- Email: [email protected]
Written Numeral Transcoding in Patients with Alzheimer's Disease
NOTE WRITTEN NUMERAL TRANS CODING IN PATIENTS WITH ALZHEIMER'S DISEASE J. Kessler and E. Kalbe (Max-Planck-Institut ftir neurologische Forschung, KOln)
ABSTRACf
Transcoding verbal numerals into Arabic numerals and vice versa was studied in 21 patients with probable Alzheimer's disease and 22 healthy subjects. Transcoding errors were analyzed according to the classification schemes by Deloche and Seron (1982a, 1982b) and Seron and Deloche (1983). In the control group, errors occurred only 2% of the time in transcoding Arabic numerals into verbal numerals, and in 1.4% for the reverse order. In AD patients, more than one error frequently occurred in the same item. In 672 transcriptions from Arabic numerals into verbal numerals 479 errors occurred in 312 items. In the reverse order 672 transcriptions were performed and 290 errors occurred in 238 items. One kind of error was striking in both transcoding directions, not reported in aphasic or other brain damaged patients: elements of the source code reappeared in the target code, e.g. 3436 -+ 3tausendvierhundert36 (3thousandfourhundred36), or zweihunderteIJ (211) -+ 2hindert11 (2hundred11). It is concluded that this error type is the result of a cognitive system which has multiple impairments and in which control mechanisms are temporarily failing.
INTRODUCTION
Number processing and number production are indispensable abilities in everyday life (McCloskey and Macaruso, 1995). Usually, numbers are described in two notation systems: Arabic numerals (e.g. 84) and verbal numerals (e.g. eighty-four), The transfer of one numeral system (source code) into another (target code) is called transcoding. Lesions of the brain can disturb this cognitive process at different levels. A number of single case (Cipolotti, 1995; Cipolotti, Warrington and Butterworth, 1995; Cohen and Dehaene, 1991; McCloskey, Sokol and Goodman, 1986; Noel and Seron, 1993) and group studies (Claros Salinas and von Cramon, 1987; Deloche and Seron, 1982a, 1982b; Seron and Deloche, 1983; Tegner and Nybiick, 1990) have described the failure of patients with varying brain damage during these tasks. To date there is, however, little information about performance in numeral transcoding tasks (Cipolotti, 1995; Noel and Seron, 1993; Tegner and Nybiick, 1990) orin calculation and number processing (Deloche, Hannequin, Carlomagno et aI., 1995) in patients with probable Alzheimer's disease (AD). Only one group study (Tegner and Nyback, 1990) investigated AD patients' ability for transcoding integers (written numbers) from ideographic (Arabic) to alphabetic script. The most characteristic error was the intrusion of numeric digits into alphabetic script. Such errors were explained by assuming that in using Arabic script AD patients were incapable of suppressing the more automatized behaviour. This error type was not found in aphasic patients or in other brain damaged subjects. The work of Tegner and Nyback, however, was a pilot study. In our study, the errors of written transcodings from Arabic into verbal numerals and vice versa, performed by patients with probable AD according to NINCDS-ADRDA criteria (McKhann, Drachman, Folstein et aI., 1984) and DSM-IV (American Psychiatric Association, 1994), were classified following the scheme proposed by Deloche and Seron Cortex, (1996) 32, 755-761
756
1. Kessler and E. Kalbe
(1982a, 1982b) and Seron and Deloche (1983). To keep memory requirements as low as possible only the written transcoding modality was chosen. MATERIALS AND METHODS
Patients 21 patients (14 male, 7 female) with probable Alzheimer's disease according to NINCDSADRDA (McKhann et aI., 1984) and DSM-IV (American Psychiatric Association, 1994) criteria were studied. Mean age was 65.4 (SD = 9.7). CT and/or MRI scans and ultrasound examination of the neck vessels were obtained to rule out ischemic or focal cerebral lesions. All patients underwent detailed neurological and psychiatric examinations. Additionally, the modified Hachinski score of Rosen, Terry, Fuld et aI. (1980) was used; only patients with a score of 3 or less were included in the study. Minimum duration of illness was 6 months. The control group consisted of 22 (12 male, 10 female) healthy persons. Their mean age was 59.0 years (SD = 6.4). In the screening procedure they did not show mnestic or cognitive deficits. Subjects with a history of alcoholism, drug abuse, learning disabilities or psychiatric illness were excluded. Education and profession were comparable to the AD group. Tests and Procedure Neuropsychological Test Battery The cognitive and mnestic status of the patients was determined by neuropsychological testing. Briefly, the following tests were used: A German version (Kessler, Markowitsch and Denzler, 1990) of the Mini-Mental-State-Examination (Folstein, Folstein and McHugh, 1975), a verbal selective reminding task with delayed recall after 15 minutes (Schaaf, Kessler, Grond et aI., 1992), Corsi's block span test (Milner, 1971) administered to the instructions and sequences of Smirni, Villardita and Zappala (1983), the Rey-Osterrieth figure (form A) (Spreen and Strauss, 1991) with delayed production after 30 minutes and use of Taylor's scoring criteria (1989), subtest 4 of the Token-test which is part of the Aachener-Aphasie-Test (AAT) (Huber, Poeck, Weniger et aI., 1983), a number of fluency tasks, and a German version of the Stroop colour word test (Baumler, 1985). Writing and reading abilities and reading comprehension were examined with subtests of the AAT (Huber et aI., 1983). For testing of number processing and calculation, the German modification (Claros Salinas, 1994) of the acalculia test battery of Deloche, Seron, Metz-Lutz et aI. (1993) was used. Number processing tasks were number recognition, number reading, number writing to dictation, magnitude comparison of two numbers, and number positioning on an analogous scale. All number processing tasks included Arabic numerals with 1, 2, 3, and 4 digits and a parallel list of verbal numerals. Transcoding Tasks Two written transcoding tasks were applied: one requested transcoding from Arabic to verbal numerals, and the other requested transcoding from verbal to Arabic numerals. The first task consisted of 32 items, including eight I-digit, eight 2-digit, eight 3-digit, and eight 4-digit numbers. For the other transcoding tasks, a parallel list was used. Items differed in complexity. The items of each list were first randomized and then presented in a fixed order throughout. The patients always started with transcoding Arabic into verbal numerals. Before each task, subjects were given for practice 5 items that were not included in the error analysis. No feedback or any form of correction was given during the tasks. Description of Errors Error types were classified following the suggestions of Deloche and Seron (1982a, 1982b) and Seron and Deloche (1983). According to their model of the number lexicon, the elements
Transcoding in Alzheimer's Disease
757
are organized in stack structures: a unit stack (numbers 1-9), a particular stack (in English, numbers 11-19), a tens stack (numbers 10, 20, 30, ... ,) and a stack called miscellaneous (hundred, thousand etc.). Within each stack, the elements are placed in their customary order, and thus each element of the lexicon is characterized by the stack it belongs to and its position within the stack. In lexical access, both the stack and the position within the stack must be coded correctly. On this theoretical basis, Seron and Deloche classify stack errors which occur when the wrong stack of numerals was chosen while the information regarding the position within the stack was coded correctly, and stack position errors that occur when the position within the stack of the number is coded incorrectly while the stack information is preserved. Furthermore. they described partial processing errors when one ore more elements of the stimulus ·were not included in the transcoding, and term-by-term-strategies when patients subdivided the stimulus into its elements and transcoded them separately. A peculiarity of the German numeral system led to another type of error. Multiples of ten in German are constructed by inverting the unit stack and the tens stack element form ("21" corresponds to "one and twenty"). Some patients produced inversion errors, neglecting the inversion rule. and some omitted the syntactical element 'and'. Furthermore, many orthographic errors were produced. Multiple errors, i.e. a combination of the errors described above, were also often observed. Some errors were considered as non-classifiable. One error type that was not described by Deloche and Seron (1982a, b) and Seron and Deloche (1983) was the shift error. Shift errors occurred when the source code was repeated or partly repeated. as in 3436 -> 3436. zweihunderte/f (211) -> zweihunderte/f (211), 4054 -> vier054 (four054), zweitausend (2000) -> 2tausend (2thousand). A more detailled description of these errors will be given in the results section. RESULTS
The MMST-scores of AD patients had a mean of 20.71 (SD=4.30, range 12-27). In comparison with the reference scores AD patients showed significant impairment in memory scores, especially in the delayed recall tasks, constructive abilities, and word production rate. All patients had essentially preserved abilities to read and write. The group was heterogenous concerning number processing abilities. For all patients, number processing with Arabic numbers was easier than with verbal numerals. All patients managed the number reading and number recognition tasks. Writing Arabic numerals to dictation was impaired in 18% of patients, but none of them found writing impossible. 75 % of patients made errors in writing verbal numbers to dictation, especially complex verbal numerals. Magnitude comparison of Arabic numerals was mostly unimpaired, whereas comparison of verbal numbers was difficult for 67% of patients, especially when smaller numbers consisted of more words than the larger ones (i.e. four-thousand vs. three-hundred sixty-two). With the exception of 2 patients, all had severe difficulties in number positioning on an analogous scale task. Basic calculation abilities ranged from unimpaired to severely impaired. Transcoding Results Control Group
The 22 subjects trans coded 704 Arabic numerals into verbal ones (22 subjects X 32 items); I shift error and 13 orthographic errors occurred (error rate = 2%). The transcoding of 704 verbal into Arabic numerals resulted in 4 stack position errors, 2 partial processing errors, 3 term-by-term errors, 1 inversion error and 1 shift error (error rate = 1.6%). No multiple errors were observed. Though the control group was slightly younger than the AD group, it is unlikely that the small age difference played a role in the performance. AD Group
Altogether the 21 AD patients carried out 1344 transcodings in both directions (21 patients X 64 items). Multiple errors occurred in almost every single transcoding act. Not a single item was transformed correctly by all patients.
758
1. Kessler and E. Kalbe
Arabic Numerals into Verbal Numerals: In 672 transfonnations, 479 errors occurred in 312 items (error rate =46%, multiple error rate =71%). They were classified as follows: 112 shift errors, 44 shift errors with self-corrections, 7 stack errors, 19 stack position errors, 35 partial processing errors, 47 tenn-by-tenn errors, 8 inversion errors, 11 omissions of the syntactical element 'and', and 164 orthographic errors. 32 errors were non classificable. Severity of dementia expressed by the MMST scores correlated significantly with stack position errors (r = - 0.56; p
The high error rate in both transcoding directions indicates that this process is significantly impaired in AD patients. 315 errors occurred (orthographic errors excluded) in transcoding
Transcoding in Alzheimer's Disease
759
from Arabic into verbal numerals, and 290 errors occurred in the reverse order in the 672 transcodings. Degree of dementia expressed in MMST -scores, errors in the Token test, and Stroop-interference-scores were significantly correlated with the total sum of errors in both transcoding directions. It was not the aim of this study to interpret the results found here in the light of the various semantic (McCloskey, Caramazza and Basili, 1985), asemantic (Seron and Deloche, 1987), and semantic and asemantic models (Dehaene, 1992; Campbell and Clark, 1988, 1992) of number processing and transcoding. Only very detailed and extensive single case studies would permit an evaluation of these cognitive models. It is difficult to say what are the mechanisms underlying these errors. Deloche and Seron (1987) postulated that interactions between the several components of the transcoding process were monitored by a task supervisor or master scheduler whose function would be to solve incompatibilities or to look for new strategies. Correspondingly, Jorm (1986) argued that controlled processes which require the attentional resources of the individual (e.g. transcoding) are disturbed in the early stages of AD whereas automatic processes are still preserved. Shallice (1988) postulated a supervisory attentional system which would not directly control behaviour but modulate lower level systems directly involved in action and thought operation. Its primary function is assumed to be the production of non-routine responses that must be planned in advance. Its anatomical basis is thought to be in the frontal lobes whose damage leads to typical symptoms such as perseveration, disinhibition or inability to shift behaviour. According to Sandson and Albert (1987), perseveration is not an unitary phenomenon but consists of three types, each with its own (specific) neuroanatomic basis. In their terminology the inappropriate maintenance of a category or framework of activity is called stuck-in-setperseveration. Shift errors correspond to this form of perseveration, in contrast to recurring perseveration or continuous perseveration. Such behaviour is closely related to frontal lobe dysfunction. PET studies with 18-FDG have shown that in AD patients the typical metabolic impairment occurs in the temporoparietal and frontal association cortex and relatively spares primary sensorimotor and visual cortex, striatum and cerebellum (Herholz, 1995; Herholz, Adams, Kessler et aI., 1990). It is plausible that the frontal hypometabolism in combination with lesions of the heteromodal association cortices leads to an impairment of the supervisory or attentional system and results in shift errors although present data are inadequate to explain at which level of information processing the transcoding erros occur. Tegner and Nyback (1990) suggested trans coding tasks as simple bedside tests for dementia. We concur that transcoding errors are absent in a healthy control group and frequent in AD patients. Whether they are specific to this dementia group remains to be determined. Acknowledgements. We would like to thank Dr. K. Willmes from the RTHC Aachen and M. Dom from the Albert-Ludwig-University of Freiburg i. Br. for helpful comments on an earlier draft of this paper and H. Bonner for statistical assistance.
REFERENCES AMERICAN PSYCHIATRIC ASSOCIATION. Diagnostic and Statistical Manual of Mental Disorders, 4th revised edition. Washington DC, 1994. BAUMLER, G . Farbe-Wort-1nterferenztest nach Stroop (FW1TJ. Gottingen : Hogrefe Verlag, 1985. CAMPBELL, 1.1.D., and CLARK, I.M. An encoding complex view of cognitive number processing: Comment on McCloskey, Sokol and Goodman. Journal of Experimental Psychology, 117: 204214, 1988. CAMPBELL, 1.1.D., and CLARK, I .M. Cognitive number processing: An encoding-complex perspective. In 1.I.D. Campbell (Ed.), The Nature and Origins of Mathematical Skills. Amsterdam: North Holland, 1992, pp. 475-491. CIPOLOTII, L. Multiple routes for reading words, why not numbers ? Evidence from a case of Arabic numerals dyslexia. Cognitive Neuropsychology, 12: 313-342, 1995. CIPOLOTII, L., WARRINGTON, E.K. , and BUTTERWORTH, B. Selective impairment in manipulating Arabic numerals. Cortex, 31: 73-86, 1995. CLAROS SALINAS, D. EC 301 R: Untersuchungsmaterial zu Storungen des Rechnens und der
760
J. Kessler and E. Kalbe Zahlenverarbeitung. Deutsche Adaption von Deloche et aI., 1993, The EC301 assessment battery
for brain-damaged adults. Konstanz: Kliniken Schmieder Gailingen, 1994. CLAROS SALINAS, D., and VON CRAMON, D. Diagnostik von Stiirungen im Umgang mit Zahlen (Akalkulie). Fortschritte der Neurologie und Psychiatrie, 55: 239-248, 1987. COHEN, L., and DEHAENE, S. Neglect dyslexia for numbers? A case report. Cognitive Neuropsychology, 8: 39-58, 1991. DEHAENE, S. Varieties of numerical abilities. Cognition, 44: 1-42, 1992. DELocHE, G., HANNEQUIN, D., CARLOMAGNO, S., AGNIEL, A., DORDAIN, M., PASQUIER, F., PELLAT, 1., DENIS, P., DESI, M., BEAUCHAMP, D., METz-LuTZ, M.N., CESARO, P., and SERON, X. Calculation and number processing in mild Alzheimer's disease. Journal of Clinical and Experimental Neuropsychology, 17: 634-639, 1995. DELocHE, G., and SERON, X. From one to I: an analysis of transcoding process by means of neuropsychological data. Cognition, 12: 119-149, 1982a. DELocHE, G., and SERON, X. From three to 3: a differential analysis of skills in transcoding quantities between patients with Broca's and Wernicke's aphasia. Brain, 105: 719-733, 1982b. DELocHE, G., and SERON, X. Numerical transcoding: A general production model. In G. Deloche and X. Seron (Eds.), Mathematical Disabilities - A Cognitive Neuropsychological Perspective. London: Lawrence Erlbaum Associates, Publisher, 1987, Ch. 7, pp. 137-l70. DELocHE, G., SERON, X., METZ-LuTZ, M.N., BAETA, E., BASSO, A., CLAROS SALINAS, D., GAILLARD, F., GOLDENBERG, G., STACHOWIAK, F., TEMPLE, C., TZAVARAS, A., and VENDRELL, J. Calculation and number processing: the EC301 assessment battery for brain-damaged adults. In F. Stachowiak (Ed.), Developments in the Assessment and Rehabilitation of Brain-Damaged Patients. Tiibingen: Narr-Verlag, 1993. FOLSTEIN, M.F., FOLSTEIN, S.E., and McHUGH, P.R. "Mini-Mental-State" - A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12: 189198, 1975. HERHoLz, K. FOG-PET and differential diagnosis of dementia. Alzheimer's Disease and Associated Disorders, 9: 6-16, 1995. HERHOLZ, K., ADAMS, A., KESSLER, 1., GROND, M., and HEISS, W.O. Criteria for the diagnosis of Alzheimer's disease with positron emission tomography. Dementia, 1: 156-164, 1990. HUBER, W., POECK, K., WENIGER, D., and WILLMES, K. Der Aachener Aphasie Test. Gottingen: Hogrefe Verlag, 1983. JORM, A.F. Controlled and automatic information processing in senile dementia: a review. Psychological Medicine, 16: 77-88, 1986. KESSLER, 1., MARKOWITSCH, H.J., and DENZLER, P. MMST Mini Mental Status Test. Weinheim: Beltz Verlag, 1990. McCLOSKEY, M., CARAMAZZA, A., and BASILI, A. Cognitive mechanisms in number processing and calculation: evidence from dyscalculia. Brain and Cognition, 4: l71-196, 1985. MCCLOSKEY, M., and MACARUSO, P. Representing and using numerical information. American Psychgologist, 50: 351-363, 1995. McCLOSKEY, M., SOKOL, S.M., and GOODMAN, R.A. Cognitive processes in verbal-number production: inferences from the performance of brain-damaged subjects. Journal of Experimental Psychology, 115: 307-330, 1986. McKHANN, G., DRACHMAN, D., FOLSTEIN, M.F., KATZMAN, R., PRICE, D., and STADLAN, E.M. Clinical diagnosis of Alzheimer's disease: Report of the NINCDS-ADRDA work group under the auspices of department of health and human services task force on Alzheimer's disease. Neurology, 35: 939-944, 1984. MILNER, B. Interhemispheric differences and psychological processes. British Medical Bulletin, 27: 273-277, 1971. NOEL, M.P., and SERON, X. Arabic number reading deficit: A single case study or when 236 is read (2306) and judged superior to 1258. Cognitive Neuropsychology, 10: 3l7-339, 1993. ROSEN, W.G., TERRY, R.D., FULD, P.A., KATZMANN, R., and PECK, A. Pathological verification of ischemic score in differentiation of dementi as. Annals of Neurology, 7: 486-488, 1980. SANDSON, J., and ALBERT, M.L. Perseveration in behavioral neurology. Neurology, 37: 1736-l741, 1987. SCHAAF, A., KESSLER, J., GROND, M., and FINK, G. Memo Test. Ein verbaler Gediichtnistest nach der Methode des selektiven Erinnerns. Weinheim: Beltz-Test-Verlag, 1992. SERON, X., and DELocHE, G. From 4 to four: A supplement to 'From three to 3'. Brain, 106: 732744, 1983. SHALLICE, T. From Neuropsychology to Mental Structure. Cambridge: Cambridge University Press, 1988. SMIRNI, P., VILLARDITA, c., and ZAPPALA, G. Influence of different paths on spatial memory performance in the Block-Tapping Test. Journal of Clinical Neuropsychology,S: 355-359, 1983.
Transcoding in Alzheimer's Disease
761
SPREEN, 0" and STRAUSS, E, A Compendium of Neuropsychological Tests, New York, Oxford: Oxford University Press, 1991, Ch. 6, pp. 157-167. TAYLOR, L.B. (1989) Scoring criteria for the Rey Complex Figure. In O. Spreen and E. Strauss (Eds.), A Compendium of Neuropsychological Tests. New York, Oxford: Oxford University Press, 1991, Ch. 6, pp. 161-164. TEGNER, R., and NYBAcK, H. To hundred and twenty4our': A study of transcoding in dementia. Acta Neurologica Scandinavica, 81: 177-178, 1990. PO Dr. 1. Kessler. Max-Planck-Institut fUr neurologische Forschung, Gleueler Str. 50. D-50931 K61n. Gennany.
ABSTRACf
Transcoding verbal numerals into Arabic numerals and vice versa was studied in 21 patients with probable Alzheimer's disease and 22 healthy subjects. Transcoding errors were analyzed according to the classification schemes by Deloche and Seron (1982a, 1982b) and Seron and Deloche (1983). In the control group, errors occurred only 2% of the time in transcoding Arabic numerals into verbal numerals, and in 1.4% for the reverse order. In AD patients, more than one error frequently occurred in the same item. In 672 transcriptions from Arabic numerals into verbal numerals 479 errors occurred in 312 items. In the reverse order 672 transcriptions were performed and 290 errors occurred in 238 items. One kind of error was striking in both transcoding directions, not reported in aphasic or other brain damaged patients: elements of the source code reappeared in the target code, e.g. 3436 -+ 3tausendvierhundert36 (3thousandfourhundred36), or zweihunderteIJ (211) -+ 2hindert11 (2hundred11). It is concluded that this error type is the result of a cognitive system which has multiple impairments and in which control mechanisms are temporarily failing.
INTRODUCTION
Number processing and number production are indispensable abilities in everyday life (McCloskey and Macaruso, 1995). Usually, numbers are described in two notation systems: Arabic numerals (e.g. 84) and verbal numerals (e.g. eighty-four), The transfer of one numeral system (source code) into another (target code) is called transcoding. Lesions of the brain can disturb this cognitive process at different levels. A number of single case (Cipolotti, 1995; Cipolotti, Warrington and Butterworth, 1995; Cohen and Dehaene, 1991; McCloskey, Sokol and Goodman, 1986; Noel and Seron, 1993) and group studies (Claros Salinas and von Cramon, 1987; Deloche and Seron, 1982a, 1982b; Seron and Deloche, 1983; Tegner and Nybiick, 1990) have described the failure of patients with varying brain damage during these tasks. To date there is, however, little information about performance in numeral transcoding tasks (Cipolotti, 1995; Noel and Seron, 1993; Tegner and Nybiick, 1990) orin calculation and number processing (Deloche, Hannequin, Carlomagno et aI., 1995) in patients with probable Alzheimer's disease (AD). Only one group study (Tegner and Nyback, 1990) investigated AD patients' ability for transcoding integers (written numbers) from ideographic (Arabic) to alphabetic script. The most characteristic error was the intrusion of numeric digits into alphabetic script. Such errors were explained by assuming that in using Arabic script AD patients were incapable of suppressing the more automatized behaviour. This error type was not found in aphasic patients or in other brain damaged subjects. The work of Tegner and Nyback, however, was a pilot study. In our study, the errors of written transcodings from Arabic into verbal numerals and vice versa, performed by patients with probable AD according to NINCDS-ADRDA criteria (McKhann, Drachman, Folstein et aI., 1984) and DSM-IV (American Psychiatric Association, 1994), were classified following the scheme proposed by Deloche and Seron Cortex, (1996) 32, 755-761
756
1. Kessler and E. Kalbe
(1982a, 1982b) and Seron and Deloche (1983). To keep memory requirements as low as possible only the written transcoding modality was chosen. MATERIALS AND METHODS
Patients 21 patients (14 male, 7 female) with probable Alzheimer's disease according to NINCDSADRDA (McKhann et aI., 1984) and DSM-IV (American Psychiatric Association, 1994) criteria were studied. Mean age was 65.4 (SD = 9.7). CT and/or MRI scans and ultrasound examination of the neck vessels were obtained to rule out ischemic or focal cerebral lesions. All patients underwent detailed neurological and psychiatric examinations. Additionally, the modified Hachinski score of Rosen, Terry, Fuld et aI. (1980) was used; only patients with a score of 3 or less were included in the study. Minimum duration of illness was 6 months. The control group consisted of 22 (12 male, 10 female) healthy persons. Their mean age was 59.0 years (SD = 6.4). In the screening procedure they did not show mnestic or cognitive deficits. Subjects with a history of alcoholism, drug abuse, learning disabilities or psychiatric illness were excluded. Education and profession were comparable to the AD group. Tests and Procedure Neuropsychological Test Battery The cognitive and mnestic status of the patients was determined by neuropsychological testing. Briefly, the following tests were used: A German version (Kessler, Markowitsch and Denzler, 1990) of the Mini-Mental-State-Examination (Folstein, Folstein and McHugh, 1975), a verbal selective reminding task with delayed recall after 15 minutes (Schaaf, Kessler, Grond et aI., 1992), Corsi's block span test (Milner, 1971) administered to the instructions and sequences of Smirni, Villardita and Zappala (1983), the Rey-Osterrieth figure (form A) (Spreen and Strauss, 1991) with delayed production after 30 minutes and use of Taylor's scoring criteria (1989), subtest 4 of the Token-test which is part of the Aachener-Aphasie-Test (AAT) (Huber, Poeck, Weniger et aI., 1983), a number of fluency tasks, and a German version of the Stroop colour word test (Baumler, 1985). Writing and reading abilities and reading comprehension were examined with subtests of the AAT (Huber et aI., 1983). For testing of number processing and calculation, the German modification (Claros Salinas, 1994) of the acalculia test battery of Deloche, Seron, Metz-Lutz et aI. (1993) was used. Number processing tasks were number recognition, number reading, number writing to dictation, magnitude comparison of two numbers, and number positioning on an analogous scale. All number processing tasks included Arabic numerals with 1, 2, 3, and 4 digits and a parallel list of verbal numerals. Transcoding Tasks Two written transcoding tasks were applied: one requested transcoding from Arabic to verbal numerals, and the other requested transcoding from verbal to Arabic numerals. The first task consisted of 32 items, including eight I-digit, eight 2-digit, eight 3-digit, and eight 4-digit numbers. For the other transcoding tasks, a parallel list was used. Items differed in complexity. The items of each list were first randomized and then presented in a fixed order throughout. The patients always started with transcoding Arabic into verbal numerals. Before each task, subjects were given for practice 5 items that were not included in the error analysis. No feedback or any form of correction was given during the tasks. Description of Errors Error types were classified following the suggestions of Deloche and Seron (1982a, 1982b) and Seron and Deloche (1983). According to their model of the number lexicon, the elements
Transcoding in Alzheimer's Disease
757
are organized in stack structures: a unit stack (numbers 1-9), a particular stack (in English, numbers 11-19), a tens stack (numbers 10, 20, 30, ... ,) and a stack called miscellaneous (hundred, thousand etc.). Within each stack, the elements are placed in their customary order, and thus each element of the lexicon is characterized by the stack it belongs to and its position within the stack. In lexical access, both the stack and the position within the stack must be coded correctly. On this theoretical basis, Seron and Deloche classify stack errors which occur when the wrong stack of numerals was chosen while the information regarding the position within the stack was coded correctly, and stack position errors that occur when the position within the stack of the number is coded incorrectly while the stack information is preserved. Furthermore. they described partial processing errors when one ore more elements of the stimulus ·were not included in the transcoding, and term-by-term-strategies when patients subdivided the stimulus into its elements and transcoded them separately. A peculiarity of the German numeral system led to another type of error. Multiples of ten in German are constructed by inverting the unit stack and the tens stack element form ("21" corresponds to "one and twenty"). Some patients produced inversion errors, neglecting the inversion rule. and some omitted the syntactical element 'and'. Furthermore, many orthographic errors were produced. Multiple errors, i.e. a combination of the errors described above, were also often observed. Some errors were considered as non-classifiable. One error type that was not described by Deloche and Seron (1982a, b) and Seron and Deloche (1983) was the shift error. Shift errors occurred when the source code was repeated or partly repeated. as in 3436 -> 3436. zweihunderte/f (211) -> zweihunderte/f (211), 4054 -> vier054 (four054), zweitausend (2000) -> 2tausend (2thousand). A more detailled description of these errors will be given in the results section. RESULTS
The MMST-scores of AD patients had a mean of 20.71 (SD=4.30, range 12-27). In comparison with the reference scores AD patients showed significant impairment in memory scores, especially in the delayed recall tasks, constructive abilities, and word production rate. All patients had essentially preserved abilities to read and write. The group was heterogenous concerning number processing abilities. For all patients, number processing with Arabic numbers was easier than with verbal numerals. All patients managed the number reading and number recognition tasks. Writing Arabic numerals to dictation was impaired in 18% of patients, but none of them found writing impossible. 75 % of patients made errors in writing verbal numbers to dictation, especially complex verbal numerals. Magnitude comparison of Arabic numerals was mostly unimpaired, whereas comparison of verbal numbers was difficult for 67% of patients, especially when smaller numbers consisted of more words than the larger ones (i.e. four-thousand vs. three-hundred sixty-two). With the exception of 2 patients, all had severe difficulties in number positioning on an analogous scale task. Basic calculation abilities ranged from unimpaired to severely impaired. Transcoding Results Control Group
The 22 subjects trans coded 704 Arabic numerals into verbal ones (22 subjects X 32 items); I shift error and 13 orthographic errors occurred (error rate = 2%). The transcoding of 704 verbal into Arabic numerals resulted in 4 stack position errors, 2 partial processing errors, 3 term-by-term errors, 1 inversion error and 1 shift error (error rate = 1.6%). No multiple errors were observed. Though the control group was slightly younger than the AD group, it is unlikely that the small age difference played a role in the performance. AD Group
Altogether the 21 AD patients carried out 1344 transcodings in both directions (21 patients X 64 items). Multiple errors occurred in almost every single transcoding act. Not a single item was transformed correctly by all patients.
758
1. Kessler and E. Kalbe
Arabic Numerals into Verbal Numerals: In 672 transfonnations, 479 errors occurred in 312 items (error rate =46%, multiple error rate =71%). They were classified as follows: 112 shift errors, 44 shift errors with self-corrections, 7 stack errors, 19 stack position errors, 35 partial processing errors, 47 tenn-by-tenn errors, 8 inversion errors, 11 omissions of the syntactical element 'and', and 164 orthographic errors. 32 errors were non classificable. Severity of dementia expressed by the MMST scores correlated significantly with stack position errors (r = - 0.56; p
The high error rate in both transcoding directions indicates that this process is significantly impaired in AD patients. 315 errors occurred (orthographic errors excluded) in transcoding
Transcoding in Alzheimer's Disease
759
from Arabic into verbal numerals, and 290 errors occurred in the reverse order in the 672 transcodings. Degree of dementia expressed in MMST -scores, errors in the Token test, and Stroop-interference-scores were significantly correlated with the total sum of errors in both transcoding directions. It was not the aim of this study to interpret the results found here in the light of the various semantic (McCloskey, Caramazza and Basili, 1985), asemantic (Seron and Deloche, 1987), and semantic and asemantic models (Dehaene, 1992; Campbell and Clark, 1988, 1992) of number processing and transcoding. Only very detailed and extensive single case studies would permit an evaluation of these cognitive models. It is difficult to say what are the mechanisms underlying these errors. Deloche and Seron (1987) postulated that interactions between the several components of the transcoding process were monitored by a task supervisor or master scheduler whose function would be to solve incompatibilities or to look for new strategies. Correspondingly, Jorm (1986) argued that controlled processes which require the attentional resources of the individual (e.g. transcoding) are disturbed in the early stages of AD whereas automatic processes are still preserved. Shallice (1988) postulated a supervisory attentional system which would not directly control behaviour but modulate lower level systems directly involved in action and thought operation. Its primary function is assumed to be the production of non-routine responses that must be planned in advance. Its anatomical basis is thought to be in the frontal lobes whose damage leads to typical symptoms such as perseveration, disinhibition or inability to shift behaviour. According to Sandson and Albert (1987), perseveration is not an unitary phenomenon but consists of three types, each with its own (specific) neuroanatomic basis. In their terminology the inappropriate maintenance of a category or framework of activity is called stuck-in-setperseveration. Shift errors correspond to this form of perseveration, in contrast to recurring perseveration or continuous perseveration. Such behaviour is closely related to frontal lobe dysfunction. PET studies with 18-FDG have shown that in AD patients the typical metabolic impairment occurs in the temporoparietal and frontal association cortex and relatively spares primary sensorimotor and visual cortex, striatum and cerebellum (Herholz, 1995; Herholz, Adams, Kessler et aI., 1990). It is plausible that the frontal hypometabolism in combination with lesions of the heteromodal association cortices leads to an impairment of the supervisory or attentional system and results in shift errors although present data are inadequate to explain at which level of information processing the transcoding erros occur. Tegner and Nyback (1990) suggested trans coding tasks as simple bedside tests for dementia. We concur that transcoding errors are absent in a healthy control group and frequent in AD patients. Whether they are specific to this dementia group remains to be determined. Acknowledgements. We would like to thank Dr. K. Willmes from the RTHC Aachen and M. Dom from the Albert-Ludwig-University of Freiburg i. Br. for helpful comments on an earlier draft of this paper and H. Bonner for statistical assistance.
REFERENCES AMERICAN PSYCHIATRIC ASSOCIATION. Diagnostic and Statistical Manual of Mental Disorders, 4th revised edition. Washington DC, 1994. BAUMLER, G . Farbe-Wort-1nterferenztest nach Stroop (FW1TJ. Gottingen : Hogrefe Verlag, 1985. CAMPBELL, 1.1.D., and CLARK, I.M. An encoding complex view of cognitive number processing: Comment on McCloskey, Sokol and Goodman. Journal of Experimental Psychology, 117: 204214, 1988. CAMPBELL, 1.1.D., and CLARK, I .M. Cognitive number processing: An encoding-complex perspective. In 1.I.D. Campbell (Ed.), The Nature and Origins of Mathematical Skills. Amsterdam: North Holland, 1992, pp. 475-491. CIPOLOTII, L. Multiple routes for reading words, why not numbers ? Evidence from a case of Arabic numerals dyslexia. Cognitive Neuropsychology, 12: 313-342, 1995. CIPOLOTII, L., WARRINGTON, E.K. , and BUTTERWORTH, B. Selective impairment in manipulating Arabic numerals. Cortex, 31: 73-86, 1995. CLAROS SALINAS, D. EC 301 R: Untersuchungsmaterial zu Storungen des Rechnens und der
760
J. Kessler and E. Kalbe Zahlenverarbeitung. Deutsche Adaption von Deloche et aI., 1993, The EC301 assessment battery
for brain-damaged adults. Konstanz: Kliniken Schmieder Gailingen, 1994. CLAROS SALINAS, D., and VON CRAMON, D. Diagnostik von Stiirungen im Umgang mit Zahlen (Akalkulie). Fortschritte der Neurologie und Psychiatrie, 55: 239-248, 1987. COHEN, L., and DEHAENE, S. Neglect dyslexia for numbers? A case report. Cognitive Neuropsychology, 8: 39-58, 1991. DEHAENE, S. Varieties of numerical abilities. Cognition, 44: 1-42, 1992. DELocHE, G., HANNEQUIN, D., CARLOMAGNO, S., AGNIEL, A., DORDAIN, M., PASQUIER, F., PELLAT, 1., DENIS, P., DESI, M., BEAUCHAMP, D., METz-LuTZ, M.N., CESARO, P., and SERON, X. Calculation and number processing in mild Alzheimer's disease. Journal of Clinical and Experimental Neuropsychology, 17: 634-639, 1995. DELocHE, G., and SERON, X. From one to I: an analysis of transcoding process by means of neuropsychological data. Cognition, 12: 119-149, 1982a. DELocHE, G., and SERON, X. From three to 3: a differential analysis of skills in transcoding quantities between patients with Broca's and Wernicke's aphasia. Brain, 105: 719-733, 1982b. DELocHE, G., and SERON, X. Numerical transcoding: A general production model. In G. Deloche and X. Seron (Eds.), Mathematical Disabilities - A Cognitive Neuropsychological Perspective. London: Lawrence Erlbaum Associates, Publisher, 1987, Ch. 7, pp. 137-l70. DELocHE, G., SERON, X., METZ-LuTZ, M.N., BAETA, E., BASSO, A., CLAROS SALINAS, D., GAILLARD, F., GOLDENBERG, G., STACHOWIAK, F., TEMPLE, C., TZAVARAS, A., and VENDRELL, J. Calculation and number processing: the EC301 assessment battery for brain-damaged adults. In F. Stachowiak (Ed.), Developments in the Assessment and Rehabilitation of Brain-Damaged Patients. Tiibingen: Narr-Verlag, 1993. FOLSTEIN, M.F., FOLSTEIN, S.E., and McHUGH, P.R. "Mini-Mental-State" - A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12: 189198, 1975. HERHoLz, K. FOG-PET and differential diagnosis of dementia. Alzheimer's Disease and Associated Disorders, 9: 6-16, 1995. HERHOLZ, K., ADAMS, A., KESSLER, 1., GROND, M., and HEISS, W.O. Criteria for the diagnosis of Alzheimer's disease with positron emission tomography. Dementia, 1: 156-164, 1990. HUBER, W., POECK, K., WENIGER, D., and WILLMES, K. Der Aachener Aphasie Test. Gottingen: Hogrefe Verlag, 1983. JORM, A.F. Controlled and automatic information processing in senile dementia: a review. Psychological Medicine, 16: 77-88, 1986. KESSLER, 1., MARKOWITSCH, H.J., and DENZLER, P. MMST Mini Mental Status Test. Weinheim: Beltz Verlag, 1990. McCLOSKEY, M., CARAMAZZA, A., and BASILI, A. Cognitive mechanisms in number processing and calculation: evidence from dyscalculia. Brain and Cognition, 4: l71-196, 1985. MCCLOSKEY, M., and MACARUSO, P. Representing and using numerical information. American Psychgologist, 50: 351-363, 1995. McCLOSKEY, M., SOKOL, S.M., and GOODMAN, R.A. Cognitive processes in verbal-number production: inferences from the performance of brain-damaged subjects. Journal of Experimental Psychology, 115: 307-330, 1986. McKHANN, G., DRACHMAN, D., FOLSTEIN, M.F., KATZMAN, R., PRICE, D., and STADLAN, E.M. Clinical diagnosis of Alzheimer's disease: Report of the NINCDS-ADRDA work group under the auspices of department of health and human services task force on Alzheimer's disease. Neurology, 35: 939-944, 1984. MILNER, B. Interhemispheric differences and psychological processes. British Medical Bulletin, 27: 273-277, 1971. NOEL, M.P., and SERON, X. Arabic number reading deficit: A single case study or when 236 is read (2306) and judged superior to 1258. Cognitive Neuropsychology, 10: 3l7-339, 1993. ROSEN, W.G., TERRY, R.D., FULD, P.A., KATZMANN, R., and PECK, A. Pathological verification of ischemic score in differentiation of dementi as. Annals of Neurology, 7: 486-488, 1980. SANDSON, J., and ALBERT, M.L. Perseveration in behavioral neurology. Neurology, 37: 1736-l741, 1987. SCHAAF, A., KESSLER, J., GROND, M., and FINK, G. Memo Test. Ein verbaler Gediichtnistest nach der Methode des selektiven Erinnerns. Weinheim: Beltz-Test-Verlag, 1992. SERON, X., and DELocHE, G. From 4 to four: A supplement to 'From three to 3'. Brain, 106: 732744, 1983. SHALLICE, T. From Neuropsychology to Mental Structure. Cambridge: Cambridge University Press, 1988. SMIRNI, P., VILLARDITA, c., and ZAPPALA, G. Influence of different paths on spatial memory performance in the Block-Tapping Test. Journal of Clinical Neuropsychology,S: 355-359, 1983.
Transcoding in Alzheimer's Disease
761
SPREEN, 0" and STRAUSS, E, A Compendium of Neuropsychological Tests, New York, Oxford: Oxford University Press, 1991, Ch. 6, pp. 157-167. TAYLOR, L.B. (1989) Scoring criteria for the Rey Complex Figure. In O. Spreen and E. Strauss (Eds.), A Compendium of Neuropsychological Tests. New York, Oxford: Oxford University Press, 1991, Ch. 6, pp. 161-164. TEGNER, R., and NYBAcK, H. To hundred and twenty4our': A study of transcoding in dementia. Acta Neurologica Scandinavica, 81: 177-178, 1990. PO Dr. 1. Kessler. Max-Planck-Institut fUr neurologische Forschung, Gleueler Str. 50. D-50931 K61n. Gennany.