptics (N = 63)
.90- .10 percent. (115 msec.)
.90 percentile (330 msec.)
.50 percentile (288 msec.)
.10 percentile (122 msec.)
Below
Above
Below
Above
Below
Above
Below
Above
92%
8%
80%
20%
80%
20%
80%
20%
16%
84%
20%
80%
31%
69%
28%
72%
30%
70%
29%
71%
43%
57%
38%
62%
best overall discrimination, using optimal cut-off scores, is given .10 measure of variability which identified correctly 92% of 84% of the BD and 70% of the Epileptic patients. The two could not be distinguished from each other on the .90 - .10 any of the others in Table V.
by the .90 the Controls, latter groups measures vs.
DISCUSSION
The present investigation was devoted primarily to the study of RT variability in epileptics and non-epileptic BD Ss. RT variability as the depen dent variable was selected to represent an instance of intra-individual response variability due to the o~inion of the authors, that the mean RT scores traditionally used in most studies conceal very important information on the underlying disturbances of the CNS. Fiske and Rice (1955) define "pure intraindividual variability... as the difference between the two responses of an individual at two points in time under the following conditions: (a) the individual is exposed each time to the same stimulus or to objectively indistinguishable stimuli; (b) the total situation in which the responses are made is the same on both occasions." Obviously, the continuous RT task fulfills this model in a nearly perfect way. The first question concerned RT variability as a function of a number of variables known to be important in epilepsy.
Reaction time variability
381
Is RT variability related to the presence, severity and type of EEG ab normality? The present study, contrary to our expectations, has demonstrated no differences between Ep with and without EEG abnormalities (as was the case with BD Ss with and without EEG disturbances). Nor were there any indications of an increase in variability in the more severely abnormal EEG groups. One explanation to account for these results in comparison with other studies (Browne, Penny, Porter and Dreifuss, 1974; Friedman and Taub, 1969; Porter, Penry and Dreifuss, 1973; Kooi and Hovey, 1957; Mirsky and Van Buren, 1965; Tizard and Margerison, 1963; Woodruff, 1974) is that we used the standard iEEG examination, recorded when the S was relaxed, with eyes closed etc. This obviously is a recording situation very different from the aroused state of the individual participating in a visual RT experiment. Also, only surface-detectable abnormalities picked up from the scalp were recorded. Thus it is possible that patterns of electro physiological abnormalities of a more subtle nature were present. Finally, as the EEG records used here were read by the eye, and not computer-analyzed, only rather prominent disturbances are detected. The observation that many Ss with a definite abnormality of brain function, epilepsy, have a normal EEG pattern (N = 25 or 39%) leads one to conclude that the EEG investigation results in many "false negatives." As stated by Rodin ( 1968 ), " ... a normal EEG in an epileptic does not automatically guarantee normal intellect and/or personality" (p. 155). It may be added: neither does it guarantee stable RT performance. Our results suggest that a continuous RT procedure deserves the status of a supplementary clinical method for diagnostic purposes. Is RT variability related to type of seizures? The generalized conclusion from previous studies seems to be that temporal lobe or psychomotor epi leptics have their major problems in respect to learning and memory (Quad fasel and Pruyser, 1954 ), while centrencephalics (epileptics with grand mal seizures without aura and with generalized paroxysms in their EEGs) have maximal "attentional" problems (Mirsky, Primae, Marsan, Rosvold and Ste vens, 1960). It is noteworthy that, in the present study, patients with grand mal seizures also were more variable. The grand mal seizure is one characterized by loss of consciousness, indicating that the basal arousal systems of the brain are temporarily affected. Psychomotor seizures, quite different in their appearance as well as in their pathophysiological mechanisms, are not characterized by loss of consciousness (although conosciusness is certainly altered). The differential performance found in previous as well as the present study do indicate that interictal performance of these groups are characterized by the same kind of disturbances - although of less intensity - as will be seen during actual seizures.
382
P. Bruhn and 0. A. Parsons
Etiology and age of onset of the epileptic disorder
In neurology Ep Ss will usually be placed in one of three etiological classes: ( 1) the idiophatic (or cryptogenetic) group where no exogeneus etiology is known; (2) a posttraumatic group, where the first seizures appear time related to concussion or head trauma and (3) a symptomatic group where seizures are looked upon as symptoms of other kinds of verified brain damage or disease. From a neurological point of view, the idiopathic group will usually not be considered "brain-damaged." Although this type of Eps certainly demonstrate an abnormal tendency for ·seizures, there is usually no structural evidence to suspect that they show signs of interictal cerebral dysfunction. It is therefore surprising to notice that the group with idiopathic epilepspy are no different from Ep Ss with known brain damage as well as the BD Ss without epilepsy. The RT experiment certainly indicates that their interictal RT pedormances (and probably their brain functions) are impaired to a considerable degree. This is also the conclusion from studies by Kl0ve and Matthews ( 197 4 ), who found that idiopathic Eps were significantly impaired, compared to non-epileptic controls in more complex neuropsycho logical functions. Age of onset was not found to be related to RT variability. As our Ep patients were primarily outpatients and thus able to adapt outside of insti tutions, we may have a biased sample. However, our findings do suggest that generalizations about age of onset and severity of impairment (Kl0ve and Matthews, 197 4) must be limited. The second major question investigated RT pedormance of epileptics and non-epileptic BD Ss. It was suspected on the basis of previous research findings (Bruhn, 1970; Bruhn and Parsons, 1971) that "lapses in pedormance seem to be related to paroxysmal electrobiological disturbances in cortical activation, whereas a general delay seems to be related to structural damage of nervous tissue." This hypothesis was derived on the basis of different studies using different patients as well as different experimental set-ups. This statement was not supportetd by the direct empirical test in the present study. The BD RT distributions were as skewed as those of the Ep and prolonged RTs were found to the same amount in both groups, indicating that RT "lapses" were not phenomena specific to Ep. A nonsignificant difference between the fast end of the individual RT distributions for Ep and BD groups suggests that general slowness is characteristic of both groups to an equal degree. Considering the age differences between the two groups (Ep younger than DB), the deleterious effect of epilepsy on RT performance is rather marked, but caution must be used in interpreting these findings in that drug intake across groups was not equated. In a similar vein it is of interest to note that only in the Ep group was age related to RT performance.
Reaction time variability
383
Whether this reflects an increasing vulnerability of the aging brain to seizures or reflects a complex interaction of anti-convulsant drugs and age remain to be explored in the future. The third question concerned the discriminating power of the RT test. Although we are not able to define those disturbed neurofunctions shared by the two groups, the behavioral similarities observed in RT variability do suggest a very basic disturbance of the GNS functions in both groups. Ep and BD Ss were separated from controls by this 61;2 minute procedure at a level comparable to more sophisticated and time consuming neuro psychological methods. This reliable discrimination of cerebral dysfunction should be of value to the clinical neuropsychologist, but further research in the area is obviously needed. From a neurophysiological approach, one should focus on subtle fluctuations in neural processes underlying intra-individual response variability. Within a behavioral context, the consequences or rela tionships of fluctuating RT latencies on more complex information processing and behavior still has to be determined.
SUMMARY
Median reaction times and intra-individual variability were studied in epi leptic (N = 63 ), brain-damaged (non-epileptic) (N = 25) and control patients (N = 25) using a six and one half minute visual, continuous reaction time task. Epileptic and brain-damaged groups were significantly slower than control patients on median reaction times at the lOth, 50th and 90th percentiles and on the differences between the lOth and 90th percentiles. Thus both general slowing and greater intra-individual variability were found in the epileptic and brain damaged patients. Reaction times were not related to presence, type and severity of EEG abnormality or to age of onset of epilepsy. Grand mal patients did have significantly greater variability than other types of seizure patients. Epileptic and brain-damaged patients did not differ significantly on any reaction time variables. Both groups were discriminated significantly from the controls on all reaction time measures, especially on the intra-individual variability measure. REFERENCES BROWNE, T. R., PENRY,]. K., PoRTER, R.]., and DREIFUSS, F. E. (1974) Responsiveness before, during and after spike-wave paroxysms, "Neurology," 24, 659-665. BRUHN, P. (1970) Disturbances of vigilance in subcortical epilepsy, "Acta Neurologica Scandi navica," 46, 442-454. - , and PARSONS, 0. A. (1971) Continuous reaction time in brain damage, "Cortex," 7, 278-291. BucHTHAL, F., and LENNOX, M. (1953) The EEG effect of Metrazol and photic stimulation in 682 subjects, "EEG Clin. Neurophysiol.," 5, 545-558. DIKMEN, S., MATTHEWS, C. G., and HARLEY, ]. P. (1975) The effect of early versus late onset of major motor epilepsy upon cognitive-intellectual performance, "Epilepsia," 16, 73-81. FISKE, D. W., and RICE, L. (1955) Intra-individual response variability, "Psycho!. Bull.," 52, 217-251. FRIEDMAN, H., and TAUB, H. A. (1969) The transcephalic DC potential and reaction time performance, "Psychophysiology," 5, 504-509.
384
P. Bruhn and 0. A. Parsons
KLjljVE, H., and MATTHEWS, C. G. (1974) Neuropsychological studies of patients with epilepsy, in Clinical neuropsychology: Current status and applications, ed. by R. Reitan and L. A. Davidson, Winston and Wiley, Washington, D.C. Kom, K. A., and HovEY, H. B. (1957) Alterations in mental function and paroxysmal cerebral activity, "A.M.A. Archives of Neurology and Psychiatry," 78, 264-271. MIRSKY, A. F., PRIMAe, D. W., MARSAN, C. A., RosvoLD, H. E., and STEVENS,]. R. (1960) A comparison of the psychological test performance of patients with focal and nonfocal epilepsy, "Experimental Neurology," 2, 75-89. -, and VAN BuREN, ]. M. ( 1973) On the nature of the "absence" in centrencephalic epilepsy: a study of some behavioral, electroencephalographic and autonomic factors, "Electroenceph. Clin. Neurophysiol.," 34, 239-245. PoRTER, R. ]., PENRY,]. K., and DREIFUss, F. E. (1973) Responsiveness at the onset of spikewave bursts, "Electroenceph. Clin. Neurophysiol.," 34, 239-245. QuADFASEL, A. F., and PRuYSER, P. W. (1954) Cognitive deficit in patients with psychomotor epilepsy, "Epilepsia," 3, 80-90. RoDIN, E. A. (1968) The prognosis of patients with epilepsy, Thomas, Springfield, Illinois. TIZARD, B., and MARGERISON, J. H. (1963) The relationship between generalized paroxysmal BEG discharges and various test situations in two epileptic patients, "]. Neural. Neurosurg. Psychiat.," 26, 308-313. WooDRUFF, M. L. (1974) Subconvulsive epileptiform discharge and behavioral impairment, "Behavioral Biology," 11, 431458.
Peter Bruhn, Chief psychologist, Department of Neurology and Neurosurgery, Rigshospitalet, 9 Blegdamsvej, DK-2100 Copenhagen 0, Denmark. Oscar A. Parsons, Ph.D., Professor and Vice Chairman, The University of Oklahoma Health Sciences Center, Department of Psychiatry and Behavioral Sciences, Post Office Box 26901, Oklahoma Ciry, Oklahoma 73190, U.S.A
