DOI 10.1007/s00702-002-0806-4 J Neural Transm (2003) 110: 537–543
Preserved memory traces within diencephalic amnesia P. Walla1, J. P. Lehrner1, C. Nasel2, C. Baumgartner1, L. Deecke1, and W. Lang1 2
1 Department of Clinical Neurology, and Department of Clinical Radiodiagnostics, University of Vienna, Vienna, Austria
Received August 1, 2002; accepted December 3, 2002 Published online February 19, 2003; © Springer-Verlag 2003
Summary. A male patient with bilateral thalamic lesions (medio-ventral nuclei) was investigated. Despite explicit memory impairments his lexical ability was normal. We recorded magnetic field changes (magnetoencephalography, MEG) during the performance of an animate/inanimate discrimination task in which some words where repeated after long delays. Normally, repeated items are classified significantly faster than their first presentations which is accomplished by an unconscious process called priming. The patient did not show any behavioural evidence of priming but the physiological data indicated preservation of this robust form of memory. Brain activation associated with repetitions was attenuated at early stages. The activity difference was posteriorly distributed which is consistent with previous reports about repetition priming. The present study indicated that the bilateral thalamic lesions of our patient disconnected the information processing stream between the primed information and the behavioural response. Keywords: Thalamic lesion, amnesia, repetition priming, magnetoencephalography. Introduction
Priming represents a well known and robust form of implicit memory (e.g. Schacter, 1987; Schacter and Buckner, 1998). From the psychological point of view it is an unconscious process leading to faster information processing of repeated items. The comparison with explicit memory processes is a major topic within human brain research. It has been shown that these two forms of memory are supported by different neural structures (Rugg et al., 1998). In healthy subjects it is difficult to investigate implicit memory processes like priming alone because the contribution of explicit memory processes is hard to control. To investigate unconscious information processing amnesic patients who have known explicit memory deficits yield the most reliable results.
538
P. Walla et al.
In most of the cases in which memory deficits of amnesic patients have been reported their performance on priming tests usually was normal (Hamann and Squire, 1997; Fleischmann et al., 1998). We investigated a patient suffering from bilateral medio-ventral thalamic lesions after a stroke. The goal of the present study was to characterise the functional significance of the malfunctioning thalamic regions within repetition priming in a lexical decision task. During the performance of the lexical decision task, magnetic field changes were recorded with a whole-head magnetoencephalograph (MEG). Response times and magnetic field distributions associated with first and second presentations were analysed. Materials and methods Subject (patient) The patient was a 57-year-old man, right handed (Salmaso and Longoni, 1985) and with an unremarkable medical history. He was found unresponsive by family members. On admission the patient was in a comatose state. General examination revealed only a mild pneumonia. CT of the head and CSF examination were normal. During the next four days the unresponsiveness gradually resolved, but the patient displayed a severe amnesic syndrome with anterograde and retrograde deficits. With the exception of horizontal ocular motility, neurolocic exam was otherwise normal. MRT and 3D-PC-angiography revealed a bilateral symmetric lesion of the medio-ventral thalamus (Fig. 1a) which was presumably caused by a deep venous thrombosis although at the time of examination MR angiography could not show an occlusion of the basal vein of Rosenthal. Prior testing showed that he was able to read words as they were presented within the present study without his normally worn glasses. The patients´ left eye was covered because of ocular motility thus visual perception was restricted to the right eye. Neuropsychological assessment performed 4 weeks after the event showed a severe amnesic syndrome. Formal testing revealed a WAIS-Verbal IQ of 53 and the Mini-mental State Examination score was 18 (Folstein). Using the WAIS-R digit span subtest shortterm memory assessment revealed a digit span of 4 (Tewes, 1994). Episodic verbal memory measured by means of California Verbal Learning Test (CVLT) (Delis et al., 1987) and Wechsler Memory Scale (WMS)–Logical Memory was severely impaired. Additionally, the patient showed poor non-verbal memory tested with the REY-figure (Spreen and Strauss, 1991).
Procedure The patient was seated in a comfortable chair and viewed a screen where words were visually presented. Each word appeared for 500 ms with an inter stimulus interval of 2.2 s. Totally, 130 words were presented. 50 words were single presentations and 40 words were repeated after an average lag of 44 intervening items (range from 37 to 49). The instruction given to the patient was to decide whether the presented words (concrete nouns) were animate or inanimate. Responses were made by pressing a left key with the left hand for animate or by pressing a right key with the right hand for inanimate. Classification response times associated with first and second presentations were statistically compared by calculating a paired t-test. Afterwards, an explicit recognition test was performed. Within this test, 60 words were repetitions from the prior animate/inanimate discrimination task (30 words were repetitions from previous single presentations and 30 words were repetitions from previous repeated words) and 30 words were new. The patient was instructed to discriminate between repeated and new words. Recognition accuracy was determined.
Preserved memory traces within diencephalic amnesia
539
MEG recordings The MEG was a 143 channel whole-head-system manufactured by CTF Systems Inc. (Canada) and situated in a magnetically shielded room. The signals were sampled with a rate of 250/s and filtered online with a bandpass from 0.3 to 80 Hz. A bandpass from 0.3 to 50 Hz was used offline (zero phase filters from CTF Systems Inc.). The period of 500 ms before stimulus onset served as baseline. The whole recording epoch of one trial was 2.3 s. Trials were visually inspected and excluded if contaminated with artefacts. They were averaged according to their nature of being related to first or second presentations. MEG maps were produced to compare the magnetic field distributions elicited by both word classes. A difference map was created to demonstrate the brain regions where differential activity occurred.
Statistical analysis As the region of interest we selected 6 sensors around the sensor MRP43 which showed the largest activity difference (Fig. 1b). The mean amplitudes of a 50 ms interval (from 160 ms to 210 ms after stimulus onset including the difference peak) averaged across these 6 sensors were compared calculating a t-test.
Results Behaviour
83.8% of all presented words were correctly classified as animate or inanimate. This result indicates that the lexical ability of the patient was in the normal range (although not perfect). The mean response time for first presentations was 1351 ms (SD ⫽ 267.4) and for second presentations it was 1301 ms (SD ⫽ 250.3). A paired t-test revealed no significant difference between these two means (t ⫽ .667; p ⫽ .509). The analysis of the explicit word recognition test resulted in a hit rate (correctly identified repetitions) of 0.33 and a false alarm rate (falsely classified new words) of 0.36. According to Snodgrass and Corwin (1988) we calculated a discrimination index by subtracting the false alarm rate from the hit rate. The analysis revealed a memory score of ⫺0.03 indicating that the recognition performance of the patient was at chance level reflecting a severe loss of explicit memory. Physiology (MEG data)
The averaged event-related fields (ERFs) produced by first and second presentations differed at early stages of word processing. To demonstrate these activity differences the averaged ERFs produced by second presentations were subtracted from the averaged ERFs produced by their first presentations. The resulting activity differences were mainly distributed over posterior recording sites and due to higher activation associated with first presentations. The period of 50 ms (ranging from 160 ms to 210 ms after stimulus onset) including the peak of the activity difference at approximately 180 ms was chosen for the MEG maps presented in Fig. 1b. The largest activity difference occurred at the right parietal sensor MRP43 which is marked in Fig. 1b. The statistical analysis revealed that across the 6 sensors (region of interest) the averaged activity associated with first presentations was significantly higher
540
P. Walla et al.
a Maps
Difference Map 160ms – 210ms after stimulus onset
First presentations
First presentations minus Second presentations Second presentations
Difference
200fT -200fT
b
0s
0.6s
First presentations
Second presentations
200fT
200fT
-200fT
-200fT 0s
0.6s
0s
0.6s
Preserved memory traces within diencephalic amnesia
541
than the averaged activity associated with second presentations (t ⫽ 4.095, p ⬍ 0.000). Discussion
The present data were collected in order to contribute to the understanding of human memory functions. Formal neuropsychological testing revealed poor postevent memory functions of the patient. The poor explicit recognition performance (chance level) for words presented during the lexical decision task confirmed the severe loss of explicit memory. These behavioural findings are consistent with previous reports about patients with thalamic lesions (Frey, 1995; Hodges and McCarthy, 1993). On the basis of the explicit memory impairment of our patient we conducted our study to investigate unconscious brain processes. The absence of explicit memory was an important advantage to test the phenomenon of priming which is usually contaminated by explicit memory. In the animate and inanimate lexical decision task the patient did not show a behavioural repetition priming effect. The response times associated with first and second presentations did not differ. The absence of priming on the behavioural level is striking because this form of memory has often been reported to be robust and still evident in amnesic patients (e.g. Hamann and Squire, 1997; Fleischmann et al., 1998). Also in patients with thalamic lesions priming has been found to be similar to healthy subjects (Cermak et al., 1991; Markowitsch et al., 1993). In normal subjects repetition priming effects have been described as being associated with differential neural activity at posterior brain areas (e.g. Beauregard et al., 1997). Repetition priming is usually accompanied by significant reductions in posterior brain activity. Our patient suffered from selective bilateral medio-ventral thalamic lesions having all posterior parts of the brain intact. Therefore, repetition priming would be expected to be normal. In fact, the analysis of the physiological data revealed evidence of a repetition priming effect indicated by different brain activity associated with first and second presentations of words. Over posterior brain areas the activity associated with repeated items 䉳
Fig. 1. a Magnetic Resonance Imaging (MRI) of the bilateral thalamic lesions. Left: Axial flair MR image (TR/TE/NEX: 7383/130/3) with 4 mm slice thickness showing hyperintense lesions in the medial ventral part of both thalami (white arrows). Right: T2w coronal MR image (TR/TE/NEX: 4000/110/4) with 2 mm slice thickness. The hyperintense thalamic bilateral lesions are clearly depicted showing an involvement of the centromedian and parafascicular nuclei, which can be assumed to be perfused from the thalamoperforating arterial group. b Averaged event-related magnetic field maps. The difference map on the left hand side demonstrates the distribution of the mean differential activity within a period of 50 ms (160 ms to 210 ms after the stimulus onset) over posterior areas. The greatest difference occurred at the marked sensor (MRP43). The curve below shows the averaged activity at this sensor across the whole trial length. As can be seen on this curve the difference maximum was at approximately 180 ms after the stimulus onset. On the right hand side the averaged event-related magnetic field maps of both first and second presentations are shown separately. The averaged curves from the sensor MRP43 are shown below. Note that the differential activity is due to higher activation associated with first presentations than with second presentations
542
P. Walla et al.
was decreased at quite early stages of word processing which is consistent with previous findings (Beauregard et al., 1997; Buckner et al., 1998). The difference peak occurred at approximately 180 ms after stimulus onset. However, the behavioural analysis did not yield evidence of priming. Generally, behavioural impairments have been found related to bilateral thalamic lesions (e.g. Bogousslavsky et al., 1991). Under normal circumstances the primed information mediated by posterior brain areas has to be transformed into a faster motor response to result in a decrease in response time. It might be that in our patient the relevant functional impairment which was responsible for a lack of such a decrease of response time was at this stage of processing. Perhaps, faster processing of repeated items was not transformed into shortened reaction times because the bilateral medio-ventral thalamic lesions possibly interrupted critical information processing pathways. The present study underlines the reliability of brain imaging methods for investigating unconscious brain processes. Priming, a normally robust form of unconscious memory was found to be impaired by means of reaction times within a lexical decision task. However, with the MEG we could detect physiological evidence for preserved priming reflected by activity reductions related to repeated stimulation over posterior brain areas. This fact may represent an interesting contribution to the understanding of human memory functions emphasising the dissociation between physiological and behavioural events with regard to priming. Acknowledgements The authors want to thank the Austrian Science Fund, the Austrian National Bank and the Human Frontier Science Program for financial support. Many thanks also go to Prof. Imhof for providing the MR-images of the patient.
References Beauregard M, Chertkow H, Gold D, Karama S, Benhamou J, Babins L, Faucher A (1997) Word priming with brief multiple presentation technique: preservation in amnesia. Neuropsychologia 35(5): 611–621 Bogousslavsky J, Regli F, Delaloye B, Delaloye Bischof A, Assal G, Uske A (1991) Loss of psychic self-activation with bithalamic infarction. Neurobehavioural, CT, MRI and SPECT correlates. Acta Neurol Scand 83(5): 309–316 Buckner RL, Goodman J, Burock M, Rotte M, Koutstaal W, Schacter DL, Rosen B, Dale A (1998) Functional-anatomic correlates of object priming in humans revealed by rapid presentation event-related fMRI. Neuron 20: 285–296 Cermak LS, Verfaellie M, Milberg W, Letourneau L, Blackford S (1991) A further analysis of perceptual identification priming in alcoholic Korsakoff patients. Neuropsychologia 29(8): 725–736 Delis D, Kramer J, Kaplan E, Ober B (1987) California Verbal Learning Test (CVLT). Testmanual, The Psychological Corporation, San Antonio Fleischman DA, Gabrieli JDE, Reminger SL, Vaidya CJ, Bennett DA (1998) Object decision priming in Alzheimer’s disease. J Int Neuropsychol Soc 4(5): 435–446 Frey HR (1995) Das bilaterale paramediane Thalamusinfarktsyndrom. Schweiz Med Wochenschr 125: 1294–1298 Hamann SB, Squire LR (1997) Intact perceptual memory in the absence of conscious memory. Behav Neurosci 111(4): 850–854
Preserved memory traces within diencephalic amnesia
543
Hodges JR, McCarthy RA (1993) Autobiographical amnesia resulting from bilateral paramedian thalamic infarction. Brain 116: 921–940 Markowitsch HJ, von Cramon DY, Schuri U (1993) Mnestic performance profile of a bilateral diencephalic infarct patient with preserved intelligence and severe amnesic disturbances. J Clin Exp Neuropsychol 15(5): 627–652 Rugg MD, Mark RE, Walla P, Schloerscheidt AM, Birch CS, Allan K (1998) Dissociation of the neural correlates of implicit and explicit memory. Nature 392: 595–598 Salmaso D, Longoni AM (1985) Problems in the assessment of hand preference. Cortex 21(4): 533–549 Schacter DL (1987) Critical review: Implicit memory: history and current status. J Exp Psych 13(3): 501–518 Schacter DL, Buckner RL (1998) Priming and the brain. Neuron 20: 185–195 Snodgrass JG, Corwin J (1988) Pragmatics of measuring recognition memory: applications to dementia and amnesia. J Exp Psychol 117: 34–50 Spreen O, Strauss E (1991) A compendium of neuropsychological tests: administration, norms, and commentary. Oxford University Press, New York Tewes U (1994) Hamburg-Wechsler-Intelligenztest für Erwachsene – Revision 1991 (HAWIE-R). Hans Huber, Bern Authors’ address: Dr. P. Walla, Department of Clinical Neurology (Ebene 6), AKH Wien, Währinger Gürtel 18-20, A-1090 Wien, Austria, e-mail:
[email protected]
