Neuropsyehologia, 1965, Vol. 3, pp. 317 to 338. Pergamon Press Ltd. Printed in England
VISUALLY-GUIDED MAZE LEARNING IN MAN:
EFFECTS OF BILATERAL HIPPOCAMPAL, BILATERAL
FRONTAL, AND UNILATERAL CEREBRAL LESIONS
BRENDA MILNER
Montreal Neurological Institute, McGill University, Montreal, Canada
(Received4 August 1965)
Abstract--Seventy-nine patients with different cerebral lesions were trained on a visuallyguided
stylus maze. The main findingswere: (1) bilateral hippocampal lesionsproduced the
most severeimpairment, with one patient showingno progressin 215 trials; (2)right temporal
lobectomy produced a significantdeficit,particularly whenthe underlyinghippocampus was
radically excised, whereas patients with left temporal-lobe lesions of equal extent, or with
small parietal-lobe lesions of either hemisphere, obtained normal scores; (3) larger lesionsof
frontal or right posterior cortex severelyimpaired maze learning, suggestingthat sizeas well
as locus of cortical lesion may be a factor in maze-performance; (4) only frontal-lobe lesions
affectedthe ability to follow test instructions.
1. INTRODUCTION
THIS study arose from the need to investigate more systematically the learning ability of
patients with bilateral hippocampal lesions, preferably on a task which was not primarily
verbal and which permitted the use of massed trials. On clinical examination, patients
with these lesions show a severe and generalized memory disorder, those with the most
extensive lesions being unable to recall or recognise test material after a lapse of five minutes
or less, if their attention has been diverted elsewhere in the meantime, and showing a
continuous anterograde amnesia for the events of daily life (ScovILLE and MILER [I];
PENFIELDand MILNER[2]; MILNER[3]). On formal testing with verbal material, they fail to
progress over many trials. Maze-learning seemed a convenient and objective method of
study, the results of which would complement our other findings.
The maze-learning performance of the patients with hippocampal lesions was eompared,
not only with that of normal subjects, but also with that of a long series of patients who had
undergone cortical excisions of varying locus and extent, for the relief of focal seizures.
The excisions sampled all parts of the fight cerebral cortex and the left frontal and anterior
temporal and parietal regions of the dominant hemisphere, but spared Broca's area and the
posterior temporal and parietal speech zones. This restriction on elective surgery means
that we do not have, and cannot expect to have, any lesions comparable to those of the
left posterior-parietal and parieto-temporal gunshot-wound eases in Prof. Teuber's
population. This study does, however, include extensive parietal-lobe lesions in the left
hemisphere, when that hemisphere is not dominant for speech.
The task chosen was a stylus-maze problem of the stepping-stone kind (BARKER,[4]),
the subjects being trained over many trials until they could traverse the maze without error.
The resulting data confirmed the profound learning deficit of the hippocampal patients but
317
318 BRENDAMILNER
also brought to light important differences in maze-learning proficiency among the different
cortical-lesion groups, depending upon the locus, laterality and possibly the size of the
lesion. These findings form the substance of the present paper.
Route-finding difficulties have traditionally been associated with parietal-lobe lesions
and SEMMES,WEINSTEIN, G~ENT and TEUBER[5] have demonstrated residual impairment
on spatial orientation tasks in men who had sustained penetrating missile wounds of the
parietal lobes many years before, men with wounds sparing this region being unimpaired.
The tasks required the subject to walk along specific routes on the basis of a series of maps,
which could be either visually or tactually presented. Left and right parietal-lobe wounds
were equally damaging to performance on these locomotor tasks. However, on a penciland-paper
lattic-maze task, ELtrnORN [6, 7] and BENTON,ELIT~ORN, FOaEL and KERR [8]
found that patients with right-hemisphere lesions were more impaired than patients with
left, a finding which is consistent with many clinical observations concerning topographical
orientation in acute cerebral disease (e.g. PATERSONand ZAN~WILL[9]; MCFIE, PIERCVand
ZANGWILL[10]; HI~CAEN,PENFIELD,BERTRANDand MALMO[11]; ETTLINGER,WARRINGTON
and ZANGWILL[12]; H~CAEN[13]; WHITTYand NEWCOMBE[14]).
HI~CAEN[13] discusses post-Rolandic lesions only, and SEMMESet al. [5] find no evidence
of anterior brain wounds impairing orientation in extrapersonal space, although orientation
with respect to the subject's own body may be disturbed (SEMMES,WEINSTEIN,GHENTand
TEUBER[15]). ELITHORN[7], however, reports equal impairment with frontal and posterior
cerebral lesions, but finds some maze items differentially sensitive to frontal, others to
posterior lesions, suggesting a different source of difficulty for the two groups. The Porteus
mazes resemble ELITHORN'S[7] perceptual mazes in requiring implicit trial-and-error for their
solution; it is therefore of interest that a transient defect in Porteus-maze performance
follows frontal lobotomy (LANDISand ERLICK [16]; PORTEUS[17] pp. 52-82). There may,
however, be many factors other than spatial ones which could account for this deficit.
The maze task in the present study poses a learning problem primarily, since it requires
the repeated tracing of a constant path until the most direct route from starting-point to
goal has been mastered. In this it resembles the maze-learning problems used in animal
work rather than the more perceptual tasks devised for the study of human patients with
cerebral lesions. For this reason, performance may be less affected by minor deficits of
spatial orientation than by difficulty in remembering the correct sequence of turns from one
trial to the next, although these two facets of maze performance cannot be isolated too
strictly from one another, and patients with marked spatial disorientation would be expected
to find any spatial maze difficult to learn. These issues will be further discussed in the context
of the data to be presented here.
2. SUBJECTS
The main group consisted of 74 patients with unilateral cortical excisions carried out for
the relief of focal epilepsy due to static, atrophic lesions. Cases of brain tumour were
excluded. A further five patients, two with bilateral frontal and three with bilateral hippocampal
lesions, complete the patient population. Table 1 gives the sex, age and Wechsler
intelligence ratings for the 79 patients, subdivided according to locus of lesion.
The 11 normal control subjects (3 male, 8 female) had had a high school education and
were probably a little above average in intelligence; they ranged in age from 18 to 45 years,
with a mean of 22.2 years.
VISUALLY-GUIDEDMAZELEARNINGINMAN 319
Table 1. Sex, age, and intelligence data for patients grouped according to locus of cerebral lesion
Group
No. Sex
of
cases M F
Age Wechsler I.Q.
Mean Range Mean Range
Unilateral
Left temporal 26 20 6 26.2 13-54 102.2 79-128
Right temporal 15 8 7 25.0 13-51 107.9 94-124
Parietal 12 9 3 22.2 16--31 98.4 71-114
Right parietotemporo-occipital
6 4 2 27.2 19-38 102.8 96--118
Frontal 15 11 4 22.3 13-33 94.3 74--124
Bilateral
Frontal 2 2 0 47.0 42, 50 111.2 126, 97
Hippocampal 3 3 0 43.0 34-55 117.7 109-123
2.1 Cases of Unilateral Lesion
Table 2 indicates the probable etiology of the original lesions in patients classified on
the basis of subsequent cerebral excision as right temporal, left temporal, parietal, right
parieto-ternporo-occipital, or frontal-lobe cases. These epileptogenic lesions typically date
from early life, with onset of habi.tual seizures in late childhood or early adolescence, and
Table 2. Probable etiology of epileptogenic lesions in unilateral groups
Brain injury Febrile Brain A-V Pial
illness abscess malf'n angloBirth
or 1-7 8-15 21 matosis
Group 1st yr yr yr yr 0-9 yr
Right temporal 11 1 - 2 - 1 Left
temporal 14 3 1 - 4 - 1 3
Parietal 8 - 1 2 - 1 Right
parietotemporo-occipital
2 1 - 1 - 1 - 1
Frontal 8 1 4 - - 1 1 Total
43 6 6 I 8 2 4 4
the distribution of lesions is similar in the various lesion groups, except for a rather high
incidence of late-childhood trauma in the frontal-lobe group. Table 3 shows the mean
number of years of habitual preoperative seizures for patients classified in the same way.
There are no salient group differences, the left and right temporal-lobe groups being
particularly well-matched.
2.1.1. Cortical excisions
2.1.1.1. Temporal-lobe groups. The 41 temporal lobectomies were relatively uniform,
usuallyinvolvingtheanterior5 to 6 cm of the temporal lobe and including at least the uncus
and amygdala on the mesial surface. In 10 left- and 7 right-sided cases, a radical excision
of the hippocampus was carried out. In the left temporal group, all 26 removals were in the
dominant hemisphere for speech and all 15 right temporal lobectomies were in the minor
hemisphere, five additional temporal-lobe patients having been omitted because they were
found to have bilateral speech representation. Only patients with well-lateralized epileptogenic
foci were included.
320 BRENDAMILNER
Table 3. Mean duration of preoperative seizures (yr) for different unilaterallesion groups
Preoperative seizures
No. (yr)
of
Group cases Mean S.D.
Right temporal 15 11.5 8.68
Left temporal 26 11.2 8.61
Parietal 12 9.4 6.73
Right parietotemporo-occipital
6 11.3 4.54
Frontal 15 9.5 6.76
TRIALS 9 TRIALS 74
ERRORS 61 ERRORS 528
TRIALS 33 TRIALS 46
ERRORS 116 ERRORS 307
TRIALS 21 TRIALS 20
ERRORS 69 ERRORS 100
~o. 1. Diagrams based on the surgeon's drawings at the time of operation, showing the
estimated extent of temporal-lobe ablation (lateral and inferior views) in 6 representative
cases. Case J. B., left temporallobectomy, hippocampus excised; Case J. C., small left temporal
lobectomy, sparing hippocampus; Case P. W., large left temporal lobectomy, sparing
hippocampus; Case P. A., right temporal lobectomy, hippocampus excised; Case E.D., right
temporal lobectomy, invading anterior hippocampus only; Case L. B., large right temporal
lobectomy, sparing hippocampus. In this figure (and in Figs. 2-4), the maze scores for each
patient are indicated beside the appropriate brain map.
VISUALLY-GUIDED MAZE LEARNING IN MAN 321
Figure 1 shows some representative temporal lobectomies. The removals on the left
ranged from 4 to 9 cm along the Sylvian fissure, with a mean of 5.2 cm, and from 4-10.5 cm
along the base of the brain, with a mean of 6.1 cm. The corresponding figures for the right
temporal lobe are 4-7 cm along the Sylvian fissure, with a mean of 5.7 cm, and 5-7.5 cm
along the base, with a mean of 6.2 cm. Thus, some lesions on the left were actually larger
than any lesion on the right and there are no significant differences between the two temporallobe
groups with respect to lesion size.
Seven left temporal-lobe patients were tested in follow-up study, from one to five
years postoperatively, at which time only one was still having seizures; the other 34 patients
were tested from two to three weeks after operation.
2.1.1.2. Parietal-lobe group. The parietal-lobe removals were quite variable in locus and
extent. The group comprised 12 patients, 9 with left-sided lesions and 3 with right, 3 of the
left parietal excisions being in the non-dominant hemisphere for speech.
TRIALS
ERRORS
18 TRIALS 12
101 ERRORS 73
TRIALS 23 TRIALS 20
ERRORS 83 ERRORS 93
TRIALS 11 TRIALS 59
ERRORS 64 ERRORS 237
• Case M.R.
FIG. 2. Diagrams showing estimated extent of cortical excision in 6 of the 12 parietal-lobe
cases. Case G. G., left central; Case H. Ja., right Sylvian; Cases P. Mo., S. La., left and right
parasagittal, respectively0ateral and mesial views); Cases W. St., M. R., left, non-dominant,
parietal lobectomies.
322 BRENDA MILNER
Figure 2 illustrates the range of lesions sampled. In all instances the parietal lobe was
primarily involved, although occasionally, as in Cases H. Ja. and M. R., the lesion encroached
upon the neighbouring superior temporal or inferior frontal cortex. In the dominant left
parietal lobe there were three lower central excisions (cf. Case G. G.) and three parasagittal
ones (cf. Case P. Mo.). In the right parietal lobe there were two parasagittal excisions
(cf. Case S. La.) and one, Case H. Ja., in the posterior Sylvian region. The three nondominant
left parietal lesions were more extensive (cf. Cases W. St. and M. R.).
Nine patients (including the six whose removals are shown in Fig. 2) were tested from
one to seven years after operation, at which time they were seizure-free; the other three
patients were tested from two to three weeks post-operatively.
2.1.1.3. Right parieto-temporo-occipital group. The removals in these six patients are
shown in Fig. 3. All were large lesions, involving parts of the parietal, temporal and occipital
cortex and, in Case D. L., invading also the orbito-frontal region. Case J. St. M. is the
one instance of adult brain injury in the unilateral lesion groups. This 38-year old man had
developed seizures as a sequel to a penetrating missile wound sustained in World War I1,
at the age of 21.
r___ ~, r... PaIA
TRIALS 29 TRIALS 31
ERRORS 128 ERRORS 165
Case J.St, M. Case A.W.
TRIALS 44 TRIALS 38
ERRORS 395 ERRORS 178
TRIALS 65 TRIALS 68
ERRORS 265 ERRORS 538
FIG.3. Diagrams showingestimated extent of ablation (lateral and inferior) in the 6 cases of
right parieto-temporo-occipitalexcision.
VISUALLY-GUIDEDMAZELEARNINGINMAN 323
c . ~ TRIALS 4 TRIALS 15
ERRORS 20 ERRORS 49
C o ~ TRIALS 12
ERRORS 58 TRIALS 45
ERRORS 270
C a ~ TRIALS 15
ERRORS 75
TRIALS 59
ERRORS 288
TRIALS 10
ERRORS 82
TRIALS 35
ERRORS 226
TRIALS 44
ERRORS 304
TRIALS 49
ERRORS 308
TRIALS 58
ERRORS 458
TRIALS 66 TRIALS 78
ERRORS434 ERRORS 542
TRIALS 17 TRIALS 45
ERRORS 117 ERRORS 440
FIG. 4. Diagrams showing lateral extent of excision in the 15 cases of unilateral frontal
lobectomy. Cases R. B. to D. K., dominant left frontal; CasesJ. Ed. to M. S.,right frontal;
Case L. C., left frontal, possiblynon-dominant; Case Y. P., left frontal, non-dominant.
324 BRENDA MILNER
Cases N. Le., J. St. M. and N. E. were tested in the early post-operative period, the
other three patients in follow-up study from 4 to 13 years later, at which time Cases D. L.
and D. Jo. were seizure-free and Case A. W. was having only occasional minor attacks.
2.1.1.4. Frontal-lobe group. The lateral extent of removal in each of the 15 cases is
shown in Fig. 4. Seven left frontal lobectomies were in the dominant hemisphere for
speech, two in the non-dominant; all six right frontal lobectomies were in the minor hemi-
sphere.
Except for Case L. A., the right frontal removals were large, typically extending back to
within 1 cm of the precentral gyrus on the lateral surface and including parts of the anterior
cingulate and subcallosal gyri on the mesial aspect of the hemisphere. The dominant left
frontal lobectomies spared Broca's area (as mapped out by speech testing at operation), but,
except for Cases R. B. and J. Mo. with small lesions, the mesial extent of removal was the
same as in the right frontal group. All lesions involved at least the dorsolateral aspect of the
frontal lobe.
Seven patients (Cases R. B., D. K., J. Mo., M. S., L. C. and Y. P.) were tested from
1 to 14 years after operation, the remaing 8 patients in the early postoperative period.
Case D. K. was still having some seizures, but the other follow-up patients were essentially
seizure-free.
2.2. Cases of"Bilateral Lesion
2.2.1. Bilateralfrontal-lobe lesions. Two patients were tested. The first, Case K. M.,
had developed seizures after a severe brain injury at the age of 16. Ten years later, in 1938,
a bilateral excision of the anterior one-third of both frontal lobes was carried out (HEBB
and PENFmLD[18]). Since then he has been seizure-free, except for one attack when he
was in the Canadian army and all anti-convulsant medication had been withdrawn. The
second patient, Case A. N., sustained a penetrating shell wound of both frontal lobes in
1945, at the age of 22, and necrotic tissue amounting to roughly one-third of the right and
one-fifth of the left frontal lobe was excised. This patient belongs to the brain-injured
group studied so extensively by Prof. Teuber and his associates. He now has occasional
seizures.
2.2.2. Bilateral hippocampal lesions. The three patients in this group all had generalized
postoperative memory impairment, which was most marked in Case H. M., who in
1955, at the age of 27, had undergone a bilateral medial temporal-lobe resection
(ScoVILLE and MILNER [1]). The removal was said to extend back 8 cm along the mesial
surface of both temporal lobes, destroying the uncus, amygdala and anterior two-thirds of
the hippocampus and hippocampal gyrus bilaterally, but sparing the temporal neocortex.
This radical operation was carried out by Dr. William Scoville in an attempt to relieve
intractable seizures, and these have in fact been considerably reduced.
Cases F. C. and P. B. (PENFIELDand MILNER[2]) had undergone unilateral (left) temporal
lobectomy for the relief of long-standing seizures, but they differed from our regular series
of temporal-lobe cases in having active epileptogenic loci in the opposite,-unoperated
temporal lobe. PENFIELDand MILNER [2] attributed the postoperative memory disorder
to a pre-existing lesion in the right hippocampal region, so that the left hippocampal excision
effectively produced a bilateral lesion. Support for this view comes from the preliminary
pathological findings on Case P. B. who died on April 24, 1965, of a massive pulmonary
VISUALLY-GUIDED MAZE LEARNING IN MAN 325
embolism. On section of the brain, Dr. Gordon Mathieson found that the right hippocampus
was shrunken and pale and had lost its normal structure. On the operated (left) side, approximately
22 mm of the posterior part of the hippocampus remained; it was slightly reduced in
size in the coronal plane but its normal pattern was preserved. Histological study remains
to be carried out.
3. APPARATUS AND PROCEDURE
• • • • • • • •--0--0
I
• • • • • 0--0--0 • •
I
• • • • O • O • • O
I
• • • O O • O • O O
I
• • • O O O O • O O
I
• • • • • 0--0-0--0 •
I
• • O O O O O O • O
I
0--0--0-0 • • 0--0--0 •
I I I
• • • • • O • • O O
I I I
• • • 0--0--0--0 • • •
sLc,.rL
Flo. 5. Plan of maze. The black line indicates the correct path.
The stylus maze used is shown diagrammatically in Fig. 5. The dots represent boltheads
spaced one inch apart to make up a 10 by 10 array. They are mounted on a 13 in. square
wooden board, painted black, thus constituting visible "stepping-stones". The subject's
task is to discover and remember the one correct path leading from the lower left-hand corner
to the upper right-hand corner of the array. In the diagram, this path is indicated by a
black line, but for the subject of course the blind alleys and the correct path are visually
indistinguishable. With a metal-pointed stylus in his preferred hand, he proceeds, one step
at a time, from bolthead to bolthead, with the loud click of an electrical error-counter
informing him whenever he leaves the correct path. The rules are few: he must go back to
the preceding bolthead whenever the counter clicks; he must not retrace portions of the
correct path; and he must not move diagonally across the board. At the end of each trial he
is told how many errors he has made and he knows that he must continue until the criterion
of three successive errorless runs has been attained. Time scores are unobtrusively recorded.
Training was carried out in blocks of 25 trials, twice daily, until the criterion was reached.
The only exception to this rule was in the case of H. M., the patient with a bilateral hippocampal
excision, where the limited time available and the need to get as much information as
possible necessitated extending the testing to 75 trials a day.
326 BRENDA MILNER
4. RESULTS
4.1. Trials to criterion and error-scores: unilateral lesion groups
No relationship was found between maze performance and either etiology of lesion,
duration of habitual seizures, or time since operation. For 49 patients and 11 normal
control subjects who were tested both on this visual maze and, later, on a tactual maze
(CORKIN [19]), the Pearson product-moment correlation between age and trials to criterion
on the visual maze was --0.07 and between age and errors it was +0.02. For the same
49 patients, the correlation between both trials and errors on the visual maze and I.Q. was
-0.20.
In contrast to these negative findings, locus of lesion proved to be a highly significant
factor in maze performance. Table 4 shows the mean trials and errors to criterion for the
various unilateral lesion groups. The scores for the parietal and left temporal-lobe groups
approximate those of the normal control group, whereas the performance of the right
temporal, right parieto-temporo-occipital, and frontal-lobe groups appears to be markedly
impaired.
Table 4. Mazelearningafter differentunilateral cortical excisions
No. Trials Errors
Group of
cases Mean Range Mean Range
Left temporal 26 22.4 7-57 103.0 48-248
Right temporal 15 33.8 5-89 224.7 23-605
Parietal (6L, 3R,
3LND) 12 22.0 11-59 97.3 64-237
Right parietotemporo-occipital
6 45.8 29-68 278.2 128-538
Frontal (7L. 6R,
2LND) 15 36.8 4-78 244.7 20-542
Normal control 11 17.0 6-27 91.8 21-175
Simple analysis of variance confirmed that there were significant group differences in
trials to criterion (F=4.46, P<0.01). Analysis of variance was also carried out on the
error scores after these had been subjected to the square-root transformation, y=v/x,
where x is the original score and y the score analysed, thus reducing the skewness of the
distributions (BARTLETT[20]). The resulting F-score of 6.43 is significant beyond the 0.001
level. Subsequent t-tests showed that the parietal and left temporal-lobe groups did not
differ from the normal control group with respect to either trials or errors. The other
three lesion groups were however significantly impaired relative to the normal
subjects on both these measures (Right temporal group: Trials, t =2.55, P<0.02; errors,
t =2.71, P<0.02. Right parieto-temporo-occipital group: Trials, t=3.42, P<0.01; errors,
t=3.39, P<0.01. Frontal group; Trials, t=3.02, P<0.01; errors, t=2.95, P<0.01).
Although from inspection of the mean scores in Table 4 the right parieto-temporo-occipital
group appears to be more impaired than either the right temporal or frontal-lobe groups,
there are in fact no significant differences in score between these three groups, presumably
because of the small number of subjects with right parieto-temporo-occipital lesions.
VISUALLY-GUIDED MAZE LEARNING IN MAN 327
4.1.1. Comparison of left and right temporal-lobe groups. Since the right and left temporal
lobe groups were so well-matched with respect to lesion size and clinical history, any
consistent differences in task performance between these groups are of particular interest.
On this maze-learning test, the right temporal group required significantly more trials
(t=2.09, P<0.05) and made significantly more errors (t=3.19, P<0.01) than the left
temporal-lobe group; nevertheless, there was considerable overlap in the distribution of
scores tbr the two groups, with five right temporal-lobe patients obtaining scores near to the
mean of the normal control group. The individual differences in maze performance in the
right temporal group were unrelated to extent of removal of lateral temporal cortex but
were related to how much of the underlying hippocampus had been excised. When the
right temporal group was subdivided into those patients with radical excision of the hippocampus
and those in whom the hippocampus had been spared or minimally destroyed, the
patients with radical excision were found to take significantly more trials to learn the maze
(Mann-Whitney U = 13, P < 0.05 for a two-tailed test) than did those with the hippocampus
essentially spared. For the error scores, the difference between the right hippocampusspared
and right hippocampus-removed groups just failed to reach significance (U=I4,
P=0.064, for a two-tailed test). There was no significant difference in score between the
left temporal-lobe patients with the hippocampus spared and those with hippocampus
radically excised. For example, in Fig. 1, both Case J. B. with radical excision of the
left hippocampus and Case P. W. with hippocampus spared but with an extensive removal
of the left temporal neocortex had scores better than the average of the normal control
subjects. Case P. A., on the other hand, with a radical excision of the right hippocampus,
made 528 errors and required 78 trials to learn the maze, whereas Case L. B., with a slightly
larger removal on the lateral surface but with the right hippocampus spared, obtained a
normal score.
200
~150
100
3O
E
20
D.:-
I--
10
m• Left temporal
] Right temporal
40
300 --
250
SO
0
Hippoc. Hippoc. Hippoc. Hippoc.
Spared Removed Spored Removed
FIG.6. Maze learning after temporal lobectomy: detrimental effectsof hippocampal excision
in the right temporal group. Left temporal: hippocampus spared, N=16; removed, N= 10.
Right temporal: hippocampus spared, N=8, removed, N=7.
328 BRENDA MILNER
These findings are summarized in Fig. 6, which shows the detrimental effect of right
hippocampectomy on maze performance. The scores for the right temporal group with
hippocampus excised are significantly worse than those of the corresponding left temporal
group (trials : U = 5, P < 0.002; errors: U = 8, P < 0.02, for two-tailed tests), whereas there is
no significant difference between the right and left temporal groups with hippocampus
essentially spared. Thus, whether or not a major portion of the underlying hippocampus
is included in the removal is a critical factor in maze learning after right temporal lobectomy.
4.2. Trials to criterion and error scores: bilateral lesion groups
Table 5 lists the individual learning scores of the five patients with bilateral lesions.
Of the two frontal-lobe patients, Case K. M. is clearly impaired, with scores outside the
normal range; his performance is, however, well within the range of the patients with
radical unilateral frontal lobectomies. Case A. N. has normal scores, although, as will be
seen below, other aspects of his maze perfol~ance were impaired. The three patients with
bilateral hippocampal lesions show a profound learning disability, which correlates with the
relative severity of their memory disorder to clinical examination. Case H. M. failed to
learn the correct path through the maze within the limits of testing; Cases F. C. and P. B.
did eventually reach criterion, but the number of trials required falls outside the range for the
other subjects tested.
Table 5. Maze-learning scores for 5 subjects with bilateral lesions
No. of
Group cases Trials Errors
Bilateral frontal 2
Case K.M. 73 381
Case A.N. 18 91
Bilateral hippocampal 3
Case H.M. 215+ 2,877+
Case F.C. 183 828
Case P.B. 102 377
Case H. M. not only failed to learn the maze; he failed to show any improvement within
each testing session of 25 massed trials, so that he was making as many errors per trial at
the end of three days of testing as he had done at the outset.
These findings are illustrated in Fig. 7 which shows the error curve for Case H. M. plotted
over 215 trials.
4.2.1. Time scores: unilateral lesion groups. For most patients, the time scores added
little new information and they were therefore not subjected to any detailed statistical
treatment. All subjects reduced their time per trial considerably with repeated practice on
the maze, but there was a high correlation between initial and final scores, with some subjects
consistently preferring to work slowly and others to work fast. Table 6 shows the mean
time scores for the first and last five trials for the different unilateral lesion groups.
It should be pointed out that the high initial time-scores of the frontal-lobe reflects the
inability of these patients to obey the rules in the early stages of training, a point which will
be discussed more fully below.
Iv{C~I~ ~ r ~ o ~ 5
per .5 tricta
18
16
14
lO
8
6
4
2
0
VISUALLY-GUIDED MAZE LEARNING IN MAN
5 IO 15 1 S 10 1 5 IO 15 ~O
Tric~ bl~och 5
Fzo. 7. Learning curve for Case H. M. (bilateral hippocampal excision), showing no decrease
in errors over a 3-day period. Each point on the curve represents 5 sucessive trials.
Table. 6. Mean time scores (sec) for different unilateral lesion groups
Group
First 5 trials Last 5 trials
Mean Range Mean Range
Left temporal 70 36-154 25 13-70
Right temporal 88 52-148 31 14-70
Parietal 74 42-131 27 15-56
Right parietotemporo-occipital
88 51-138 25 17-31
Frontal 108 54-187 25 15-72
Normal control 66 35-170 20 15-56
329
4.3. Time scores: bilateral lesion group
The time scores at the beginning and end of training for the five patients with bilateral
lesions are shown individually in Table 7. The scores for Case K. M., with a bilateralfrontaMobe
lesion, are unremarkable, comparing favourably with the average performance
Table 7. Mean time scores (sec) for first 5 and last 5 trials, for patients with bilateral lesions
First lLast
Group 5 5
trials trials
Bilateral frontal
Case K.M. 87 13
Case A.N. 226 67
Bilateral hippocampal
Case H.M. 70 58
Case F.C. 88 19
Case P.B. 144 46
330 BRENDA MILNER
of the unilateral frontal-lobe group. Case A. N., however, has a higher initial time-score
than any other subject tested. This patient had a strong tendency to back-track all the
way to the starting-point during the first few trials, in order, as he explained, to rehearse the
correct moves before proceeding further. When he was prevented from doing this, he
remained "frozen" in the middle of the maze, reluctant to move farther away from the
starting-point. This is a striking instance of the difficulty in adapting themselves to test
instructions shown by many patients with frontal-lobe lesions.
In the bilateral hippocampal group, Cases F. C. and P. B. showed an appreciable
reduction in time score by the time they had completed their lengthy training, although
P. B. characteristically worked slowly on this as on many other tasks. Case H. M., on the
other hand, showed almost as little progress with respect to time as he did with respect
to errors, as can be seen from Fig. 8, in which the mean time for successive blocks of five
trials is plotted for the 3-day period.
Hear,. tlro.e C~c5)
par .5 trials
6O
SO
4o1
V-7
Dc~ Da~j Da~ a
J
'''~ .... ib .... ~3 i'"~ .... io i''~" .... 1o.... I~" ao
Trial Blocks
FIG.8. Learning curve for Case H. M., showingno reduction in time score after the first block
of trials, for a 3-day period.
There was some reduction in time from the first trial to the third, which reflected
increasing dexterity in the use of the stylus, but from then on there was no consistent
change.
4.4. Broken rules and repetitive errors
We turn now to a consideration of certain aspects of test behaviour which were specific
to the frontal-lobe group. These concern compliance with test instructions. The instructions
for the maze task seemed simple enough and, until the first patient with a frontal lobectomy
was tested, there was no reason to believe that they might prove a stumbling-block. This
patient, however, repeatedly failed to go back to the preceding bolthead when the error
counter clicked, although he knew that he was supposed to do so. Thenceforth all such
instances of"rule-breaking" were systematically recorded. Some patients with frontal-lobe
lesions persisted in back-tracking towards the starting-point (¢f. Case A. N.); others omitted
some of the steps on the route, or made diagonal moves in an attempt to approach the
VISUALLY-GUIDED MAZE LEARNING IN MAN 331
goal more directly. Many disregarded the clicking of the counter and the subsequent
warning cries of the experimenter. These findings are illustrated in Fig. 9, which depicts the
mean number of broken rules for the different lesion groups. It can be seen that subjects
with intact frontal lobes rarely, if ever, failed to carry out the test instructions, and the
ability of the patients with bilateral hippocampal lesions to follow instruction is particularly
noteworthy, in view of their difficulty in learning the correct path. Simple analysis of variance
on these scores yielded an F-ratio of 17.6 (P<0.001) and subsequent t-tests showed the
frontal-lobe group's mean score to be significantly higher than that of the other subjects
tested. (Frontal group vs Normal control group: t =5.39, P<0.001).
201 -
"d
o 15-
E
10--
0 ~
F T P RPo H N Group
No.
17 39 12 6 3 11 of Cases
FIG.9. Incidence of "broken rules", as related to locus of lesion. F, frontal; T, temporal; P,
parietal; RPo, right parieto-temporo-occipital; H, bilateral hippocampal; N, normal control.
One of the consequences of the rule-breaking was an increase in repetitive errors within
a single trial. Thus, if a patient retraced a portion of the correct path and then approached
the same choice-point for the second time, he was liable to repeat his previous error. Not all
instances of intratrial repetitive errors were of this type, however. Some frontal-lobe
patients would go back only one step after making an error (thus complying with the rules),
but then, after a brief pause, would repeat their previous incorrect response. This kind of
behaviour was rarely exhibited by the other subjects, although occasionally a patient with
a large right-posterior lesion might appear to lose his bearings in the maze and either backtrack
or make the same error twice, after trying another incorrect bolthead first, and being
seemingly quite unaware that he was repeating a response that had already proved
unsuccessftd.
Figure 10 shows the mean number ofintratrial repetitive errors for the various groups of
subjects. Apart from the right posterior group, only the frontal-lobe patients show any
tendency to make repetitive responses and, as described above, the behaviour appears to
have a different basis in the two groups. Analysis of variance showed the group differences
to be significant beyond the 0.001 level (F= 16.2). The right posterior group's mean score
was not significantly higher than that of the temporal, parietal, or normal control groups
and was significantly lower than that of the frontal group (t=2.81, P<0.02). The frontallobe
mean differed from that of the temporal, parietal and normal groups at well beyond
the 0.001 level of probability.
332 BRENDA MILNER
r'-
O
E
0
U
r-
0
F 1" P RPo H N Group
No.
17 39 12 6 3 11 of Cases
Incidence of intratrial repetitive errors, as related to locus of lesion. F, frontal;Fio. 10.
T, temporal; P, parietal; RPo, right parieto-temporo-occipital; H, bilateral hippoeampal;
N, normal control.
5. DISCUSSION
The findings reported here attest to the severity of the learning defect in patients with
bilateral hippocampal lesions. Case H. M., with a bilateral medial temporal lobectomy,
not only failed to progress from one training session to another, but also showed no improvement
within each set of 25 massed trials, although each trial lasted only about 60 see. At the
end of eight training sessions, he was still "having a little argument with himself" about
which direction to take at the first choice-point. This lack of progress is presumably related
to the complexity of the maze, in which there are 28 choice-points. Although it is possible
by various mnemonics to reduce considerably the number of turns which have to be
remembered, the whole route can still not readily be brought within the normal attention
span. This means that experience in the latter part of the maze interferes with the effective
rehearsal of the first part of the path, so that a patient such as H. M., who seems to rely
entirely upon verbal rehearsal to bridge a temporal gap, must start each new trial as if it
were a fresh problem.
All subjects but H. M. reduced their time scores in the course of training, but the
reduction was due essentially to a decrease in decision time rather than to a gain in motor
skill. It is therefore understandable that, in the case of H. M., not only the error scores,
but also the time scores failed to improve after the slight degree of additional dexterity
needed to manipulate the stylus accurately has been acquired. In contrast to these mazelearning
data, the results for the same patient on a mirror-drawing task showed a normal
reduction in time and errors with repeated practice, and this reduction carried over from
one day to the next, although the patient himself did not remember ever having done the
task before (MILNER [21]).
There have been numerous attempts to replicate the human memory loss in experimental
animals with hippocampal lesions, the results of which have been largely negative. It is
therefore of interest that KAnDA, RASMUSSENand KVEIM[22] have demonstrated a marked
impairment of multiple-choice maze-learning in rats with hippocampal lesions, a defect
VISUALLY-GUIDED MAZE LEARNING IN MAN 333
which seems not to be accountable for in terms of emotional or motivational change
(KvEIM, SETERLEIVand KAADA[23]). These maze-learning tests are of course all sequential
tasks, and KIMBLEand PRIBRAM[24] have data consistent with the notion that, in the monkey
hippocampal lesions interfere with the acquisition of behaviour sequences, whether
internally or externally regulated, whereas on a non-sequential task, discrimination learning,
animals with these lesions are unimpaired. However, although the maze-learning difficulty
of patients with bilateral hippocampal lesions, considered in isolation, could be described
as a disturbance of sequential behaviour, so restrictive a concept can hardly do justice to the
wide diversity of situations in which these patients are abnormally forgetful.
The severe impairment of learning exhibited by the bilateral hippocampal patients
answered the question which initially motivated this study, but some equally interesting
findings emerged from consideration of the scores of the different cortical-lesion groups
which had been intended as patient controls.
The impairment seen on this spatial learning task after right-temporal lobectomy, but
not after left, is consistent with earlier findings of visual, nonverbal learning deficits in
patients with right temporal-lobe lesions (KIMuRA [25]) and it contrasts with the verbal
learning deficits seen after left temporal lobectomy, but not after right (MEYERand YATES
[26], MILNER[27]; MILNERand KIMURA[28]). In the context it should be noted that many of
the patients in the left temporal-lobe group (and two in the left parietal group) were still
markedly dysphasic at the time they were learning the maze, and even those who were not
dysphasic had sufficient verbal memory impairment to prevent them from learning the
maze by the use of verbal mediating cues. Despite these verbal limitations, the maze learning
scores of the left-temporal group did not differ from those of normal control subjects. The
right temporal patients, on the other hand, repeatedly resorted to overt counting and other
verbal devices to learn the correct path, and yet proved to be inefficient learners.
The further discovery in the present study, that the occurrence of an appreciable
maze-learning impairment after right temporal lobectomy is contingent upon the inclusion
of a large part of the underlying hippocampus in the removal, emphasizes afresh the
importance of this structure for human learning processes. With unilateral excisions,
however, the learning deficit is specific rather than general, suggesting that each hippocampus
is concerned in the major functions of its own hemisphere.
The frontal-lobe group, which was not originally subdivided into left- and right-sided
cases, was also found to be significantly impaired. There were, however, marked individual
differences in maze-learning scores within the group, which may be related to side of lesion,
size of lesion, or both. Figure 4 shows the extent of removal for each patient, together with
the corresponding maze scores, the drawings being arranged in order of decreasing proficiency
of maze performance within the left and right groups, respectively. The right
frontal-lobe group tends to be the more impaired, with all scores but one outside the
normal range, but there is much overlap in the scores of the two groups, and no significant
difference in mean score was found. It is interesting, however, that the right-left difference
was more clear-cut in the same patients on a tactual maze learning task (CoRKIN [19]).*
*There weretwo additional cases,each representing a combination of lesionsin frontal and temporal
areas, one on the left, Case J. Eu., and the other on the right, Case L. P. Their scores werenot included
in the statistical analyses, sincethe lesions were unique. However, their scores on the visual maze were
instructive: for the left fronto-temporal case, J. Eu., there were 11 trials and 61 errors; for the right frontotemporal
case, L. P., 29 trials, 202 errors. The brain diagrams for these casescan be found in the paper by
CORK1N[19],Fig. I, this issue,p. 341.
334 BRENDA MILNER
The interpretation of any differences between the right and left frontal groups is rendered
more difficult by the fact that, with the exception of Case L. A., all the removals on the
right are large, whereas some of those on the left (and notably those in patients with
normal maze-learning scores) are small. Thus, it could be argued that lesion size, rather
than laterality, is the important factor in lowering maze performance after frontal lobectomy.
The fact that the two patients with bilateral injury to the frontal lobes, Cases K. M. and
A. N., were less impaired than some cases of right frontal lobectomy, might suggest that
side and not size is the important variable; however, this argument is not conclusive,
since it is not certain that more tissue was destroyed in these bilateral lesions than in a
radical unilateral lobectomy.
What relationship the high error scores of the frontal-lobe patients bear to their
inability to follow rules is unclear. Breaking rules does not directly increase the error score
to any appreciable extent. However, this tendency may reflect an underlying dysfunction
which is responsible also for the slow learning, but which is different in kind fi'om that
produced by other lesions. It may be, for example, that the frontal-lobe patients are initially
less interested than others in avoiding errors (Mn~NER[29]).
It should be noted that this behaviour on the maze task is dissociable, in individual
cases, from the perseverative behaviour shown by frontal-lobe patients on the Wisconsin
Card Sorting Test (MILNER [30]). Thus, Case M. S., with a radical right frontal lobectomy
obtained a combined score of 70 broken rules and intratrial repetitive errors on the maze,
but showed normal flexibility of response on the sorting test, whereas Cases R. B. and J. Mo.,
with small left superior-frontal lesions, performed normally on the maze, but made an
exceedingly high number of perseverative errors on the sorting test.
The good performance of the group of patients with parietal-lobe lesions was unexpected,
in view of the clinical literature. It is true that 9 of the 12 parietal patients were tested
several years after operation, when they had been seizure-free for some time; however,
the three parietal patients who were tested only 18 days after operation also obtained normal
scores, even though one of them was severely aphasic at the time. Inspection of the individual
scores suggests, rather, that the small size of the lesions may be the main reason
why the parietal-lobe patients performed so well; the only patient with scores outside the
normal range is Case M. R., who had what was certainly the largest lesion. This patient,
moreover, was tested four years after operation and was no longer having seizures, but his
scores indicate a marked impairment of maze learning.
The possibility the lesion size in the posterior cortex is a relevant factor in maze learning
derives further support from the very poor performance of the small group of patients
with large right posterior lesions involving the parietal, temporal and occipital lobes.
It is true that four of the six patients in this group had also had the hippocampus radically
excised, but the tendency of some patients to lose their bearings in the maze suggested
more than a learning deficit.
The defective maze performance of the right posterior group also raises the question of
the possible interfering role of visual field defects in maze performance. Partial homonymous
upper quadrant defects on the side contralateral to the lesion were demonstrated in about
half the cases of unilateral temporal lobectomy. These defects occurred no more often with
right-sided lesions than with left and bore no relation to maze performance. There were
fewer patients with lower quadrant or hemianopic defects, but these of course are the defects
which might be expected to produce most disturbance on visual route-following tasks. Two
left parietal patients had right lower-quadrant defects: one, Case P. Mo., performed
VISUALLY-GUIDED MAZE LEARNING IN MAN 335
normally; the other, Case M. R., (with a larger lesion) was impaired. Amongst the righthemisphere
cases, only the six patients in the right posterior cortical group had lower
quadrant or hemianopic defects, and, although their error scores on the maze were high,
the mean was a little smaller than that of the right frontal-lobe group. Moreover, these
patients did not show any neglect of the left half of visual space. These data suggest that
the field defects of the right posterior group, though coincidental with maze defects, were
not influential in producing them.
Since WEINSTEIN,SEMMES,GHENT and TEUBER [31] had found impairment on their
tactual and visual route-finding tasks to be positively correlated with deficits in twopoint
discrimination on the hand, it was decided to investigate the relationship between
maze-learning scores and two-point discrimination thresholds for the 62 patients for whom
such data were available. It can be seen from Table 8 that the mean number of trials to
criterion and the mean number of errors for patients with significant discrimination defects
Table 8. Comparison of maze-learning
scores for unilateral groups with and
without defects in 2-point discrimination.
No. Mean scores
Group of
cases Trials Errors
With defect 18 24.2 156
Without defect 44 30.9 179
are slightly lower than those for the group without such defects. Since sensory defects,
in this patient population, occurred most frequently with lesions invading the Rolandic
cortex, and since most of the small parietal-lobe lesions were of this kind, it is not surprising
that two-point discrimination defects did not correlate positively with impaired maze
performance. It seems likely, then, that the positive relationship between route-finding
difficulty and two-point loss, found by WEINSTEINet aL, merely reflects a partial coincidence
of the loci of lesion responsible for the two defects, as the authors themselves suggest
(SEMMESet al. [15]).
The maze used here differs from the route-finding tasks used by SEMMES et al. [5]
not only in being a learning task (as has already been pointed out), but also in being a
stylus and not a locomotor task. It is known from the work of ORBACH[33] that locomotormaze
performance of peripherally blinded monkeys is more vulnerable than stylus-maze
performance to occipital lesions. Moreover, in the stylus-maze there was no need for the
subject to be constantly shifting from one set of coordinates to another. At all times the
entire maze was visible and its position relative to the subject's own body remained fixed.
It seems reasonable, therefore, that this stepping-stone stylus maze would be less likely
to uncover mild disorders of spatial orientation than would the map-reading tasks used by
SEMMES and her co-workers [5]. This difference in task could perhaps account for the
apparent contradiction between the finding of SEMMESet al. [5] that patients with parietallobe
lesions (and only these patients) showed a residual impairment on the map-reading
tasks years after penetrating brain wounds and the present finding of normal performance
on the stylus maze in the parietal-lobe group. Nevertheless the possible importance of
size of parietal-lobe lesion in producing this difference cannot lightly be dismissed. We
have seen that some of the patients with large right-posterior cortical excisions did in
fact lose their bearings in the stylus maze during the early stages of training, and, more
336 BRENDA MrLNER
recently, we have had a striking example of this behaviour from a patient with a large
cortical excision limited to the parietal lobe.
The data for this 16-year-old-girl, Case J. Du., were not included in the present analyses
because she is one of a new series of patients who are learning the visual maze only after
having first been trained on Dr. Suzanne Corkin's tactual maze. However, her test behaviour
is so pertinent to the present discussion that it will be described here.
FIa. 11. CaseJ. Du., right parietal lobectomy. Not includedin this study becausetested after
previoustraining on tactual maze. Visualmazescores: 52 trials, 400 errors.
Figure 11 shows the lateral extent of the parietal cortical excision, in this case, the
removal on the mesial surface being small. It can be seen that the area excised is roughly
the sum of the two small right parietal-lobe excisions of Cases S. La. and H. Ja. shown in
Fig. 2, though sparing the temporal lobe completely.
It should be pointed out that this patient did not lose her way in the hospital; her
visual fields showed only very slight peripheral constriction, in the left lower quadrant, and
she had no tendency to neglect the left half of extrapersonal space. Her performance on a
number of visual learning tasks and on the Wisconsin Card Sorting Tests was entirely
normal. Yet both maze tasks proved extremely difficult and stressful for her.
Three weeks after operation, after prior training on the tactual maze, she learned the
visual maze in 52 trials with 400 errors. During the course of this slow acquisition, she
lost her way many times and became very discouraged, the testing being interrupted by
frequent bouts of weeping. She referred to the maze as a "jungle" and doubted that she
would ever learn it. In the early trials she built up a total of 20 broken rules and 4 intratrial
repetitive errors, mainly at the most difficult choice points, in the middle of the maze,
where she clearly did not know which way she had come nor which way to turn next. This
behaviour shows that the stylus-maze task is sensitive to relatively mild forms of spatial
disorientation, even if less so than a locomotor maze.
The findings in Case J. Du., whose parietal-lobe lesion was large, suggest that lesion
size may indeed by a critical factor in determining the degree of spatial disorientation after
a parietal-lobe lesion. This fact, together with the differing spatial demands of the two tasks
may reconcile the present findings with those of SEMMESet aL [5].
The finding of an impairment in maze learning after large lesions of the right posterior
cortex is consistent with earlier work on topographical orientation in cases of cerebral
disease (PATERSONand ZANGWILL[9]; McFm et al. [10]; HI~CAENet al. [11]; ETTLINGERet al.
[12]; HI~CAEN[13]; WHITTY and NEWCOMBE[14]). The fact that the present study does
not include comparable lesions in the left posterior cortex prevents an assessment of the
relative contribution of the two hemispheres to maze learning. For the temporal lobe,
however, a greater effect from right-sided lesions is clearly shown and there is a suggestion
that this holds true for the frontal lobe also. Although large removals anywhere in the
right hemisphere impair maze performance, it does not follow that they do so for the same
reasons; instead, it seems likely that the anterior and posterior regions contribute in
qualitatively different ways to performance on this spatial learning task.
VISUALLY-GUIDEDMAZELEARNINGIN MAN 337
Acknowledgements--I wish to thank Dr. WILDERPENFIELD,Dr. THEODORERASMUSSEN,Dr. CHARLESBRANCH
and Dr. WILLIAMFEINDELfor the continuing opportunity to study their patients at the Montreal Neurological
Institute. I am also most grateful to Dr. WILLIAMSCOVILLEfor making possible the extensive testing of
Case H. M., and to Professor HANS-LUKASTEUBERfor so kindly allowing me to include his case, A. N., in
my study. Dr. M. P. BRYDENbuilt the apparatus and I am greatly appreciative of his help. This work was
supported in part by Canadian Federal-Provincial Health Grant No. 604-5-89, in part by U. S. Public
Health Service Research Grant NB 02831 from the National Institute of Nervous Diseases and Blindness
and Small Grant M5 774 from the National Institute of Mental Health, and in part by the Medical Research
Council of Canada.
REFERENCES
1. SCOVILLE,W. B. and MILNER,BRENDA. Loss of recent memory after bilateral hippocampal lesions.
J. Neurol. Neurosurg.Psychiat. 20, 11-21, 1957.
2. PENFIELD,W. and MILNER,BRENDA. Memory deficits produced by bilateral lesions in the hippocampal
zone. A.M.A. Archs NeuroL Psychiatry 79, 475-497, 1958.
3. MILNER,BRENDA. The memory defect in bilateral hippocampal lesions. Psychiat. Res. Rep. 11, 43-52,
1959.
4. BARKER,R. G. The stepping-stone maze: a directly visible space-problem apparatus. J. gen. PsychoL
5, 280-285, 1931.
5. SEMMES,JOSEPHINE,WEINSTEIN,S., GHENT,LILAand TEUBER,H.-L. Spatial orientation in man after
cerebral injury--I: Analyses by locus of lesion. J. PsychoL 39, 227-244, 1955.
6. ELITHORN,A. A preliminary report on a perceptual maze test sensitive to brain damage, d. NeuroL
Neurosurg. Psychiat. 18, 287-292, 1955.
7. ELITHORN,A. Intelligence, perceptual integration and the minor hemisphere syndrome. Neuropsychologia
2, 327-332, 1964.
8. BENTON,A. L., ELITHORN,A., FOGEL,M. L. and KERR,M. A perceptual maze test sensitive to brain
damage. J. Neurol. Neurosurg.Psychiat. 26, 540-544, 1963.
9. PATERSON,A. and ZANGWILL,O. L. Disorders of visual space perception associated with lesions of
the right cerebral hemisphere. Brain 67, 331-358, 1944.
10. MCFIE, J., PIERCY,M. F. and ZANGWILL,O. L. Visual spatial agnosia associated with lesions of the
right cerebral hemisphere. Brain 73, 167-190, 1950.
11. HI~CAEN,H., PENFIELD,W., BERTRAND,C. and MALMO,R. The syndrome of apractognosia due to
lesions of the minor cerebral hemisphere. A.M.A. Archs Neurol. Psychiatry 75, 40(0-434, 1956.
12. ETTLINGER,G., WARRINGTON,ELIZABETHand ZANGWILL,O. L. A further study of visual-spatial
agnosia. Brain 80, 335-361, 1957.
13. H~CAEN, H. Clinical symptomatology in right and left hemispheric lesions. In Interhemispheric
Relations and Cerebral Dominance, MOLrNTCASTLE,V. B. (Editor), pp. 215-243, Johns Hopkins Press,
Baltimore, 1962.
14. WmTTY, C. W. M. and NEWCOMBE,FREDA. Disabilities associated with lesions in the posterior parietal
region of the non-dominant hemisphere. Neuropsychologia3, 175-186, 1965.
15. SEMMES,JOSEPHINE,WEINSTEIN,S., GHENT,LILAand TEUBER,H.-L. Correlates of impaired orientation
in personal and extrapersonal space. Brain 86, 747-772, 1963.
16. LANDIS,C. and ERLICK,D. An analysis of the Porteus maze-test as affected by psychosurgery. Am. d.
PsychoL 63, 557-566, 1950.
17. PORTEUS,S. D. The Maze Test and ClinicalPsychology. Pacific Books, Pale Alto, Calif., 1959.
18. HEBB,D. O. and PENHELO,W. Human behavior after extensive bilateral removal from the frontal
lobes. Archs NeuroL Psychiat, Chicago44, 421-438, 1940.
19. CORKIN,SUZANN~. TactuaUy-guided maze-learning in man: Effects of unilateral cortical excisions and
bilateral hippocampal lesions. Neuropsychologia3, 339-351, 1965.
20. BARTLETt,M. J. The use of transformations. Biometrics 3, 39-51, 1947.
21. MILNER,BRENDA. Les troubles de la m6moire accompagnant des 16sions hippocampiques bilat6rales.
In Physiologie de l'Hippocampe, pp. 257-272. C.N.R.S., Paris, 1962.
22. KAADA,B. R., RASMUSSEN,E. W. and KVEIM,O. Effects of hippocampal lesions on maze learning and
retention in rats. ExpL Neural 3, 335-355, 1961.
23. KVEIM,O., SETEKLEIV,J. and KAADA,B. R. Differential effects of hippocampal lesions on maze and
paSSive avoidance learning in rats. Expl. Neurol. 9, 59-72, 1964.
24. KIMaLE, D. P. and PRmRAM,K. H. Hippocampectomy and behavior sequences. Science,N. Y. 139,
824-825, 1963.
25. KIMORA,DOREEN, Right temporal-lobe damage. Archs Neurol., Chicago8, 264-271, 1963.
26. MEYER,V. and YATES, A. J. Intellectual changes following temporal lobectomy for psychomotor
epilepsy, d. Neural. Neurosurg.Psychiat. 18, 44-52, 1955.
338 BRENDA MILNER
27. MILNER, BRENDA. Psychological defects produced by temporal-lobe excision. Res. Pubis Ass. Res.
nerv. merit. Dis. 36, 244-257, 1958.
28. MILNER,BRENDAand KIMURA,DOREEN. Dissociable Visual Learning Dejects aJter Unilateral Temporal
Lobectomy in Man. Paper read at East. Psychol. Ass., Philadelphia, April, 1964.
29. MILNER, BRENDA. Some effects of frontal lobectomy in man. In The Frontal Granular Cortex and
Behavior, WARRENJ. M. and AKERT K. (Editors), pp. 315-354, McGraw-Hill, New York, 1964.
30. MILNER,BRENDA. Effects of different brain lesions on card sorting. Archs Neurol., Chicago 9, 90-100,
1963.
31. WEINSTEIN,S., SEMMES,JOSEPHINE, GHENT, ElLA and TEUBER, H.-L. Spatial orientation in man after
cerebral injury--II: Analysis according to concomitant defects. J. Psychol. 42, 249-263, 1956.
32. CORKIN,SUZANNE,MILNER,BRENDAand RASMUSSEN,T. Effects of different cortical excisions on sensory
thresholds in man. Trans. Am. neuroL Ass. 89, 112-116, 1964.
33. ORBACH,J. "Functions" of striate cortex and the problem of mass action. Psychol. Bull. 56, 217-292,
1959.
R6sum6--L'apprentissage d'un labyrinthe sous contr61e visuel a 6t6 etudi6 chez 79 malades avec
diff6rentes 16sions c6r6brales. R6sultats: (1) les 16sions bilat6rales de l'hippocampe produisent
le d6ficit le plus important, un malade ne manifestant aucun apprentissage au cours de 215
essais; (2) les lobectomies temporales droites produisent un d6ficit significatif, plus particuli6rement
lorsqu'elles sont accompagn6es de l'ablation radicale de l'hippocampe sous-jacent.
Par contre, les malades ayant une 16sion temporale gauche de m6me 6tendue, ou une 16sion
pari6tale (droite ou gauche) de faibles dimensions, donnent des r6sultats normaux; (3) des
16sions plus grandes d'un lobe frontal ou du cortex post6rieur droit ralentissent fortement
l'apprentissage----ces donn6es sugg6rent que non seulement la localisation mais 6galement
l'6tendue d'une 16sion corticale influencent l'apprentissage du labyrinthe; (4) seules les 16sions
frontales affectent la capacit6 b. suivre les consignes reques.
Zusammenfassung--Das Erlernen eines Labyrinthes mit visueller Darbietung wurde bei 79
Patienten mit verschiedenen Hirnlfisionen untersucht. Die wichtigsten Befunde sind: (1)
beidseitige L/isionen des Hippocampus fiihren zur gr6ssten Lernschwierigkeit; im Verlauf von
215 Einzelversuchen war bei einem Patienten iJberhaupt kein Lerneffekt sichtbar; (2) Abtragungen
des rechten Temporallappens fiihren zu einer signifikantenVerlangsamung des Lernens, ganz
besonders wenn sie von einer radikalen Abtragung des daruntergelegenen Hippocampus
begleitet sind. Hingegen weisen Patienten mit gleich grossen L/isionen des linken Temporallappens
oder mit geringen Lfisionen des linken oder rechten Parietallappens normale Resultate
auf; (3) gr/Sssere L/isionen eines Frontallappens oder der rechten hinteren Hirnrinde verlangsamen
das Lernen stark--dies fiihrt zum Schluss, dass nicht nut die Lokalisation, sondern
auch die Ausdehnung einer Hirnrindenlfision das Erlernen des Labyrinthes beeinflussen;
(4) nur Frontallappenl/isionen erschweren das korrekte Ausfiihren der erhaltenen lnstruktionen.