,Nuurop~ycholoyia. Vol. 29, No. 9, pp. 849-859, Prmled m Great Bntain.
THE FRONTAL
CORTEX
MARY Montreal
Neurological
0028--3932191 .$3.N+0.00 0 1991 Pergamon Press plc
1991.
AND MEMORY ORDER
PAT MCANDREWS*
Institute
and BRENDA
and the Department of Neurology Montreal, Quebec, Canada
(Receiced
20 July
1990; accepted
FOR TEMPORAL
MILNER
and Neurosurgery,
6 April
McGill
University,
1991)
Abstract-Patients with unilateral frontal- or temporal-lobe excisions and normal control subjects made relative-recency decisions about objects presented sequentially. Several objects within each series were presented in the context of actions to be performed using them, such as “squeeze the sponge”, whereas others had only to be named. Both left and right frontal-lobe groups were impaired on order judgements for named items, but their performance was normal for action items. The results suggest that providing salient and distinctive items, involving meaningful actions and multimodal cues, helps compensate for deficits in memory for temporal information associated with frontal-lobe damage.
INTRODUCTION
MANY CURRENTmodels
of frontal-lobe functioning propose that this region of cortex plays an important role in memory, although there is no general agreement as to the specific mnemonic processes involved [7, 11,24,42]. JACOBSEN [ 151 was the first to demonstrate a memory impairment on delayed-response and delayed-alternation tasks in monkeys with bilateral removals from the frontal cortex. In these tasks, the animal must remember, over a brief interval, which of two identical food wells had been baited (delayed response) or which had been associated with reward on the previous trial (delayed alternation). Jacobsen’s discovery of a profound impairment in animals with frontal-lobe lesions has been replicated so often that normal performance on these tasks is considered a direct reflection of the functional integrity of dorsolateral frontal cortex. Although several studies have shown that the capacity for short-term spatial memory is critical to success on these tasks [lo, 261, others have emphasized the requirement for adequate registration and retention of temporal information [S, 25, 331. That is, the same two events and possible choices occur repeatedly, and the animal must remember which event occurred on the most recent trial in order to respond correctly. This interpretation, which is consistent with Jacobsen’s own explanation of the deficit, has led to a focus on the specific role of the frontal cortex in memory for the temporal order of events. Over the past two decades, research by Milner and her colleagues has shown that patients *Please address all correspondence to: M. P. McAndrews, Research Foundation, 33 Russell Street, Toronto, Ontario, 849
Department of Sociobehavioral Canada M5S 2Sl.
Research,
Addiction
850
M.
P. MCANUKEWS and B. MILNFK
with unilateral frontal-lobe lesions are impaired at monitoring and remembering the temporal order of contextually similar events. Direct evidence of this impairment was first obtained using a relative-recency discrimination task 1211. Subjects were shown a long series of items sequentially (words, representational drawings, or abstract paintings), and were asked periodically to decide which of two test items had appeared more recently in the series. Some test cards contained one item from the series and one new foil, in which case the discrimination was based solely on item recognition. Although patients with frontal-lobe lesions were able to discriminate between old and new test items as well as normal control subjects, their performance was markedly impaired on the recency discriminations. The observed impairments varied with the laterality of the excision and the type of material, in that patients with right frontal-lobe lesions showed marked deficits in recency discrimination for representational drawings and paintings, whereas those with left frontal-lobe excisions were impaired only on the verbal task. Patients with temporal-lobe lesions showed the opposite pattern; normal recency discrimination but poor item recognition in some cases [21,22]. PETRIDESand MILNER [3 I] reported a similar deficit on a task that required patients to select a different item from a repeated stimulus array by keeping track of their responses across successive trials. Similar deficits have subsequently been reported on tasks involving recency discrimination or reconstruction of temporal order in patients with epileptogenic frontal-lobe lesions [17] and in other groups of patients with unilateral or bilateral frontallobe excisions [39]. There have been few attempts to characterize more precisely the processes or mechanisms that could account for deficits observed on tasks requiring memory for temporal order. PRIBRAM and his colleagues [32,33] have argued that deficits on delayed alternation reflect a failure to parse or segment the ongoing stream of experience into discrete “temporal moments”. This notion arose from studies showing that monkeys with dorsolateral frontallobe excisions were unimpaired when the experimenter imposed external “temporal landmarks” by asymmetrically manipulating the duration of the delay period between trials (see 1341 for a criticism of this interpretation). MILNER[21,24] speculated that frontal-lobe damage might compromise encoding or retrieval of “time tags” hypothesized to be laid down as part of the mnemonic record of experienced events 1461. SCHACER 1371 argued for an impairment in automatic encoding of spatiotemporal information, whilst other theorists have emphasized the role of active strategies and reconstruction in memory for temporal order 120, 27, 42, 4.51. The research reported here examined further the nature of this deficit by introducing a new type of material that might benefit the frontal-lobe patients in the recencydiscrimination task. We refer to these new memoranda as action items, although they have been called “subject-performed tasks” in previous research [4]. For each action item, the subject enacts a simple instruction for using an object, such as “break the stick”, or “bounce the ball”. Several considerations led us to believe that action items would be especially powerful memoranda in the recency-discrimination task. Previous research has shown that recall for such items is generally superior to recall of materials presented in other ways, such as lists of words or experimenter-performed actions. In addition, recall of action items, unlike recall of most other types of material, is relatively unaffected by large variations in subject characteristics and presentation variables, such as age, level of general intelligence, or the extent of semantic elaboration of items at study [ 1,4,5, 14,301. Given that such variables are thought to influence strategic or effortful encoding and retrieval processes, COHEN [S] has
THE FRONTAL
CORTEX
AND
MEMORY
FOR TEMPORAL
ORDER
851
suggested that memory for enacted events is largely insensitive to the use of mnemonic strategies. As well, the action items we used involved distinctive actions and multimodal cues, features that should increase the salience of the material. Finally, GENTILE and S~AMM [9] reported improved performance by monkeys with frontal-lobe lesions in the delayedalternation task when different movements were required to displace manipulanda over the food wells, a finding that indicates the contribution of polysensory action cues to performance on this task. In the present experiment, patients with unilateral frontal- or temporal-lobe excisions and normal control subjects made relative-recency discriminations about pairs of objects presented in a study series. Only a few objects were manipulated by subjects, the remainder being merely named. Given our expectation that patients with frontal-lobe excisions would be impaired at recency judgements for named items, the principal hypothesis tested here was that they would perform normally on the action items. It was necessary, however, to examine the potential influence of action items on memory for named items presented in the same series. For example, it was possible that the action items could enhance processing of certain named objects, by serving as potential temporal “landmarks” separating the named items into those presented “before” and “after” a given manipulation. To evaluate this possibility, discrimination tests were constructed such that some pairs of named items were separated by an action “landmark” in the presentation sequence, whereas other pairs consisted of items that had not been separated by an intervening action. A second concern was that the enacted items might interfere with the processing of adjacent named items, as has been shown with other unique or high-priority events embedded in a series of neutral homogeneous events 1381. To control for this possibility, subjects were also tested on additional series of objects (i.e. “homogenous” series) in which there were no action items. Recent findings 1401 indicate that patients with frontal-lobe lesions may have difficulty with other kinds of memory judgements, such as those concerning frequency of occurrence. Accordingly, we included another type of memory decision that required subjects to remember and compare the sizes of the objects. Although size can be considered an intrinsic property of objects, rather than a contextual attribute, it is conceivable that the operations involved in making comparative judgements about remembered events along this dimension would also be impaired in patients with frontal-lobe lesions.
METHOD The participants in this study comprised 52 patients of the Montreal Neurological Hospital, each of whom had undergone a unilateral brain operation for the relief of pharmacologically intractable epilepsy, and 20 normal control subjects. Except where stated. the epileptogenic lesions were static and atrophic. Patients with a Full-Scale IQ rating below 75 on the Wechsler Adult Intelligence Scale-Revised were excluded from the study, as were those with cvidcnce of diffuse cerebral damage or bilateral independent electrographic abnormalities. Table I shows the sex distribution, mean age, Full-Scale IQ rating and educational level of subjects in the various groups. Frontcll-lohe groups. Figures I and 2 illustrate the excisions of patients in the left and right frontal-lobe groups. respectively. These diagrams are based on the surgeons’ drawings at the time of operation. In all patients but one (Ho.Sp.). the excision included part of the dorsolateral frontal cortex, and in some patients the excision extended Into the medial frontal region. Broca’s area was spared in all patients with left frontal-lobe (LF) excisions. In the LF group. there were three cases of indolent brain tumor (Li.Ar.. Lo.La. and HOSP.) and one case of arteriovenous malformation(A.-M.Be.). The right frontat-lobe (RF) group also included three cases of indolent brain tumor (Br.Fe., Cy.Pi. and Ri.Pr.) and onecase ofarteriovenous malformation (Bo.Gr.). Lesions ofone patient in each group (Ra.Ri. and Su.Br.) encroached on the anterior third of the corpus callosum. All patients in the LF group were tested from 6 months to IO years after operation. Five patients in the RF group were tested within 3 months of surgery. and three patients were tested from 4 to 25 years after operation.
M. P. MCANDREWS and B. MILNER
852
Table Sex Group Normal control Left frontal Left temporal Right frontal Right temporal
M
F
8 4 12 5 9
12 4 6 3 9
1. Subjects
Age (years) Mean Range 31.7 36.4 28.0 28.4 28.6
Li.
A.-M
Th.
Ra
20 48 24 47 1643 13 49 16 40
Education Mean 13.1 14.8 12.8 12.1 Il.6
(years) Range l&16 12 18 S-18 7 19 6-16
WAIS-R IQ Mean Range Not assessed 103 92 121 100 76 131 95 80 118 95 80 114
Ja.
MO. Au.,-,_
Fig. I. Diagrams based on the surgeons’ drawings at the time of operation, showing the estimated extent of cortical excision for patients in the left frontal-lobe group. For Figs I and 2, the medial view (above) and inferior view (below) have been included whenever available, together with the lateral view.
THt
Fig. 2. Diagrams
tKONTAL
showing
CORTEX
the estimated
AND
MEMORY
FOR TEMPORAL
extent of cortical frontal-lobe group.
excision
ORDER
for the patients
in the right
Temporul-lohe yroups. The right and left temporal-lobe groups each comprised 18patients. The removals for these patients always included the anterior temporal neocortex and the amygdala, and in some cases invaded the pes hippocampi. There was extensive encroachment (at least 1.5 cm) upon the hippocampus and/or the parahippocampal gyrus in 10 patients in the left temporal-lobe (LT) group and eight in the right temporal-lobe (RT) group. Two patients in the LT group and one patient in the RT group had an indolent brain tumor, and one patient in the latter group had an arteriovenous malformation. In the temporal-lobe groups, 12 patients were tested within 1 month of surgery and 24 patients were tested from 1 to 12 years after operation. Normu/ contrd yroup. The 20 normal control (NC) subjects were either hospital support staff or relatives of patients. and were matched as a group to the patient groups with respect to age and number of years of education ;see Table I ).
Nine series were created from a set of 198 objects (everyday objects and toys). Each series consisted of 20 target objects and two fillers, which were presented at the beginning and end of the series. Within each series, 12 objects were designated as targets for the relative-recency judgements and eight objects were used for the relative-size judgemcnts. Assignment of particular items to these two discrimination tasks was constant across all subjects. Subjects did not know whether a given object was to be probed with a question about size or about temporal order. Each series contained two pairs of recency discrimination items from each of the three presentation conditions (Action Items: Named Items, Landmark; Named Items, No Landmark). Since these nine series contain all three presentation conditions. we refer to them as “mixed” series. Table 2 illustrates presentation and test items for a representative series used in the experiment. The two items making up each test pair were separated by three or four intervening items at initial presentation. Six different ordering schemes were used for the nine presentation series, to ensure, as far as possible, that the three conditions were represented equally often at various positions in the series. Assignment of particular items to the three presentation conditions was counterbalanced across subjects, as was the order of items within each pair. Pairs for the relative-size judgements were created by selecting items within each series that were closest in shape
854
M. P. MCANDKEWS and B. MILNEH Table 2. Materials Presentation
1. 2. 3. 4. 5. 6. 7. 8. 9. IO.
Bath plug Glass Staples Bell Pine cone Tape measure (use it) Chain Roller skate Cake Egg timer (invert it)
Aiotc~: Instructtons
series 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.
Test pairs --recency
Ball Horn Soap Bottle cap Spoon Mirror (look in it) Locomotive Egg carton Pacifier (swing it) cat
Action
Items: tape measure--egg timer pacifier mirror Named Items, Landmark: chain-staples soap-locomotive Named Items, No Landmark: bath plug-pine spoon-ball
cone
Test pairs-size horn-&g carton bottle cavbutton
glass-roller bell-cake
skate
for use in parentheses.
and size to one another. Also, items for pairs were selected so that the size judgement could rarely be based on the relative size of such objects in reality. e.g. by pairing a large button with a small bottle cap, or a plastic flower with a toy dinosaur. Forty-four new’ objects were used to create two additional series. These”homogeneous” series were constructed in the same manner as the nine “mixed” series described above, but they included no action items and no objects wcrc designated for size discrimination. For half of the subjects, these items were switched with others in the “mixed” series, to control for item-specific effects.
Each of the 11 series involved a presentation and a test phase. In the presentation phase, the 20 target Items and two fillers were shown one at a time and then withdrawn behind a screen. For action items. the experimenter gave the instruction to enact at the same time as the object was presented. Subjects named each object and carried out the specified actions, where appropriate. WC attempted to maintain the rate of prcscntation at approximately 4 set per item. but usually the manipulated items were in view for slightly longer (5 X XC). owing to the time taken to perform the specified action. The inter-stimulus interval was approximately 2 sec. Immediately after presentation of the last ttcm in each series. subjects performed an intervening distractor task for approximately 40 set to eliminate any potential rccency effects (cf. [4]). This task involved giving estimates about various quantities (e.g. the weight of the average automobtle) and was unrelated to any ofthe target items. The test phase was next. in which recency and size decisions were interspersed. Items were tested in a pre-determined, random order. For the recency task, subjects wcrc shown two objects and asked to decide which had been presented later in the series. For the sirejudgements, the names of two objects from the series were spoken and subjects had to decide which was the larger (or smaller) of the two, as they had appeared in the series. Approximately 1 2 min ofconversation or rest preceded the presentation of each subsequent series. The order in which the I I series were administered was random across subjects. Subjects were fully informed about the nature of the two memory decistons at the outset of the experiment, and several examples were provided to ensure that they understood the nature ofthe task. The instructions emphasized that they were to name each object and to carry out each actton. when required. as quickly as possible. Subjects were also informed about the distractor task and given a sample question bcforc presentation ofthe lirst series. The entire procedure required approximately I hr.
RESULTS One-way analyses of variance (ANOVAs) revealed no significant differences among the groups in terms of age [F(4,67)= 1.83, P>O.lO] or years of formal education [F (4, 67) = 2.33, P > 0.061. Differences among the patient groups with respect to Full Scale IQ were also non-significant [F (3, 48) = I .08, P>O.35]. The first analysis of the recency-discrimination data concerned the “mixed” series, which included the principal within-subject manipulation. The relevant data are shown in Fig. 3, which displays the proportion of correct recency judgements as a function of subject group
THE FRONTAL
COKTEX
AND
MEMORY
FOR TEMPORAL
ORDER
855
and presentation condition. A 5 x 3 mixed ANOVA, with group as the between-subjects factor and condition as the within-subject factor, was performed on these data. A Helmert contrast was performed on the within-subject factor to permit separate examination of two comparisons: the first evaluated the potential effects ofaction “landmarks” by comparing the two named item conditions, the second compared performance on action vs named items, collapsing over the “landmark” variable. 10 _
09
c
n
Action Items
fl
Named Items, Landmark
q Named
Group ”
Fig. 3. Mean
(+SEM)
Items, No landmark
NC
LF
LT
RF
RT
20
0
18
8
18
proportion correct recency discrimination conditions for each group of subjects.
for
the
three
presentation
For the first contrast, there was no significant effect of condition [F (4, 67) = 0.0341, and a non-significant group x condition interaction [F (4, 67) = 0.1171. These findings indicate that intervening action items did not serve as “landmarks” in that they had no beneficial effects on recency judgements for named items in any subject group. A different pattern was obtained for the major comparison of interest, between the action and named item conditions. Here, the main effect of condition was significant [F (4, 67)= 33.78, P
THE FRONTAL
CORTEX
MARY Montreal
Neurological
0028--3932191 .$3.N+0.00 0 1991 Pergamon Press plc
1991.
AND MEMORY ORDER
PAT MCANDREWS*
Institute
and BRENDA
and the Department of Neurology Montreal, Quebec, Canada
(Receiced
20 July
1990; accepted
FOR TEMPORAL
MILNER
and Neurosurgery,
6 April
McGill
University,
1991)
Abstract-Patients with unilateral frontal- or temporal-lobe excisions and normal control subjects made relative-recency decisions about objects presented sequentially. Several objects within each series were presented in the context of actions to be performed using them, such as “squeeze the sponge”, whereas others had only to be named. Both left and right frontal-lobe groups were impaired on order judgements for named items, but their performance was normal for action items. The results suggest that providing salient and distinctive items, involving meaningful actions and multimodal cues, helps compensate for deficits in memory for temporal information associated with frontal-lobe damage.
INTRODUCTION
MANY CURRENTmodels
of frontal-lobe functioning propose that this region of cortex plays an important role in memory, although there is no general agreement as to the specific mnemonic processes involved [7, 11,24,42]. JACOBSEN [ 151 was the first to demonstrate a memory impairment on delayed-response and delayed-alternation tasks in monkeys with bilateral removals from the frontal cortex. In these tasks, the animal must remember, over a brief interval, which of two identical food wells had been baited (delayed response) or which had been associated with reward on the previous trial (delayed alternation). Jacobsen’s discovery of a profound impairment in animals with frontal-lobe lesions has been replicated so often that normal performance on these tasks is considered a direct reflection of the functional integrity of dorsolateral frontal cortex. Although several studies have shown that the capacity for short-term spatial memory is critical to success on these tasks [lo, 261, others have emphasized the requirement for adequate registration and retention of temporal information [S, 25, 331. That is, the same two events and possible choices occur repeatedly, and the animal must remember which event occurred on the most recent trial in order to respond correctly. This interpretation, which is consistent with Jacobsen’s own explanation of the deficit, has led to a focus on the specific role of the frontal cortex in memory for the temporal order of events. Over the past two decades, research by Milner and her colleagues has shown that patients *Please address all correspondence to: M. P. McAndrews, Research Foundation, 33 Russell Street, Toronto, Ontario, 849
Department of Sociobehavioral Canada M5S 2Sl.
Research,
Addiction
850
M.
P. MCANUKEWS and B. MILNFK
with unilateral frontal-lobe lesions are impaired at monitoring and remembering the temporal order of contextually similar events. Direct evidence of this impairment was first obtained using a relative-recency discrimination task 1211. Subjects were shown a long series of items sequentially (words, representational drawings, or abstract paintings), and were asked periodically to decide which of two test items had appeared more recently in the series. Some test cards contained one item from the series and one new foil, in which case the discrimination was based solely on item recognition. Although patients with frontal-lobe lesions were able to discriminate between old and new test items as well as normal control subjects, their performance was markedly impaired on the recency discriminations. The observed impairments varied with the laterality of the excision and the type of material, in that patients with right frontal-lobe lesions showed marked deficits in recency discrimination for representational drawings and paintings, whereas those with left frontal-lobe excisions were impaired only on the verbal task. Patients with temporal-lobe lesions showed the opposite pattern; normal recency discrimination but poor item recognition in some cases [21,22]. PETRIDESand MILNER [3 I] reported a similar deficit on a task that required patients to select a different item from a repeated stimulus array by keeping track of their responses across successive trials. Similar deficits have subsequently been reported on tasks involving recency discrimination or reconstruction of temporal order in patients with epileptogenic frontal-lobe lesions [17] and in other groups of patients with unilateral or bilateral frontallobe excisions [39]. There have been few attempts to characterize more precisely the processes or mechanisms that could account for deficits observed on tasks requiring memory for temporal order. PRIBRAM and his colleagues [32,33] have argued that deficits on delayed alternation reflect a failure to parse or segment the ongoing stream of experience into discrete “temporal moments”. This notion arose from studies showing that monkeys with dorsolateral frontallobe excisions were unimpaired when the experimenter imposed external “temporal landmarks” by asymmetrically manipulating the duration of the delay period between trials (see 1341 for a criticism of this interpretation). MILNER[21,24] speculated that frontal-lobe damage might compromise encoding or retrieval of “time tags” hypothesized to be laid down as part of the mnemonic record of experienced events 1461. SCHACER 1371 argued for an impairment in automatic encoding of spatiotemporal information, whilst other theorists have emphasized the role of active strategies and reconstruction in memory for temporal order 120, 27, 42, 4.51. The research reported here examined further the nature of this deficit by introducing a new type of material that might benefit the frontal-lobe patients in the recencydiscrimination task. We refer to these new memoranda as action items, although they have been called “subject-performed tasks” in previous research [4]. For each action item, the subject enacts a simple instruction for using an object, such as “break the stick”, or “bounce the ball”. Several considerations led us to believe that action items would be especially powerful memoranda in the recency-discrimination task. Previous research has shown that recall for such items is generally superior to recall of materials presented in other ways, such as lists of words or experimenter-performed actions. In addition, recall of action items, unlike recall of most other types of material, is relatively unaffected by large variations in subject characteristics and presentation variables, such as age, level of general intelligence, or the extent of semantic elaboration of items at study [ 1,4,5, 14,301. Given that such variables are thought to influence strategic or effortful encoding and retrieval processes, COHEN [S] has
THE FRONTAL
CORTEX
AND
MEMORY
FOR TEMPORAL
ORDER
851
suggested that memory for enacted events is largely insensitive to the use of mnemonic strategies. As well, the action items we used involved distinctive actions and multimodal cues, features that should increase the salience of the material. Finally, GENTILE and S~AMM [9] reported improved performance by monkeys with frontal-lobe lesions in the delayedalternation task when different movements were required to displace manipulanda over the food wells, a finding that indicates the contribution of polysensory action cues to performance on this task. In the present experiment, patients with unilateral frontal- or temporal-lobe excisions and normal control subjects made relative-recency discriminations about pairs of objects presented in a study series. Only a few objects were manipulated by subjects, the remainder being merely named. Given our expectation that patients with frontal-lobe excisions would be impaired at recency judgements for named items, the principal hypothesis tested here was that they would perform normally on the action items. It was necessary, however, to examine the potential influence of action items on memory for named items presented in the same series. For example, it was possible that the action items could enhance processing of certain named objects, by serving as potential temporal “landmarks” separating the named items into those presented “before” and “after” a given manipulation. To evaluate this possibility, discrimination tests were constructed such that some pairs of named items were separated by an action “landmark” in the presentation sequence, whereas other pairs consisted of items that had not been separated by an intervening action. A second concern was that the enacted items might interfere with the processing of adjacent named items, as has been shown with other unique or high-priority events embedded in a series of neutral homogeneous events 1381. To control for this possibility, subjects were also tested on additional series of objects (i.e. “homogenous” series) in which there were no action items. Recent findings 1401 indicate that patients with frontal-lobe lesions may have difficulty with other kinds of memory judgements, such as those concerning frequency of occurrence. Accordingly, we included another type of memory decision that required subjects to remember and compare the sizes of the objects. Although size can be considered an intrinsic property of objects, rather than a contextual attribute, it is conceivable that the operations involved in making comparative judgements about remembered events along this dimension would also be impaired in patients with frontal-lobe lesions.
METHOD The participants in this study comprised 52 patients of the Montreal Neurological Hospital, each of whom had undergone a unilateral brain operation for the relief of pharmacologically intractable epilepsy, and 20 normal control subjects. Except where stated. the epileptogenic lesions were static and atrophic. Patients with a Full-Scale IQ rating below 75 on the Wechsler Adult Intelligence Scale-Revised were excluded from the study, as were those with cvidcnce of diffuse cerebral damage or bilateral independent electrographic abnormalities. Table I shows the sex distribution, mean age, Full-Scale IQ rating and educational level of subjects in the various groups. Frontcll-lohe groups. Figures I and 2 illustrate the excisions of patients in the left and right frontal-lobe groups. respectively. These diagrams are based on the surgeons’ drawings at the time of operation. In all patients but one (Ho.Sp.). the excision included part of the dorsolateral frontal cortex, and in some patients the excision extended Into the medial frontal region. Broca’s area was spared in all patients with left frontal-lobe (LF) excisions. In the LF group. there were three cases of indolent brain tumor (Li.Ar.. Lo.La. and HOSP.) and one case of arteriovenous malformation(A.-M.Be.). The right frontat-lobe (RF) group also included three cases of indolent brain tumor (Br.Fe., Cy.Pi. and Ri.Pr.) and onecase ofarteriovenous malformation (Bo.Gr.). Lesions ofone patient in each group (Ra.Ri. and Su.Br.) encroached on the anterior third of the corpus callosum. All patients in the LF group were tested from 6 months to IO years after operation. Five patients in the RF group were tested within 3 months of surgery. and three patients were tested from 4 to 25 years after operation.
M. P. MCANDREWS and B. MILNER
852
Table Sex Group Normal control Left frontal Left temporal Right frontal Right temporal
M
F
8 4 12 5 9
12 4 6 3 9
1. Subjects
Age (years) Mean Range 31.7 36.4 28.0 28.4 28.6
Li.
A.-M
Th.
Ra
20 48 24 47 1643 13 49 16 40
Education Mean 13.1 14.8 12.8 12.1 Il.6
(years) Range l&16 12 18 S-18 7 19 6-16
WAIS-R IQ Mean Range Not assessed 103 92 121 100 76 131 95 80 118 95 80 114
Ja.
MO. Au.,-,_
Fig. I. Diagrams based on the surgeons’ drawings at the time of operation, showing the estimated extent of cortical excision for patients in the left frontal-lobe group. For Figs I and 2, the medial view (above) and inferior view (below) have been included whenever available, together with the lateral view.
THt
Fig. 2. Diagrams
tKONTAL
showing
CORTEX
the estimated
AND
MEMORY
FOR TEMPORAL
extent of cortical frontal-lobe group.
excision
ORDER
for the patients
in the right
Temporul-lohe yroups. The right and left temporal-lobe groups each comprised 18patients. The removals for these patients always included the anterior temporal neocortex and the amygdala, and in some cases invaded the pes hippocampi. There was extensive encroachment (at least 1.5 cm) upon the hippocampus and/or the parahippocampal gyrus in 10 patients in the left temporal-lobe (LT) group and eight in the right temporal-lobe (RT) group. Two patients in the LT group and one patient in the RT group had an indolent brain tumor, and one patient in the latter group had an arteriovenous malformation. In the temporal-lobe groups, 12 patients were tested within 1 month of surgery and 24 patients were tested from 1 to 12 years after operation. Normu/ contrd yroup. The 20 normal control (NC) subjects were either hospital support staff or relatives of patients. and were matched as a group to the patient groups with respect to age and number of years of education ;see Table I ).
Nine series were created from a set of 198 objects (everyday objects and toys). Each series consisted of 20 target objects and two fillers, which were presented at the beginning and end of the series. Within each series, 12 objects were designated as targets for the relative-recency judgements and eight objects were used for the relative-size judgemcnts. Assignment of particular items to these two discrimination tasks was constant across all subjects. Subjects did not know whether a given object was to be probed with a question about size or about temporal order. Each series contained two pairs of recency discrimination items from each of the three presentation conditions (Action Items: Named Items, Landmark; Named Items, No Landmark). Since these nine series contain all three presentation conditions. we refer to them as “mixed” series. Table 2 illustrates presentation and test items for a representative series used in the experiment. The two items making up each test pair were separated by three or four intervening items at initial presentation. Six different ordering schemes were used for the nine presentation series, to ensure, as far as possible, that the three conditions were represented equally often at various positions in the series. Assignment of particular items to the three presentation conditions was counterbalanced across subjects, as was the order of items within each pair. Pairs for the relative-size judgements were created by selecting items within each series that were closest in shape
854
M. P. MCANDKEWS and B. MILNEH Table 2. Materials Presentation
1. 2. 3. 4. 5. 6. 7. 8. 9. IO.
Bath plug Glass Staples Bell Pine cone Tape measure (use it) Chain Roller skate Cake Egg timer (invert it)
Aiotc~: Instructtons
series 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.
Test pairs --recency
Ball Horn Soap Bottle cap Spoon Mirror (look in it) Locomotive Egg carton Pacifier (swing it) cat
Action
Items: tape measure--egg timer pacifier mirror Named Items, Landmark: chain-staples soap-locomotive Named Items, No Landmark: bath plug-pine spoon-ball
cone
Test pairs-size horn-&g carton bottle cavbutton
glass-roller bell-cake
skate
for use in parentheses.
and size to one another. Also, items for pairs were selected so that the size judgement could rarely be based on the relative size of such objects in reality. e.g. by pairing a large button with a small bottle cap, or a plastic flower with a toy dinosaur. Forty-four new’ objects were used to create two additional series. These”homogeneous” series were constructed in the same manner as the nine “mixed” series described above, but they included no action items and no objects wcrc designated for size discrimination. For half of the subjects, these items were switched with others in the “mixed” series, to control for item-specific effects.
Each of the 11 series involved a presentation and a test phase. In the presentation phase, the 20 target Items and two fillers were shown one at a time and then withdrawn behind a screen. For action items. the experimenter gave the instruction to enact at the same time as the object was presented. Subjects named each object and carried out the specified actions, where appropriate. WC attempted to maintain the rate of prcscntation at approximately 4 set per item. but usually the manipulated items were in view for slightly longer (5 X XC). owing to the time taken to perform the specified action. The inter-stimulus interval was approximately 2 sec. Immediately after presentation of the last ttcm in each series. subjects performed an intervening distractor task for approximately 40 set to eliminate any potential rccency effects (cf. [4]). This task involved giving estimates about various quantities (e.g. the weight of the average automobtle) and was unrelated to any ofthe target items. The test phase was next. in which recency and size decisions were interspersed. Items were tested in a pre-determined, random order. For the recency task, subjects wcrc shown two objects and asked to decide which had been presented later in the series. For the sirejudgements, the names of two objects from the series were spoken and subjects had to decide which was the larger (or smaller) of the two, as they had appeared in the series. Approximately 1 2 min ofconversation or rest preceded the presentation of each subsequent series. The order in which the I I series were administered was random across subjects. Subjects were fully informed about the nature of the two memory decistons at the outset of the experiment, and several examples were provided to ensure that they understood the nature ofthe task. The instructions emphasized that they were to name each object and to carry out each actton. when required. as quickly as possible. Subjects were also informed about the distractor task and given a sample question bcforc presentation ofthe lirst series. The entire procedure required approximately I hr.
RESULTS One-way analyses of variance (ANOVAs) revealed no significant differences among the groups in terms of age [F(4,67)= 1.83, P>O.lO] or years of formal education [F (4, 67) = 2.33, P > 0.061. Differences among the patient groups with respect to Full Scale IQ were also non-significant [F (3, 48) = I .08, P>O.35]. The first analysis of the recency-discrimination data concerned the “mixed” series, which included the principal within-subject manipulation. The relevant data are shown in Fig. 3, which displays the proportion of correct recency judgements as a function of subject group
THE FRONTAL
COKTEX
AND
MEMORY
FOR TEMPORAL
ORDER
855
and presentation condition. A 5 x 3 mixed ANOVA, with group as the between-subjects factor and condition as the within-subject factor, was performed on these data. A Helmert contrast was performed on the within-subject factor to permit separate examination of two comparisons: the first evaluated the potential effects ofaction “landmarks” by comparing the two named item conditions, the second compared performance on action vs named items, collapsing over the “landmark” variable. 10 _
09
c
n
Action Items
fl
Named Items, Landmark
q Named
Group ”
Fig. 3. Mean
(+SEM)
Items, No landmark
NC
LF
LT
RF
RT
20
0
18
8
18
proportion correct recency discrimination conditions for each group of subjects.
for
the
three
presentation
For the first contrast, there was no significant effect of condition [F (4, 67) = 0.0341, and a non-significant group x condition interaction [F (4, 67) = 0.1171. These findings indicate that intervening action items did not serve as “landmarks” in that they had no beneficial effects on recency judgements for named items in any subject group. A different pattern was obtained for the major comparison of interest, between the action and named item conditions. Here, the main effect of condition was significant [F (4, 67)= 33.78, P
