Neuropsyehologia, 1977, Vol. 15, pp. 43 to 49. Pergamon Press. Printed in England.
SEQUENTIAL MEMORY FOR FIGURES IN BRAIN-DAMAGED PATIENTS* E. DE RENZI, P. FAGLIONIand P. VILLA Clinica Neurologica, Via del Pozzo 71, Modena) Italy (Received 7 November 1975)
Abstract--Fifty control patients and 105 patients with damage restricted to one hemisphere were given two tests, one made up of 10 meaningful figures and the other of 8 meaningless figures. Study trials were alternated with test trials and the patient's task over test trials was to rearrange the items according to the same sequence as they had appeared over study trials, until a criterion of 3 consecutive errorless runs was attained. Relative to controls, patients with damage to the posterior areas of the right hemisphere were significantly impaired on both tests, whereas the corresponding left brain-damaged group scored marginally poorer on the meaningless figure test only. Since nearly half of the left-sided patients were aphasics, the inference seems justified that visual serial learning is much more dependent on the integrity of the visual memory code than on that of the verbal memory code.
THE CONTRIBUTIONof pathology to the study of memory processes is not limited to the identification of neuro-anatomical structures critical for remembering, but extends to the comprehension of mechanisms underlying storage and retrieval of traces. The debated distinction between short-term and long-term memory has, for instance, received strong support from the finding that there are patients who are selectively impaired on either of these stages, the other being preserved [1]. Valuable inferences on the psychology of memory can also be drawn from the study of patients without an overt amnesic syndrome, because they often suffer from the impairment of abilities, such as language or visual perception, which though not being primarily mnestic, are directly related to the availability of the verbal and imaginal codes, whereby information is stored [2, 3]. The bearing of these two codes on memory processing has been a subject of discussion in recent years and the view has been advanced that the extent to which they are used is dependent upon the nature of the stimulus and of the task. On the basis of intuitive assumptions and of reaction time studies, PAIVIOand CSAPO [4] propose the following ranking of the availability of each code with reference to the kind of stimulus; words have the easiest access to the verbal code and pictures to the imaginal code; second comes the verbal code for familiar (i.e. readily labeled) pictures, third the imaginal code for concrete words, and fourth the verbal code for unfamiliar pictures and the imaginal code for abstract words, A differential specialization of the two coding systems has also been assumed with respect to the nature of the representational mechanisms required by the task. The verbal code would be essential for tasks where the serial order of item presentation must be retained (e.g. memory span and serial learning), whereas the image code would show its proficiency whenever items are processed and stored as units independent from one another. Evidence * This study was supported by a C.N.R. grant. 43
44
E. DE RENZI, P. FAGLIONIand P. VILLA
in support of this hypothesis has been provided in normals by PAIVIO and CSAFO [4] and by DEL CAST1LLOand GOMENmK [5] who manipulated the rate of item presentation, stimulus concreteness and the demand on sequential processing involved in the memory task. N o t only sequential memory for pictures, but also sequential memory for unfamiliar (meaningless) forms [5] improved when an opportunity was given to produce a verbal code for them. That the availability of the verbal code is specifically instrumental to perform sequential memory tasks is suggested by the differential effect obtained by NELSON et al. [6, 7] when they manipulated the phonetic similarity of the verbal labels of pictures: pictures with similar labels were poorly learned when the task was to rearrange their order of appearance, but not when they had to be associated with their mate in a paired associates learning, whereas high phonetic similarity impaired both performances when the corresponding words were used. The superiority of the verbal code for sequential processing and of the imaginal code for parallel processing was not, however, confirmed by SNODGRASS and ANTONE [8] who presented picture and word pairs either in a spatial or in a temporal relationship and requested subjects to recognize after 5 min which pair was reversed and which was in the same order. Memory for spatial pairs was better than memory for temporal pairs and memory for picture pairs was better than memory for word pairs, but no task type by material interaction occurred. Patients with unilateral brain damage may offer an alternative approach to evaluating the relative bearing of verbal and imaginal coding to memory. Since mechanisms underlying language are subserved by the left hemisphere, patients suffering from injury to this side of the brain should show a greater impairment not only when tested for learning verbal material [9-12] but also when given familiar pictures, which can be easily labeled. If, as maintained by PAIVIO [3], learning the serial order of familiar pictures is particularly suited to call the verbal code into play, left hemisphere patients should be at disadvantage on this task with respect to right hemisphere patients. These in turn may be expected to be impaired when tested with unfamiliar, meaningless figures, since they are known to have trouble in memorizing visual nonverbal material [13-15] and to be prone to resort to verbal labeling as much as possible, a strategy difficult, though not impossible, to pursue in dealing with random patterns. If, on the other hand, the imaginal code plays a predominant role in long-term visual memory tasks, patients with posterior localized damage, expecially those with involvement of the right hemisphere [13] may be expected to fail both on meaningful and meaningless figure tasks. In the present research, learning to criterion of the order of appearance of meaningful and meaningless items was studied in normals and in two hemisphere-damaged samples. The brain-damaged groups were further subdivided in patients with and without visual field defects, on the assumption that the first had injury involving the retro-Rolandic area and the second predominantly anterior damage. MATERIALS AND METHOD Subjects
One hundred and fifty-five right-handed subjects were examined. Fifty were control patients, namely patients admitted in the wards for diseases not involving the brain; 105 were patients with damage restricted to one hemisphere on the basis of clinical, brain scanning, LEG and neuroradiological data. Fifty-one of them were suffering from right hemisphere lesion and 54 from left hemisphere lesion. Both hemispheric groups were further subdivided into those without visual field defects (VFD) and those with VFD. There were, therefore, 5 groups and the number of patients for each of them was a follows: Controls (C) = 50; left-brain damaged patients without VFD (LH ) ~ 33; left brain-damaged patients
45
SEQUENTIAL MEMORY FOR FIGURES IN BRAIN-DAMAGED PATIENTS
with VFD (LH+) = 21; right brain-damaged patients without VFD (RH--) = 27; right brain-damaged patients with VFD (RH+) = 24. Presence of VFD was ascertained by means of the confrontation method, supplemented, in any apparently negative case, by symmetrical double simultaneous stimulation in the upper and lower temporal quadrants of the binocular visual fields. Left brain-damaged patients were given the Token Test [161 to determine the presence and severity of aphasia. Twenty-three of them (10 in the group without VFD and 13 in the group with VFD) were diagnosed as aphasics on the basis of their performance on this test. Etiology was controlled and found not to differ significantly across the groups.
Testingprocedures Two tests requiring learning the order of appearance of visual stimuli were given, alternating a learning trial with a test trial. Over the learning trials stimuli were presented always in the same order one at a time for 3 see. Immediately after the last item was shown, a parallel series of the same stimuli was laid down in two rows on the table in a sequence that was always different from that used on learning trials and that systematically changed over the first 10 test trials. For each successive block of 10 test trials the same order recurred. The patient was asked to rearrange the items in the same order as that in which they had appeared on learning trials. Practice was continued until the criterion of three successive errorless runs was attained. Training was carried out for 25 trials; if the criterion was not reached, the next day a second block of 25 trials was administered. Whenever the patient failed to attain the criterion at the end of this second block, a score of 50 was given. On each test trial notice was taken of which items had been correctly placed. The :meaningfulfigure test was always given first and consisted of 10 coloured realistic drawings; the meaningless figure test was made up of 8 random shapes drawn from those published by VANDEm'LASand GARVJN[17]. Figure 1 shows their order of appearance on the learning trials. Both tests were proceeded by a 2- and a 3-items buffer test for demonstration. If the patient was unable to understand the instructions on these preliminary trials, he was discarded. RESULTS The data o f each test have been analysed by a one way analysis of covariance, with age a n d year of schooling as co-variates. The b r a i n damage effect, the hemisphere effect, the visual field defect effect, the hemisphere x V F D interaction a n d the pairwise comparisons between groups were carried out with the SCHEFF~ m e t h o d [18] which keeps the risk of type one error at the level of the general analysis. Table 1 shows the m e a n scores obtained by controls a n d the 4 hemispheric groups on the meaningful figure test. The n u m b e r of trials to criterion is approximately the same in Table 1. Mean trials to criterion on the meaningfuland meaningless figure test MeaningfulFtest Meaningless Ftest
C 9'89 12.66
LH-10"94 15.65
LH-+13-59 ~ 20.12
RH-9"75 15-79
RH+ 16"15 22.05
controls a n d in the two b r a i n - d a m a g e d groups without V F D , b u t it increases in the two groups with V F D , especially in the right-sided one. This was confirmed by the analysis of covariance, which showed that the b r a i n damage effect a n d the hemisphere effect were n o t significant ( F (4,148): 1.24 a n d < 1, respectively), while the V F D effect had a n F (4,148): 2.64, P < 0.05. W h e n the groups were compared to each other, the R H ~ - one was f o u n d to be significantly inferior to controls ( F (4,148): 3.07, P < 0.025) a n d to R H - - patients ( F (4,148): 2.53, P < 0-05). N o difference between the other groups reached the confidence level, n o t even that between L H q - patients a n d controls ( F (4,148): 1.07). The p a t t e r n o f results was n o t substantially different for the meaningless figure test (Table 1). Brain-damaged patients as a whole performed more poorly t h a n controls ( F (4,148): 3.00, P < 0.025), b u t this was due to the poorer scores of the two groups with
46
E. DE RENZI, P. FAGLIONIand P. VILLA
VFD, the F (4,148) of 1.99 being significant at the 0.05 level, if a one-tailed test is used. Neither the hemisphere effect, nor the interaction are significant (F < 1 in both cases). The R H + group is significantly inferior to controls at the 0.01 level (F (4,148): 3.88), while the comparison between the corresponding left-sided group and controls reaches the 5 significance level only if a one-tailed test is used (F (4,148): 2.20). No other inter-group comparison is significant. The number of trials to criterion of left brain-damaged patients did not exceed that of the R H + group even when the evaluation was restricted to the left sided patients who were diagnosed aphasic on the basis of their Token Test score. There were 23 such patients and their mean on the meaningful figure test was 14.91, and on the meaningless figure test was 20.91, both lower than the corresponding means of the R H + group. Moreover, severity of aphasia was not significantly correlated with the memory performance: the r between Token Test scores and meaningful figure test scores was -4).34 (P > 0.05 with 21 D.F.) and the r between the Token Test scores and the meaningless test scores was ---0.17 (P > 0.05 with 21 D.F.). It is also worth noting that the majority of patients unable to reach the criterion were found among R H + patients. There were 2 such patients in the meaningful figure test and both were RH-t- patients; there were 8 patients in the meaningless figure test: 4 were R H + , 2 R H - - , 1 L H + and 1 L H - - . Since an arbitrary score of 50 was given when the criterion was not attained within 50 trials, the mean performance of the R H + group is likely to be actually worse than it has been estimated. The correlation coefficients between the meaningful and the meaningless figure test were 0.50 in normals, 0.72 in left brain-damaged patients and 0.77 in right brain-damaged patients, all with a P < 0.001. Standard serial position curves of the meaningful figure test were obtained for each ArcsinV~- = 1 . 6 1 0 - 0 O 0 5 x - O . O 5 8 x 2 . O.OO6x ~
160 -150 - 140
130
.N~
--
- -
120 -.E
SEQUENTIAL MEMORY FOR FIGURES IN BRAIN-DAMAGED PATIENTS* E. DE RENZI, P. FAGLIONIand P. VILLA Clinica Neurologica, Via del Pozzo 71, Modena) Italy (Received 7 November 1975)
Abstract--Fifty control patients and 105 patients with damage restricted to one hemisphere were given two tests, one made up of 10 meaningful figures and the other of 8 meaningless figures. Study trials were alternated with test trials and the patient's task over test trials was to rearrange the items according to the same sequence as they had appeared over study trials, until a criterion of 3 consecutive errorless runs was attained. Relative to controls, patients with damage to the posterior areas of the right hemisphere were significantly impaired on both tests, whereas the corresponding left brain-damaged group scored marginally poorer on the meaningless figure test only. Since nearly half of the left-sided patients were aphasics, the inference seems justified that visual serial learning is much more dependent on the integrity of the visual memory code than on that of the verbal memory code.
THE CONTRIBUTIONof pathology to the study of memory processes is not limited to the identification of neuro-anatomical structures critical for remembering, but extends to the comprehension of mechanisms underlying storage and retrieval of traces. The debated distinction between short-term and long-term memory has, for instance, received strong support from the finding that there are patients who are selectively impaired on either of these stages, the other being preserved [1]. Valuable inferences on the psychology of memory can also be drawn from the study of patients without an overt amnesic syndrome, because they often suffer from the impairment of abilities, such as language or visual perception, which though not being primarily mnestic, are directly related to the availability of the verbal and imaginal codes, whereby information is stored [2, 3]. The bearing of these two codes on memory processing has been a subject of discussion in recent years and the view has been advanced that the extent to which they are used is dependent upon the nature of the stimulus and of the task. On the basis of intuitive assumptions and of reaction time studies, PAIVIOand CSAPO [4] propose the following ranking of the availability of each code with reference to the kind of stimulus; words have the easiest access to the verbal code and pictures to the imaginal code; second comes the verbal code for familiar (i.e. readily labeled) pictures, third the imaginal code for concrete words, and fourth the verbal code for unfamiliar pictures and the imaginal code for abstract words, A differential specialization of the two coding systems has also been assumed with respect to the nature of the representational mechanisms required by the task. The verbal code would be essential for tasks where the serial order of item presentation must be retained (e.g. memory span and serial learning), whereas the image code would show its proficiency whenever items are processed and stored as units independent from one another. Evidence * This study was supported by a C.N.R. grant. 43
44
E. DE RENZI, P. FAGLIONIand P. VILLA
in support of this hypothesis has been provided in normals by PAIVIO and CSAFO [4] and by DEL CAST1LLOand GOMENmK [5] who manipulated the rate of item presentation, stimulus concreteness and the demand on sequential processing involved in the memory task. N o t only sequential memory for pictures, but also sequential memory for unfamiliar (meaningless) forms [5] improved when an opportunity was given to produce a verbal code for them. That the availability of the verbal code is specifically instrumental to perform sequential memory tasks is suggested by the differential effect obtained by NELSON et al. [6, 7] when they manipulated the phonetic similarity of the verbal labels of pictures: pictures with similar labels were poorly learned when the task was to rearrange their order of appearance, but not when they had to be associated with their mate in a paired associates learning, whereas high phonetic similarity impaired both performances when the corresponding words were used. The superiority of the verbal code for sequential processing and of the imaginal code for parallel processing was not, however, confirmed by SNODGRASS and ANTONE [8] who presented picture and word pairs either in a spatial or in a temporal relationship and requested subjects to recognize after 5 min which pair was reversed and which was in the same order. Memory for spatial pairs was better than memory for temporal pairs and memory for picture pairs was better than memory for word pairs, but no task type by material interaction occurred. Patients with unilateral brain damage may offer an alternative approach to evaluating the relative bearing of verbal and imaginal coding to memory. Since mechanisms underlying language are subserved by the left hemisphere, patients suffering from injury to this side of the brain should show a greater impairment not only when tested for learning verbal material [9-12] but also when given familiar pictures, which can be easily labeled. If, as maintained by PAIVIO [3], learning the serial order of familiar pictures is particularly suited to call the verbal code into play, left hemisphere patients should be at disadvantage on this task with respect to right hemisphere patients. These in turn may be expected to be impaired when tested with unfamiliar, meaningless figures, since they are known to have trouble in memorizing visual nonverbal material [13-15] and to be prone to resort to verbal labeling as much as possible, a strategy difficult, though not impossible, to pursue in dealing with random patterns. If, on the other hand, the imaginal code plays a predominant role in long-term visual memory tasks, patients with posterior localized damage, expecially those with involvement of the right hemisphere [13] may be expected to fail both on meaningful and meaningless figure tasks. In the present research, learning to criterion of the order of appearance of meaningful and meaningless items was studied in normals and in two hemisphere-damaged samples. The brain-damaged groups were further subdivided in patients with and without visual field defects, on the assumption that the first had injury involving the retro-Rolandic area and the second predominantly anterior damage. MATERIALS AND METHOD Subjects
One hundred and fifty-five right-handed subjects were examined. Fifty were control patients, namely patients admitted in the wards for diseases not involving the brain; 105 were patients with damage restricted to one hemisphere on the basis of clinical, brain scanning, LEG and neuroradiological data. Fifty-one of them were suffering from right hemisphere lesion and 54 from left hemisphere lesion. Both hemispheric groups were further subdivided into those without visual field defects (VFD) and those with VFD. There were, therefore, 5 groups and the number of patients for each of them was a follows: Controls (C) = 50; left-brain damaged patients without VFD (LH ) ~ 33; left brain-damaged patients
45
SEQUENTIAL MEMORY FOR FIGURES IN BRAIN-DAMAGED PATIENTS
with VFD (LH+) = 21; right brain-damaged patients without VFD (RH--) = 27; right brain-damaged patients with VFD (RH+) = 24. Presence of VFD was ascertained by means of the confrontation method, supplemented, in any apparently negative case, by symmetrical double simultaneous stimulation in the upper and lower temporal quadrants of the binocular visual fields. Left brain-damaged patients were given the Token Test [161 to determine the presence and severity of aphasia. Twenty-three of them (10 in the group without VFD and 13 in the group with VFD) were diagnosed as aphasics on the basis of their performance on this test. Etiology was controlled and found not to differ significantly across the groups.
Testingprocedures Two tests requiring learning the order of appearance of visual stimuli were given, alternating a learning trial with a test trial. Over the learning trials stimuli were presented always in the same order one at a time for 3 see. Immediately after the last item was shown, a parallel series of the same stimuli was laid down in two rows on the table in a sequence that was always different from that used on learning trials and that systematically changed over the first 10 test trials. For each successive block of 10 test trials the same order recurred. The patient was asked to rearrange the items in the same order as that in which they had appeared on learning trials. Practice was continued until the criterion of three successive errorless runs was attained. Training was carried out for 25 trials; if the criterion was not reached, the next day a second block of 25 trials was administered. Whenever the patient failed to attain the criterion at the end of this second block, a score of 50 was given. On each test trial notice was taken of which items had been correctly placed. The :meaningfulfigure test was always given first and consisted of 10 coloured realistic drawings; the meaningless figure test was made up of 8 random shapes drawn from those published by VANDEm'LASand GARVJN[17]. Figure 1 shows their order of appearance on the learning trials. Both tests were proceeded by a 2- and a 3-items buffer test for demonstration. If the patient was unable to understand the instructions on these preliminary trials, he was discarded. RESULTS The data o f each test have been analysed by a one way analysis of covariance, with age a n d year of schooling as co-variates. The b r a i n damage effect, the hemisphere effect, the visual field defect effect, the hemisphere x V F D interaction a n d the pairwise comparisons between groups were carried out with the SCHEFF~ m e t h o d [18] which keeps the risk of type one error at the level of the general analysis. Table 1 shows the m e a n scores obtained by controls a n d the 4 hemispheric groups on the meaningful figure test. The n u m b e r of trials to criterion is approximately the same in Table 1. Mean trials to criterion on the meaningfuland meaningless figure test MeaningfulFtest Meaningless Ftest
C 9'89 12.66
LH-10"94 15.65
LH-+13-59 ~ 20.12
RH-9"75 15-79
RH+ 16"15 22.05
controls a n d in the two b r a i n - d a m a g e d groups without V F D , b u t it increases in the two groups with V F D , especially in the right-sided one. This was confirmed by the analysis of covariance, which showed that the b r a i n damage effect a n d the hemisphere effect were n o t significant ( F (4,148): 1.24 a n d < 1, respectively), while the V F D effect had a n F (4,148): 2.64, P < 0.05. W h e n the groups were compared to each other, the R H ~ - one was f o u n d to be significantly inferior to controls ( F (4,148): 3.07, P < 0.025) a n d to R H - - patients ( F (4,148): 2.53, P < 0-05). N o difference between the other groups reached the confidence level, n o t even that between L H q - patients a n d controls ( F (4,148): 1.07). The p a t t e r n o f results was n o t substantially different for the meaningless figure test (Table 1). Brain-damaged patients as a whole performed more poorly t h a n controls ( F (4,148): 3.00, P < 0.025), b u t this was due to the poorer scores of the two groups with
46
E. DE RENZI, P. FAGLIONIand P. VILLA
VFD, the F (4,148) of 1.99 being significant at the 0.05 level, if a one-tailed test is used. Neither the hemisphere effect, nor the interaction are significant (F < 1 in both cases). The R H + group is significantly inferior to controls at the 0.01 level (F (4,148): 3.88), while the comparison between the corresponding left-sided group and controls reaches the 5 significance level only if a one-tailed test is used (F (4,148): 2.20). No other inter-group comparison is significant. The number of trials to criterion of left brain-damaged patients did not exceed that of the R H + group even when the evaluation was restricted to the left sided patients who were diagnosed aphasic on the basis of their Token Test score. There were 23 such patients and their mean on the meaningful figure test was 14.91, and on the meaningless figure test was 20.91, both lower than the corresponding means of the R H + group. Moreover, severity of aphasia was not significantly correlated with the memory performance: the r between Token Test scores and meaningful figure test scores was -4).34 (P > 0.05 with 21 D.F.) and the r between the Token Test scores and the meaningless test scores was ---0.17 (P > 0.05 with 21 D.F.). It is also worth noting that the majority of patients unable to reach the criterion were found among R H + patients. There were 2 such patients in the meaningful figure test and both were RH-t- patients; there were 8 patients in the meaningless figure test: 4 were R H + , 2 R H - - , 1 L H + and 1 L H - - . Since an arbitrary score of 50 was given when the criterion was not attained within 50 trials, the mean performance of the R H + group is likely to be actually worse than it has been estimated. The correlation coefficients between the meaningful and the meaningless figure test were 0.50 in normals, 0.72 in left brain-damaged patients and 0.77 in right brain-damaged patients, all with a P < 0.001. Standard serial position curves of the meaningful figure test were obtained for each ArcsinV~- = 1 . 6 1 0 - 0 O 0 5 x - O . O 5 8 x 2 . O.OO6x ~
160 -150 - 140
130
.N~
--
- -
120 -.E
