Neuropsycholoym, Vol. 29. No Printed in Great Britain.
5. pp. 343-360, 1991
0028-393291 $3.00+0.00 d‘ 1991 Pergamon Press plc
RIGHT TEMPORAL-LOBE CONTRIBUTION VISUAL PROCESSING JULIEN
DOYON*
and BRENDA
TO GLOBAL
MILNER
Department of Psychology and Department of Neurology and Neurosurgery, McGill University and the Montreal Neurological Institute, 3801 University Street, Montreal, Quebec, Canada H3A 2B4 (Receiwd
9 March 1990: accepted 25 January
1991)
Abstract-The performance of 92 patients with unilateral temporal- or frontal-lobe excisions and 35 normal control subjects was tested under two experimental conditions (global, local) of a Stroop-type reaction-time task, employing either hierarchically structured letters or abstract designs as stimuli. In the local condition, subjects were asked to focus their attention on the small forms and to ignore the large form, whereas they were instructed to do the reverse in the global condition. The results showed that, in the local condition, the patients with right temporal-lobe lesions were less affected than the other groups by interference from the global aspect of the stimulus. This reduced sensitivity to the overall configuration of the stimulus was unrelated to the extent of hippocampal removal or to the presence of visual-held defects. These findings support the hypothesis that the human right temporal neocortex contributes to the global processing of visual information.
INTRODUCTION
studies in the monkey have demonstrated impaired performance on a wide variety of visual-discrimination tasks after bilateral removal of the inferior temporal (IT) cortex (for reviews see [S, 16,41,58]). In human subjects, it has also been possible to uncover a visuo-perceptual impairment after anterior temporal-lobe excisions from the right hemisphere, but not from the left [35-37, 55, 561. This perceptual impairment tends to be both milder and more variable than the well-established deficit in visual memory observed after such right anterior temporal-lobe lesions [23, 24, 34, 371 and it is most readily elicited by tasks requiring the integration of fragmented visual information [7,28,3 1,32,37,46,47]. The goal of the present study was to delineate further the type of visual processing that is deficient in patients with right anterior temporal-lobe lesions. Neurophysiological studies looking at the activity of cells in the IT cortex of the monkey have given us some clues to the kind of visual processing that takes place in this area. Several investigators [S-l 1, 18, 21, 501 have identified cells in the IT cortex that are particularly responsive to the overall shape of an object. Such findings have led to the suggestion that the IT cortex plays a major role in global visual processing; that is, in capturing the overall configuration of an object rather than in analysing its local features [ 10, 171. It thus seems possible that the visuo-perceptual deficit seen in patients with right temporal-lobe lesions may reflect an impairment in the ability to form a meaningful whole out of discrete parts. To investigate this question, we have tested patients with unilateral anterior temporal- or BEHAVIOURAL
*Correspondence to be addressed Quebec, Canada G 1K 7P4.
to: Julien
Doyon,
Department
343
of Psychology,
Lava1 University,
Ste-Foy,
344
J. DOYON and B. MILNER
frontal-lobe lesions on a reaction-time, Stroop-type task employing hierarchically structured letters or abstract designs (see Fig. 1). In the first experiment, the stimuli consisted of one large upper case letter (global level) created by the spatial arrangement of identical smaller upper case letters (local level). The stimuli could be either congruent (i.e. where the large and the small letters were the same), or incongruent (i.e. where the two letters were different). The subjects were tested under two experimental conditions (one local, the other global). In the local condition, the subjects were asked to focus their attention on the small letters and to ignore the large letter, whereas they were instructed to do the reverse in the global condition. In both conditions, the subject’s task was to indicate as quickly as possible whether the letter at the attended level was the same as, or dijfierentfiom, a target letter. It was hypothesized that, if the right temporal lobe is critically involved in global visual processing, then, in the local condition, patients with right temporal-lobe lesions should not be influenced as much as the other groups by the overall configuration of the stimulus, and hence that they would show a smaller difference in reaction time between the congruent and the incongruent stimuli. In contrast, in the global condition, it was expected that the patients in the right temporal-lobe group would be slower to respond and would demonstrate a larger interference effect from the local letters than would the other groups.
Congruent
EEEEE E EEEEE E EEEEE b)
TTTTT T T T T
H H H H
HU U HU U H UUUUU HU U HHH U U
Congruent L_lLlLJ
I-Y-h-J LJrl
LII
LJrl
LII
Lib
t_IlLJl-lLJJ
Fig.
Incongruent
1.Representative
rlr-7 rlr-7 r-7 r-9
congruent
17
and incongruent hierarchically designs.
structured
(a) letters and (b) abstract
As hierarchically structured stimuli are perceptually organized into two unequivocally definable levels, they have been employed with normal subjects to investigate the functional hemispheric dissociation between global and local processing. The results have generally indicated a left-hemisphere predominance for local processing and a right-hemisphere
GLOBAL
predominance
for global
VISUAL
PROCESSING
345
processing
(e.g. [Sl]). Since we initiated our studies [13-151, [6, 26, 27, 491 have also conducted a series of experiments in which they have tested small groups of patients with either left or right temporo-parietal lesions, as well as normal control subjects, on tasks demanding the identification of hierarchically structured stimuli. In accordance with normal studies, they found that patients with right-hemisphere damage made more errors [6] and were slower [26,27,49] in recognizing global forms than in recognizing local ones, whereas the reverse was true for patients with left-hemisphere damage. When the clinical groups were subdivided into those with lesions centred in the left superior temporal gyrus (LSTG), those with lesions centred more rostrally in the left inferior parietal lobule (LIPL), and those with lesions in the right temporo-parietal region (RTP), the results suggested that this hemispheric global/local asymmetry might be due to a perceptual mechanism associated with STG lesions and not with IPL lesions [49]. The evidence was clearer for the left hemisphere than for the right, because the patients in the RTP group could not be further subdivided. Thus, the role of the right temporal lobe in the global processing of visual information remains unclear. The two experiments reported below were designed to investigate this question by contrasting the effects of well-circumscribed excisions from the right or left anterior temporal cortex. They were also designed to explore this issue further by comparing the performance of the temporal-lobe groups to that of patients with unilateral frontal lobectomies, and by using not only hierarchically structured letters but abstract designs as well. ROBERTSON and LAMB and their colleagues
EXPERIMENT
1. HIERARCHICALLY
STRUCTURED
LETTER-TASK
The patients who participated, in this experiment, as well as those taking part in Experiment 2, had each undergone a unilateral brain operation at the Montreal Neurological Hospital, the operations being performed for the relief of medically intractable cerebral seizures. In most instances, the epileptogenic lesion had been static and atrophic, dating from birth or early life; however, a few cases ofepilepsy secondary to an indolent tumour or vascular malformation were also included. Patients in whom there was evidence of a rapidly growing neoplasm, diffuse cerebral damage or bilateral independent electrographic abnormality were excluded, as were those with atypical speech representation (as demonstrated by preoperative intracarotid sodium amobarbital tests 138, 541). Also eliminated were patients with a Full Scale IQ (nearest to the time ofexperimental testing) ofless than 80 on Form I or II of the WechslerrBellevue Intelligence Scale, or less than 75 on the Wechsler Adult Intelligence Scale-Revised. Ali the patients included had normal, or corrected-to-normal, visual acuity in both eyes. They were tested either in the early postoperative period (approximately 2 weeks after the operation) or in long-term follow-up (from 1to 22 years after surgery), but since there were no significant differences in performance on the task between those patients seen soon after surgery and those seen in follow-up, the groups were not subdivided according to time of testing. In the present experiment, 76 subjects were assigned to five groups (normal control [NC], n=20; left temporal [LT], n = 20; right temporal CRT], n = 23; left frontal [LF], n = 5; and right frontal [RF], n = 8). Table I shows the sex, age and educational level for all subject groups, and the Full Scale IQ data for the clinical groups.
Table
1. Experiment
Age (year)
Sex Group
M
F
Mean
Range
Normal control Left temporal Right temporal Left frontal Right frontal
10 12 10 1 5
IO 8 13 4 3
27.5 28.2 29.5 27.6 35.1
1641 l&39 1544 1635 2647
1. Subjects Education Mean 13.1 12.7 13.0 13.0 12.5
(year) Range IO--21 618 9918 8-17 3-22
Wechsler Mean
IQ Range
not assessed 106 84135 111 96-135 96 9G113 110 91-132
J. DOYON and B. MILNER
346
Temporal-lobe yroups.The cortical resections in the 43 patients with temporal-lobe removals always included the anterior neocortex and the amygdala but varied in the amount ofhippocampus and parahippocampal gyrus excised. Although the removals from the right hemisphere tended to be larger than those from the left, there was considerable overlap between the groups in this respect and there was no correlation in either group between the extent of removal and performance on the present task.
Left temporal-lobe
group
In this group, the mean extent of removal along the Sylvian fissure was 4.8 cm, ranging from 4.0 to 6.5 cm, and the mean extent along the base was 4.8, with a range of 3.556.5 cm. The group included one case of oligodendroglioma (grade 1). Half of the patients were tested postoperatively, whilst the other halfwere seen during long-term follow-up studies.
Right temporal-lobe
group
For these patients, the mean extent of removal along the Sylvian fissure was 5.0 cm, ranging from 4.5 to 8.0 cm, and the mean extent along the base of the brain was 5.6, with a range of4.558.0 cm. This group contained one case of ganglioglioma, one case of low-grade astrocytoma and two cases of vascular malformation. Fifteen patients were tested 2 weeks after surgery, the remaining eight being seen in follow-up. Frontul-lobe
groups.
frontal-lobe removals present the individual
In the excisions from the left frontal lobe, Broca’s area was always spared; otherwise, the from either hemisphere varied considerably in locus and extent. For this reason, Figs 2 and 3 brain maps of the patients with frontal-lobe lesions.
Left frontal-lobe
group
The LF group (see Fig. 2) contained one patient (An. Be.) with only the most anterior part of the lateral convexity removed; two patients (Mi. Mi., Ho. Sp.) who underwent excisions invading the orbito-frontal area, and two others (MO. Ma.-Au., Lo. La.) who underwent more extensive frontal lobectomies. This group included two cases of lowgrade astrocytoma (Ho. Sp., Lo. La.) and one case of ganglioglioma (Mi. Mi.). Two patients (Ho. Sp., Lo. La.) were tested approximately 2 weeks after surgery and the other three patients during follow-up study.
Right ,frontal-lohr
group
The excisions in the RF group (see Fig. 3) varied from restricted removals confined either to the dorsolaterdl and dorsomedial regions (La. Ho., Br. Fe.), or to parts of the lateral convexity (Ja. Ne., Ma. Ak.), or to the inferior and orbito-frontal regions (Br. Fo.). Three patients had more radical frontal lobectomies, involving the lateral, medial and orbito-frontal areas (J-P. Bi., Lo. Th., Jo. Ed.). The RF group contained two cases of low-grade astrocytoma (Br. Fe., Ma. Ak.), one case ofohgodendroglioma (Br. Fo.), one case of previous brain abscess (Jo. Ed.), one case of scarring subsequent to adult head trauma (J-P. Bi.) and one case of angioma (La. Ho.). Two patients (La. Ho., Ma. Ak.) were tested 2 weeks postoperatively, and the remainder were seen in follow-up.
Normal
confrol
group
A group of 20 right-handed normal control subjects was chosen to match the patient groups as closely as possible with respect to sex, age and educational level. These subjects were either hospital support staff, friends of the first author or relatives of patients. They all had normal or corrected-to-normal vision in both eyes and none had any history of trauma affecting the central nervous system. Stimuli
and apparatus
The results of previous studies [19.25,30,52] have shown that differences in various parameters, such as size of visual angle subtended, matrix size and sparsity, will determine the level (global or local) of the hierarchically structured-letter stimuli that will be processed first. Taking guidelines from HOFFMAN1191, an attempt was made to construct stimuli that would produce roughly the same interference effect from congruent to incongruent stimuli in both global and local conditions. The stimuli were composed of small, upper case letters arranged into the shape of one large upper case letter (see Fig. la). Both the small and large letters were selected from the same set (E, T, U or H). The large letter was formed by placing identical small letters in the appropriate positions of a 5 x 5 matrix. The small letters were 0.9 cm high by 0.7 cm wide, whereas the large letter was 5.2 cm high by 4.4 cm wide.
GLOBAL
VISUAL
PROCESSING
347
Mi. Mi.
m
MO. Ma.-Au ‘F-2------
Fig. 2. Diagrams based on the surgeon’s drawings at the time of operation, illustrating the estimated medial, lateral and ventral extent (when available) of cortical removal for individual patients in the left frontal-lobe group. The shaded areas indicate functionally disconnected cortex. Each composite stimulus could were used for both congruent and Apple-soft basic language on an using a 12-inch black and white number of correct and incorrect
be either congruent or incongruent. incongruent stimuli. The letters were Apple II + computer, and they were Sony receiver/monitor (Model CVM choices, as well as the reaction time
All possible combinations of the four letters generated by a computer program written in displayed in white on a black background, 131). The computer program recorded the (RT) in msec for both types of response.
Procedure
The subject was seated in front of a table in a dark room with the head positioned in a chin-rest, so as to keep the viewing distance constant at 170 cm. Under these conditions, the small letters subtended 0.30” by 0.24” of visual angle, whereas the large letter subtended I .75” by 1.48” of visual angle. The luminosity and contrast of the stimuli were kept constant throughout. Each trial (see Fig. 4) began by the presentation, at the top of the screen, of a target letter (i.e. E, T, U or H), which stayed on for the duration of the trial. The target letter measured approximately twice the size of a small letter (i.e. 1.5cm high by 1.O cm wide, or 0.51” by 0.34” of visual angle). After 1 set, a central fixation-point was displayed for 800 msec. This fixation-point then disappeared and was replaced by one of the large letters made up of small letters. The trial was terminated by the subject’s response; the stimuli were then erased, and this was followed by an intertrial interval of 2500 msec, during which the screen stayed black.
348
J. DOYON and B. MILNER
Lo. Th.
/.
Jo. Ed.
_
i
Br. Fo.
J-P. Bi.
c3
m
Fig. 3. Brain maps representing
the estimated extent of neocortical excision for the individual in the right frontal-lobe group.
patients
There were two experimental conditions: ylobal and local. The subject’s task in the global condition was to respond by pressing one of the two keys (‘7” or “/“) on the computer keyboard with the left or right index finger, depending on whether the large letter was the Same as, or differentfrom, the target letter. The subjects were instructed to respond as quickly and accurately as possible and to ignore the small letters. In the local condition, the procedure was identical to the above, except that the subjects were now asked to focus on the small letters and to ignore the large letter. These experimental conditions were administered to each subject in two sessions, which took place on consecutive days. The order of presentation of the conditions, as well as the hand used to respond “same” or “different”, were counterbalanced within each group. Each session lasted approximately 30 min and consisted ofone series of 32 practice trials, followed by three series of 64 experimental trials each. In both the practice and the experimental series, the trials were presented in a randomized order of congruent and incongruent items. Six different versions of the same randomized order of stimuli were used, but the order of administration of these six versions (1 6) was the same for all subjects.
Results
Separate one-way analyses of variance indicated no significant difference between the subject groups with respect to age or education. Although in Table 1 the LF group shows a lower mean Full Scale IQ than the other patient groups, the results of the one-way statistical analysis on the IQ data also failed to reach significance.
c
GLOBAL
VISUAL
PROCESSING
E
Congruent
Stimuli
Incongruent
n
m
I
L
1000ms
I
800 ms
c cl
H
I
2500 ms
-
RT
349
,
Press appropriate response key
Fig. 4. This diagram illustrates the procedure used for each experimental trial in the hierarchical-letter task. One example of both congruent (top) and incongruent (bottom) stimuli are shown here. For both tasks, all trials began with the presentation of a target form at the top of the screen, which stayed on for the duration of the trial. After 1000 msec, a central fixation-point was displayed, which was then replaced 800 msec later by one of the hierarchical stimuli. In the local condition, the subject’s task was to indicate whether the small forms were the same as, or dzfirent from, the target form, whereas the subject was instructed to do the reverse in the local condition. After the subject had responded, both the target and the stimulus were erased and this was followed by an inter-trial interval of 2500 msec. Only examples of “same” responses are illustrated in this figure.
The two dependent measures of interest in this experiment, as well as in Experiment 2 to be reported below, were the mean number of errors and the mean reaction time for congruent and incongruent stimuli, respectively, in both global and local conditions. Each measure was analysed separately for the global and local conditions using an analysis of variance with two repeated measures, i.e. condition (global vs local) and stimulus-type (congruent vs incongruent). Separate analyses were justified, since different predictions were made for the two conditions. The post hoc comparisons were always performed using the NewmanKeuls procedure [59]. Errors. The mean number of errors for each subject group was less than 8% of the total number of trials in both global and local conditions. In general, the analyses of variance for repeated measures yielded a significant main effect of group in the local condition only, F (4, 71)= 2.61, P~0.05, and subsequent post hoc tests indicated that the LF group had made significantly more errors than any of the other groups, NC: Q=5.04, PcO.01; LT: Q=3.89, P
5. pp. 343-360, 1991
0028-393291 $3.00+0.00 d‘ 1991 Pergamon Press plc
RIGHT TEMPORAL-LOBE CONTRIBUTION VISUAL PROCESSING JULIEN
DOYON*
and BRENDA
TO GLOBAL
MILNER
Department of Psychology and Department of Neurology and Neurosurgery, McGill University and the Montreal Neurological Institute, 3801 University Street, Montreal, Quebec, Canada H3A 2B4 (Receiwd
9 March 1990: accepted 25 January
1991)
Abstract-The performance of 92 patients with unilateral temporal- or frontal-lobe excisions and 35 normal control subjects was tested under two experimental conditions (global, local) of a Stroop-type reaction-time task, employing either hierarchically structured letters or abstract designs as stimuli. In the local condition, subjects were asked to focus their attention on the small forms and to ignore the large form, whereas they were instructed to do the reverse in the global condition. The results showed that, in the local condition, the patients with right temporal-lobe lesions were less affected than the other groups by interference from the global aspect of the stimulus. This reduced sensitivity to the overall configuration of the stimulus was unrelated to the extent of hippocampal removal or to the presence of visual-held defects. These findings support the hypothesis that the human right temporal neocortex contributes to the global processing of visual information.
INTRODUCTION
studies in the monkey have demonstrated impaired performance on a wide variety of visual-discrimination tasks after bilateral removal of the inferior temporal (IT) cortex (for reviews see [S, 16,41,58]). In human subjects, it has also been possible to uncover a visuo-perceptual impairment after anterior temporal-lobe excisions from the right hemisphere, but not from the left [35-37, 55, 561. This perceptual impairment tends to be both milder and more variable than the well-established deficit in visual memory observed after such right anterior temporal-lobe lesions [23, 24, 34, 371 and it is most readily elicited by tasks requiring the integration of fragmented visual information [7,28,3 1,32,37,46,47]. The goal of the present study was to delineate further the type of visual processing that is deficient in patients with right anterior temporal-lobe lesions. Neurophysiological studies looking at the activity of cells in the IT cortex of the monkey have given us some clues to the kind of visual processing that takes place in this area. Several investigators [S-l 1, 18, 21, 501 have identified cells in the IT cortex that are particularly responsive to the overall shape of an object. Such findings have led to the suggestion that the IT cortex plays a major role in global visual processing; that is, in capturing the overall configuration of an object rather than in analysing its local features [ 10, 171. It thus seems possible that the visuo-perceptual deficit seen in patients with right temporal-lobe lesions may reflect an impairment in the ability to form a meaningful whole out of discrete parts. To investigate this question, we have tested patients with unilateral anterior temporal- or BEHAVIOURAL
*Correspondence to be addressed Quebec, Canada G 1K 7P4.
to: Julien
Doyon,
Department
343
of Psychology,
Lava1 University,
Ste-Foy,
344
J. DOYON and B. MILNER
frontal-lobe lesions on a reaction-time, Stroop-type task employing hierarchically structured letters or abstract designs (see Fig. 1). In the first experiment, the stimuli consisted of one large upper case letter (global level) created by the spatial arrangement of identical smaller upper case letters (local level). The stimuli could be either congruent (i.e. where the large and the small letters were the same), or incongruent (i.e. where the two letters were different). The subjects were tested under two experimental conditions (one local, the other global). In the local condition, the subjects were asked to focus their attention on the small letters and to ignore the large letter, whereas they were instructed to do the reverse in the global condition. In both conditions, the subject’s task was to indicate as quickly as possible whether the letter at the attended level was the same as, or dijfierentfiom, a target letter. It was hypothesized that, if the right temporal lobe is critically involved in global visual processing, then, in the local condition, patients with right temporal-lobe lesions should not be influenced as much as the other groups by the overall configuration of the stimulus, and hence that they would show a smaller difference in reaction time between the congruent and the incongruent stimuli. In contrast, in the global condition, it was expected that the patients in the right temporal-lobe group would be slower to respond and would demonstrate a larger interference effect from the local letters than would the other groups.
Congruent
EEEEE E EEEEE E EEEEE b)
TTTTT T T T T
H H H H
HU U HU U H UUUUU HU U HHH U U
Congruent L_lLlLJ
I-Y-h-J LJrl
LII
LJrl
LII
Lib
t_IlLJl-lLJJ
Fig.
Incongruent
1.Representative
rlr-7 rlr-7 r-7 r-9
congruent
17
and incongruent hierarchically designs.
structured
(a) letters and (b) abstract
As hierarchically structured stimuli are perceptually organized into two unequivocally definable levels, they have been employed with normal subjects to investigate the functional hemispheric dissociation between global and local processing. The results have generally indicated a left-hemisphere predominance for local processing and a right-hemisphere
GLOBAL
predominance
for global
VISUAL
PROCESSING
345
processing
(e.g. [Sl]). Since we initiated our studies [13-151, [6, 26, 27, 491 have also conducted a series of experiments in which they have tested small groups of patients with either left or right temporo-parietal lesions, as well as normal control subjects, on tasks demanding the identification of hierarchically structured stimuli. In accordance with normal studies, they found that patients with right-hemisphere damage made more errors [6] and were slower [26,27,49] in recognizing global forms than in recognizing local ones, whereas the reverse was true for patients with left-hemisphere damage. When the clinical groups were subdivided into those with lesions centred in the left superior temporal gyrus (LSTG), those with lesions centred more rostrally in the left inferior parietal lobule (LIPL), and those with lesions in the right temporo-parietal region (RTP), the results suggested that this hemispheric global/local asymmetry might be due to a perceptual mechanism associated with STG lesions and not with IPL lesions [49]. The evidence was clearer for the left hemisphere than for the right, because the patients in the RTP group could not be further subdivided. Thus, the role of the right temporal lobe in the global processing of visual information remains unclear. The two experiments reported below were designed to investigate this question by contrasting the effects of well-circumscribed excisions from the right or left anterior temporal cortex. They were also designed to explore this issue further by comparing the performance of the temporal-lobe groups to that of patients with unilateral frontal lobectomies, and by using not only hierarchically structured letters but abstract designs as well. ROBERTSON and LAMB and their colleagues
EXPERIMENT
1. HIERARCHICALLY
STRUCTURED
LETTER-TASK
The patients who participated, in this experiment, as well as those taking part in Experiment 2, had each undergone a unilateral brain operation at the Montreal Neurological Hospital, the operations being performed for the relief of medically intractable cerebral seizures. In most instances, the epileptogenic lesion had been static and atrophic, dating from birth or early life; however, a few cases ofepilepsy secondary to an indolent tumour or vascular malformation were also included. Patients in whom there was evidence of a rapidly growing neoplasm, diffuse cerebral damage or bilateral independent electrographic abnormality were excluded, as were those with atypical speech representation (as demonstrated by preoperative intracarotid sodium amobarbital tests 138, 541). Also eliminated were patients with a Full Scale IQ (nearest to the time ofexperimental testing) ofless than 80 on Form I or II of the WechslerrBellevue Intelligence Scale, or less than 75 on the Wechsler Adult Intelligence Scale-Revised. Ali the patients included had normal, or corrected-to-normal, visual acuity in both eyes. They were tested either in the early postoperative period (approximately 2 weeks after the operation) or in long-term follow-up (from 1to 22 years after surgery), but since there were no significant differences in performance on the task between those patients seen soon after surgery and those seen in follow-up, the groups were not subdivided according to time of testing. In the present experiment, 76 subjects were assigned to five groups (normal control [NC], n=20; left temporal [LT], n = 20; right temporal CRT], n = 23; left frontal [LF], n = 5; and right frontal [RF], n = 8). Table I shows the sex, age and educational level for all subject groups, and the Full Scale IQ data for the clinical groups.
Table
1. Experiment
Age (year)
Sex Group
M
F
Mean
Range
Normal control Left temporal Right temporal Left frontal Right frontal
10 12 10 1 5
IO 8 13 4 3
27.5 28.2 29.5 27.6 35.1
1641 l&39 1544 1635 2647
1. Subjects Education Mean 13.1 12.7 13.0 13.0 12.5
(year) Range IO--21 618 9918 8-17 3-22
Wechsler Mean
IQ Range
not assessed 106 84135 111 96-135 96 9G113 110 91-132
J. DOYON and B. MILNER
346
Temporal-lobe yroups.The cortical resections in the 43 patients with temporal-lobe removals always included the anterior neocortex and the amygdala but varied in the amount ofhippocampus and parahippocampal gyrus excised. Although the removals from the right hemisphere tended to be larger than those from the left, there was considerable overlap between the groups in this respect and there was no correlation in either group between the extent of removal and performance on the present task.
Left temporal-lobe
group
In this group, the mean extent of removal along the Sylvian fissure was 4.8 cm, ranging from 4.0 to 6.5 cm, and the mean extent along the base was 4.8, with a range of 3.556.5 cm. The group included one case of oligodendroglioma (grade 1). Half of the patients were tested postoperatively, whilst the other halfwere seen during long-term follow-up studies.
Right temporal-lobe
group
For these patients, the mean extent of removal along the Sylvian fissure was 5.0 cm, ranging from 4.5 to 8.0 cm, and the mean extent along the base of the brain was 5.6, with a range of4.558.0 cm. This group contained one case of ganglioglioma, one case of low-grade astrocytoma and two cases of vascular malformation. Fifteen patients were tested 2 weeks after surgery, the remaining eight being seen in follow-up. Frontul-lobe
groups.
frontal-lobe removals present the individual
In the excisions from the left frontal lobe, Broca’s area was always spared; otherwise, the from either hemisphere varied considerably in locus and extent. For this reason, Figs 2 and 3 brain maps of the patients with frontal-lobe lesions.
Left frontal-lobe
group
The LF group (see Fig. 2) contained one patient (An. Be.) with only the most anterior part of the lateral convexity removed; two patients (Mi. Mi., Ho. Sp.) who underwent excisions invading the orbito-frontal area, and two others (MO. Ma.-Au., Lo. La.) who underwent more extensive frontal lobectomies. This group included two cases of lowgrade astrocytoma (Ho. Sp., Lo. La.) and one case of ganglioglioma (Mi. Mi.). Two patients (Ho. Sp., Lo. La.) were tested approximately 2 weeks after surgery and the other three patients during follow-up study.
Right ,frontal-lohr
group
The excisions in the RF group (see Fig. 3) varied from restricted removals confined either to the dorsolaterdl and dorsomedial regions (La. Ho., Br. Fe.), or to parts of the lateral convexity (Ja. Ne., Ma. Ak.), or to the inferior and orbito-frontal regions (Br. Fo.). Three patients had more radical frontal lobectomies, involving the lateral, medial and orbito-frontal areas (J-P. Bi., Lo. Th., Jo. Ed.). The RF group contained two cases of low-grade astrocytoma (Br. Fe., Ma. Ak.), one case ofohgodendroglioma (Br. Fo.), one case of previous brain abscess (Jo. Ed.), one case of scarring subsequent to adult head trauma (J-P. Bi.) and one case of angioma (La. Ho.). Two patients (La. Ho., Ma. Ak.) were tested 2 weeks postoperatively, and the remainder were seen in follow-up.
Normal
confrol
group
A group of 20 right-handed normal control subjects was chosen to match the patient groups as closely as possible with respect to sex, age and educational level. These subjects were either hospital support staff, friends of the first author or relatives of patients. They all had normal or corrected-to-normal vision in both eyes and none had any history of trauma affecting the central nervous system. Stimuli
and apparatus
The results of previous studies [19.25,30,52] have shown that differences in various parameters, such as size of visual angle subtended, matrix size and sparsity, will determine the level (global or local) of the hierarchically structured-letter stimuli that will be processed first. Taking guidelines from HOFFMAN1191, an attempt was made to construct stimuli that would produce roughly the same interference effect from congruent to incongruent stimuli in both global and local conditions. The stimuli were composed of small, upper case letters arranged into the shape of one large upper case letter (see Fig. la). Both the small and large letters were selected from the same set (E, T, U or H). The large letter was formed by placing identical small letters in the appropriate positions of a 5 x 5 matrix. The small letters were 0.9 cm high by 0.7 cm wide, whereas the large letter was 5.2 cm high by 4.4 cm wide.
GLOBAL
VISUAL
PROCESSING
347
Mi. Mi.
m
MO. Ma.-Au ‘F-2------
Fig. 2. Diagrams based on the surgeon’s drawings at the time of operation, illustrating the estimated medial, lateral and ventral extent (when available) of cortical removal for individual patients in the left frontal-lobe group. The shaded areas indicate functionally disconnected cortex. Each composite stimulus could were used for both congruent and Apple-soft basic language on an using a 12-inch black and white number of correct and incorrect
be either congruent or incongruent. incongruent stimuli. The letters were Apple II + computer, and they were Sony receiver/monitor (Model CVM choices, as well as the reaction time
All possible combinations of the four letters generated by a computer program written in displayed in white on a black background, 131). The computer program recorded the (RT) in msec for both types of response.
Procedure
The subject was seated in front of a table in a dark room with the head positioned in a chin-rest, so as to keep the viewing distance constant at 170 cm. Under these conditions, the small letters subtended 0.30” by 0.24” of visual angle, whereas the large letter subtended I .75” by 1.48” of visual angle. The luminosity and contrast of the stimuli were kept constant throughout. Each trial (see Fig. 4) began by the presentation, at the top of the screen, of a target letter (i.e. E, T, U or H), which stayed on for the duration of the trial. The target letter measured approximately twice the size of a small letter (i.e. 1.5cm high by 1.O cm wide, or 0.51” by 0.34” of visual angle). After 1 set, a central fixation-point was displayed for 800 msec. This fixation-point then disappeared and was replaced by one of the large letters made up of small letters. The trial was terminated by the subject’s response; the stimuli were then erased, and this was followed by an intertrial interval of 2500 msec, during which the screen stayed black.
348
J. DOYON and B. MILNER
Lo. Th.
/.
Jo. Ed.
_
i
Br. Fo.
J-P. Bi.
c3
m
Fig. 3. Brain maps representing
the estimated extent of neocortical excision for the individual in the right frontal-lobe group.
patients
There were two experimental conditions: ylobal and local. The subject’s task in the global condition was to respond by pressing one of the two keys (‘7” or “/“) on the computer keyboard with the left or right index finger, depending on whether the large letter was the Same as, or differentfrom, the target letter. The subjects were instructed to respond as quickly and accurately as possible and to ignore the small letters. In the local condition, the procedure was identical to the above, except that the subjects were now asked to focus on the small letters and to ignore the large letter. These experimental conditions were administered to each subject in two sessions, which took place on consecutive days. The order of presentation of the conditions, as well as the hand used to respond “same” or “different”, were counterbalanced within each group. Each session lasted approximately 30 min and consisted ofone series of 32 practice trials, followed by three series of 64 experimental trials each. In both the practice and the experimental series, the trials were presented in a randomized order of congruent and incongruent items. Six different versions of the same randomized order of stimuli were used, but the order of administration of these six versions (1 6) was the same for all subjects.
Results
Separate one-way analyses of variance indicated no significant difference between the subject groups with respect to age or education. Although in Table 1 the LF group shows a lower mean Full Scale IQ than the other patient groups, the results of the one-way statistical analysis on the IQ data also failed to reach significance.
c
GLOBAL
VISUAL
PROCESSING
E
Congruent
Stimuli
Incongruent
n
m
I
L
1000ms
I
800 ms
c cl
H
I
2500 ms
-
RT
349
,
Press appropriate response key
Fig. 4. This diagram illustrates the procedure used for each experimental trial in the hierarchical-letter task. One example of both congruent (top) and incongruent (bottom) stimuli are shown here. For both tasks, all trials began with the presentation of a target form at the top of the screen, which stayed on for the duration of the trial. After 1000 msec, a central fixation-point was displayed, which was then replaced 800 msec later by one of the hierarchical stimuli. In the local condition, the subject’s task was to indicate whether the small forms were the same as, or dzfirent from, the target form, whereas the subject was instructed to do the reverse in the local condition. After the subject had responded, both the target and the stimulus were erased and this was followed by an inter-trial interval of 2500 msec. Only examples of “same” responses are illustrated in this figure.
The two dependent measures of interest in this experiment, as well as in Experiment 2 to be reported below, were the mean number of errors and the mean reaction time for congruent and incongruent stimuli, respectively, in both global and local conditions. Each measure was analysed separately for the global and local conditions using an analysis of variance with two repeated measures, i.e. condition (global vs local) and stimulus-type (congruent vs incongruent). Separate analyses were justified, since different predictions were made for the two conditions. The post hoc comparisons were always performed using the NewmanKeuls procedure [59]. Errors. The mean number of errors for each subject group was less than 8% of the total number of trials in both global and local conditions. In general, the analyses of variance for repeated measures yielded a significant main effect of group in the local condition only, F (4, 71)= 2.61, P~0.05, and subsequent post hoc tests indicated that the LF group had made significantly more errors than any of the other groups, NC: Q=5.04, PcO.01; LT: Q=3.89, P
