- Email: [email protected]
Neurophysiological correlates of visual stimulus recognition in man
International Elsevier
Journal of Psychophysiology
NEUROPHYSIOLOGICAL
BECHTEREVA
CORRELATES
N.P. and KROPOTOV
YU.D.
Department of Human Neurophysiology, Leningrad 197022 (U.S.S.R.)
Institute
(Accepted
November
317
1 (1984) 317-324
OF VISUAL STIMULUS
of Experimental
Medicine,
U.S.S.R.
RECOGNITION
Academy
IN MAN
of Medical Sciences, Pavlov’s Street, 12,
26th, 1983)
Key words: neurophysiological correlates components - human
- visual stimulus
recognition
~ neuronal
activity
- implanted
electrodes
- early and late
This study investigated the components of evoked impulse activity of neurons and neuronal populations (NIA) in the human brain. Subjects were 5 parkinsonian patients, two patients with skull trauma and an epileptic, diagnozed and treated with implanted electrodes. NIA was recorded during the following psychological tests: (1) identification of letters and digits presented at near-threshold exposures; (2) recognition of polygonal shapes with and without semantic meaning. Peri-stimulus time histograms (PSTHs) for the cases of recognition and non-recognition in the first test and for cases of presentation of familiar and unfamiliar patterns in the second test were computed and compared with each other. PSTH components in the stimulus-response interval were classified into 3 groups: the earliest components with the latency 60-200 ms; the late components with latency 300-400 ms; and slow count rate shifts revealed 300-500 ms after stimulus presentation. No significant differences were found between the short-latency components for cases of recognition and non-recognition in the first test and for cases of presentation of familiar and unfamiliar patterns in the second test, while late components depended upon subjective estimation by the patient of the stimulus. Early components are supposed to be related to physical characteristics of the stimulus, while the late components with semantic meaning.
INTRODUCTION The last few decades have witnessed an increasing interest in the investigation of human cognitive processes, which has stimulated a great number of experimental studies in this field. Two approaches to the problem appear to be most workable at present: the psychological, based on recording different behavioural parameters (the time of the reaction, the number of correct responses, etc.) in the trials for visual and audial stimulus recognition (Neisser, 1966; Smith, 1968; Nickerson, 1972; Glezer et al., 1974), and the neurophysiological, based on analyzing bioelectric processes of the human brain. Briefly, recognition can be defined as a mental process of associating the presented object with some class (category) familiar to the subject from previous experience and fixed in his memory. 0167-8760/83/$03.00
0 1983 Elsevier Science Publishers
B.V.
Until the 1960s the scalp-recorded event-related potential (ERP) method was the only source of information concerning the neurophysiology of human cognitive processes. In spite of the progress made in studying the event-related potentials in the visual and audial stimulus recognition tests (Brown et al., 1973; John and Schwarts, 1978; Parasuraman et al., 1982; and others) the data obtained fail to provide sufficient insight into the neuronal cognitive mechanisms of the brain due to the obscure nature of the EEG and ERP genesis. The application of long-term implanted electrodes in the 1960s for diagnostics and therapy of brain disorders has opened up new opportunities for studying neurophysiological correlates of various mental activities and recognition processes in particular (Bikford et al., 1953; Bates, 1961; Walter and Crow, 1961; Bechtereva et al., 1963; Bechtereva, 1971). Taking advantage of the direct con-
318
tact with the brain, we used electrode implantation for diagnostics and treatment to record neuron and neuronal population impulse activity (NIA) of the human brain during the mental process without interruption of treatment, account being taken of all the medical and ethical considerations previously described (Bechtereva et al., 1983). In the course of the investigations the brain areas were located maintaining tests for mental operations (short-term memory in particular) (Bechtereva, 1978; Bechtereva et al., 1977; Gretchin, 1972; Ilyukhina, 1977; Halgren et al., 1978). The present study further researches this problem and aims to investigate the components of neuron evoked impulse activity of the human brain responsible for visual stimulus recognition.
METHODS Investigations were made of 5 parkinsonian patients, two patients with scull trauma and an epileptic, their age ranged from 25 to 50 years, who had electrodes implanted for diagnostics and therapy into different nuclei of the thalamus, strio-pallidar system, premotor and temporal areas of the cortex. The electrodes were implanted by neurosurgeons Khilko, Ruchkov and Gurchin who used the computer stereotaxic method developed by Anichkov (1977). The electrodes were made of gold wire 0.1 mm in diameter; throughout their length except for the active surface on the electrode tip, they were insulated by neutral plastic. The electrodes were gathered in bundles of 6-10 in a staircase like way so that their active surfaces were spaced at a distance of 2-4 mm. The active surface ranged from 0.01 to 0.15 mm*, the mean value of the electrode impedance in cell structures made up 15 k0 when recorded by bipolar lead at 10 kHz. All the patients had normal vision with no disorders of visual memory. The investigations began 2 or 3 weeks after the electrodes were implanted. During the tests the patients were sociable, felt quite well and responded with interest to psychological visual stimulus recognition trials, displaying no negative emotional attitudes to the investigation. Seated in a specially darkened room, the patients per-
formed two different psychological tests. The tests of the first type were made to separate the NIA components responsible for visual stimulus recognition. For this purpose the stimulus exposure time was so short that approximately in 50% of cases the patients failed to recognize visual stimuli due to fluctuations in their functional state though they could detect them. The method used was the following: the light matrix for visual stimulus presentation was placed at a distance of 1.5 m at eye level in front of the patients. There were a rectangular matrix of 70 X 50 mm of red LEDs, a microphone amplifier and a decipher receiving controlling signals from the computer ‘Plurimat-S’ (Intertechnique, France). The decipher presented one of 16 stimuli (10 digits, 4 letters, a regular dot pattern and a dot in the center of the screen). Our stimuli were digits 6, 8, 9 and letters E, A, N; these were chosen so that the intensity of their light flow was identical. The patient was asked to try and recognize the stimuli presented to him on the matrix. All the tests, consisting of 128 similar trials, were carried out in the morning. At the beginning of the trial a patient had one of the above stimuli presented to him, which, for convenience, will be hereafter referred to as the recognizable stimulus. Another stimulus, a regular dot pattern used as a signal for verbal response of the patient to check the correct performance of the trial, followed 1.5 s after a short presentation of the recognizable stimulus. If the visual stimulus was recognized the patient was asked to name the figure or letter presented to him, and to say ‘no’ if it was not recognized. In the rare cases (l-2%) when the patient failed to notice the stimulus he was supposed to answer ‘I did not notice’ and such trials were not taken into account in further analysis. Before the beginning of the test all the patterns were presented to the patient with a long exposure time (150 ms). Further the exposure time was gradually shortened so that finally the patient failed to recognize half of the visual stimuli presented to him. On this basis the test exposure time varied individually within the range of 500-2000 ps. After the preliminary trials were made and the patient had adapted to the surroundings we recorded the NIA during the
319
Fig. 1 illustrates the polygonal shapes used in the investigations. Approximately 50% of the whole number of the polygowith familiar nal shapes (for example: l-5) were associated patterns, while others (for example: 6-10) were identified as unfamiliar.
psychological recognition trials and the data were fed into the computer. Psychological tests of the second type were used to differentiate NIA evoked components related with the semantic features of visual stimuli. For this purpose the patient had various polygonal shapes presented to him (white lines against a dark background - the negatives are presented in Fig. l), which in approximately half of the cases could be associated with familiar patterns. Physical characteristics of the stimuli (illumination, colour, number of lines and angles) were on average identical for the group of semantically significant and insignificant stimuli. The shapes were presented by means of a 3-channel tachistoscope. A white screen with projected shapes of 50 x 50 cm was placed 1.5 m from the patient at his viewing level. The tests were carried out in the morning and consisted of 80-160 similar trials. In each trial one of the above shapes referred to as recognizable was presented during 150 ms. 1.8 s after the presentation of recognizable shape, another shape, a checkerboard pattern was presented as a signal for the patient’s vocal response. The patient was asked to watch the shape and if it was familiar he was to say ‘yes’ on presentation of the checkerboard pattern; in the reverse case he was expected to say ‘no’. The NIA of the human brain was amplified by means of the g-channel polyneurograph with input impedance 100 MO and the input noise level not exceeding 25 PV from peak-to-peak (Danko,
Kaminskij, 1982). A reference electrode, a silver-silver chloride plate with the active surface of 15 cm2, was placed on the patient’s forehead. We succeeded in recording NIA, with amplitude range from 80 to 300 PV peak-to-peak, on the average in 20% of the total number of 312 brain areas where the electrodes were implanted. The total number of neuronal populations was 62. The NIA was fed into the computer memory through 8 channels by amplitude threshbld discriminators. Standard impulses from the threshold discriminators were applied to separate lines of input register of the ‘Plurimat-S’ special data acquisition interface. The contents of the register was transferred into the computer memory at the rate of 2 kHz, which means that the moment of spike generation was determined with the accuracy up to 500 ps. The equipment used for the NIA computer processing is dealt with at length in the book by Gogolitsin and Kropotov (1983). The data were not fed continuously into the computer, but in parts, each trial lasting 4 s. Each 4-s record included a background fragment when the patient performed no activity preceding the stimulus presentation. In each test of every patient the psychological trials were divided into two groups depending on the patient’s individual response (‘recognized - not recognized’ in the first round of tests and ‘familiar pattern - unfamiliar pattern’ in the second). In each group of trials the k-neuronal population in the i-trial and within the j-time interval (bin) was characterized by value flJ”- the number of discharges in the given neuron population in the appropriate trial and the appropriate bin (in our tests the bin was equal to 64, 128, 150 and 256 ms (Fig. 2)). The mean values and variances of fi estimated for each neuronal population were the object of our analysis. These values were estimated in all the trials of the same group by averaging. Reliable count rate reactions in the neuronal populations in response to a trial were revealed in the following way. The first to be determined were the statistic characteristics of the background activity - the mean value and count rate variance. Then comparison was made for each bin using the t-test between the mean count rate of the given population in this bin and the mean value of the back-
320
n
&
R NO
I
Fig. 2 illustrates the procedure for computing the PSTHs and differences between mean count rates of evoked responses for familiar (YES) and unfamiliar (NO) visual patterns. On the top the presentations of polygonal shape (S) and checkerboard pattern (R) are indicated. l-3-evoked count rates in single trials. PSTH, per&stimulus-time histogram. For one chosen (blackened) bin count rate histograms are shown (N, number of trials; F, count rate in single trial and in the given bin).
ground fragment thus estimating the reaction validity. To compare the NIA reactions for two different groups of the trials comparison was made in each bin by the t-test of the count rate mean values calculated for these two groups. The validity standard was taken to be P < 0.01.
RESULTS Fig. 3 shows variants of mean count rate dynamics obtained for different neuronal populations in different patients when visual stimuli were presented in the tests on the recognition threshold (cases of recognition). In 90% of the neuronal populations the evoked reactions were found to be multiphase and were observed on the peri-stimulus time histograms (PSTHs) as peak with a pronounced extreme and as slow shift of the count rate. In the interval between the stimulus presentation and a regular dot pattern characteristic PSTH
I
Fig. 3. Examples of the count rate responses to the familiar patterns in 6 different neuronal populations. Bin = 150 ms. The interrupted vertical lines correspond to the presentations of the recognizable stimulus and checkerboard pattern. The horizontal line, background count rate level. The interrupted horizontal line corresponds to significant deviations (P < 0.01) from mean count rate of the background fragment. F, mean background count rate. Abbreviations: D, right hemisphere; S, left hemisphere; VL, ventrolateral nucleus of the thalamus; Cd, caudate nucleus; LP, posterior lateral nucleus of the thalamus: PI. globus pallidum. On the bottom, short-, long-latency components and slow count rate shift are schematically shown.
components could be classified into 3 groups: the earliest components with the latency 60-200 ms and duration of about 200 ms; the late components with the latency 300-400 ms and duration 300-600 ms and slow count rate shifts revealed 300-500 ms after the stimulus presentation and persisted until the patient’s verbal response (Fig. 4). Besides these components of the count rate, evoked changes during the patient’s verbal response were recorded in 70% of the neuronal populations (Fig. 3). Fifteen percent of the investigated neuronal populations (located in the posterior and anterior ventral nuclei, lateral posterior of the thalamus, corpus callosum and caudate nucleus) were shown by PSTHs to have components with short latency, their amplitude ranging from 10 to 30% of the mean count rate of the background fragment.
321
. ... . 1..
““ii’“”
““‘i”“’
. ..
Fig. 4. Examples of the PSTHs in the stimulus-response interval for recognizable (top) and non-recognizable (bottom) stimuli in 8 different neuronal populations. Arrows indicate bins with significant (P -c 0.01) differences between mean count rates of evoked responses to recognizable and non-recognizable stimuli. Abbreviations: AV, anterior ventral nucleus; VP, posterior ventral nucleus; CM, centrum median of the thalamus, CC, corpus callosum. Other notations as in Fig. 3.
PSTH stimulus-response interval had long-latency components with the amplitude 20-40% of the background fragment mean count rate in 40% of the investigated neuronal populations (located in the anterior ventral, posterior lateral, reticular nuclei, centrum median of the thalamus and caudate nucleus). Slow count rate shifts with the maximum amplitude ranging from 20 to 50% of the background fragment count rate were observed in 45% of the investigated neuronal populations (in anterior ventral, posterior lateral, posterior ventral, ventralateral nuclei, centrum median of the thalamus, caudate nucleus and corpus callosum). No significant differences were found in any neuronal populations under study between the mean count rate in the background fragments followed by presentations of recognizable and non-recognizable stimuli. Neither did we record any significant differences between short-latency components of the evoked responses to recognition and non-recognition.
Significant differences between mean count rate of the neuronal populations for cases of recognition and non-recognition were recorded 300-500 ms after the stimulus presentation in 25% of the investigated neuronal populations (in the posterior ventral, posterior lateral, centrum median of the thalamic nucleus, in the corpus callosum and caudate nucleus). Half of the above neuronal populations were found to differ significantly within 300-800 ms after the stimulus presentation. The amplitude of the long-latency components of the evoked reactions in these neuronal populations appeared to be larger in cases of recognition as compared with cases of non-recognition (histograms 2-5 in Fig. 4). In the other half of the above mentioned neuronal populations significant differences between mean count rate for cases of recognition and nonrecognition were observed (histogram 7 in Fig. 4) between count rate slow shifts, the amplitude being larger in case of recognition as compared to nonrecognition. It is noteworthy that in the evoked NIA of some brain areas from one to three components could be observed. Thus besides the responsebound component, the PSTH showed a shortlatency component and a slow shift of the count rate in the neuronal population of the posterior lateral nucleus of the thalamic nucleus (histogram 1 in Fig. 4); in the neuronal populations of ventrolateral nucleus of the thalamic nucleus we observed simultaneously short and long-latency components (histogram 2 in Fig. 4). However only one of the above components was observable in the evoked activity of other neuronal populations (histogram 7 in Fig. 4). The same patient was found to have both neuronal populations with differences and without any for cases of recognition and non-recognition (histogram 8 in Fig. 4). As was mentioned in Methods the patients had polygonal shapes with and without meaning presented to them with the view to studying neuronal reactions reflecting semantic characteristic of visual stimuli. PSTH obtained in different patients for different neuronal populations had the same characteristic features as those elicited in the first series of investigations when the visual stimuli were presented on the perception threshold. Sig-
322
nificant differences between mean count rates of the neuronal populations for recognition of familiar and unfamiliar shapes were observed only 300-500 ms after the stimulus presentation. They were observable in 20% of all the investigated neuronal populations in the ventrolateral nucleus of the thalamus, in the globus pallidum, the caudate nucleus and the corpus callosum, in whose evoked activity long-latency components or count rate slow shifts could be recorded. The differences were in the amplitude of the long-latency components in these neuronal populations which were found to be larger on presentation of familiar shapes (histograms 2, 3 in Fig. 5), as for the count rate slow shifts, they were either recorded 300-600 ms later in case of unfamiliar shapes or their amplitude appeared to be smaller (histograms 5, 7 in Fig. 5). It is noteworthy that although long-latency components could be different for some neuronal populations on presentation of familiar and unfamiliar shapes, their phases remained unchanged (histograms 3, Fig. 5). Besides the neuronal populations in whose evoked activity we recorded significant differences between the late components
.
...iiiiiiii “”
..I
Fig. 5. Examples of the PSTHs in the stimulus-response interval for familiar (top) and unfamililar (bottom) shapes in 8 different neuronal populations. Abbreviations: T COT,temporal cortex. Other notations as in Figs. 3 and 4.
of evoked reactions for familiar and unfamiliar shapes, the same patient was found to have neuronal populations whose long-latency components and the count rate slow shifts of the evoked reactions did not differ for two varients of trials (histograms 1, 4, 6, 8 in Fig. 5).
DISCUSSION Multiphase count rate changes during the presentation of visual stimuli were shown by our investigations to evolve in the neuronal populations of different nuclei of the thalamus and the striopallidar system, i.e. in the structures not involved in primary visual analyzer areas. Count rate changes correlating in time with the patient’s verbal response were also recorded in those neuronal populations. These data confirm our concept concerning the polyfunctional nature of many neuronal populations in the human brain based on studying the oxygen dynamics, impedance and some other indices of the brain’s activity during various psychological trials (Bechtereva, 1971; 1978; 1980). In analyzing the data, of special interest were similar responses in different distant populations of the brain obtained from the same and different patients, which proves that there must be some typical neuronal mechanisms maintaining the performance of the psychological trials. Similar data indicative of the similarity of reactions in the neuronal populations of different brain structures in animals were obtained in studying the impulse activity and evoked potential during various behavioural acts (John, 1972; Shvyrkov, 1978). Short- and long-latency components as well as count rate slow shifts evolved in the evoked reactions of the neuronal populations of the human brain when visual stimuli were presented. These components could be recorded simultaneously in the NIA evoked changes of the same brain area. We also revealed neuronal populations with only one of the above components recorded in the count rate dynamics. None of the investigated neuronal populations had different shortlatency components for cases of recognition and non-recognition. However, the long-latency components and the count rate slow shifts, at least for
323
some neuronal populations, did differ for these two cases of recognition trials. The difference was as follows: when the patient recognized the pattern, the amplitude of the long-latency components and the count rate slow shifts appeared to be larger than in the reverse case when the pattern was not recognized by the patient. Note should be made that none of the neuronal populations revealed any significant differences between the count rates in the background segments. Thus the original functional states of the brain which served as background for the visual stimulus presentation did not differ, according to the count rate dynamics data, when the patient recognized the pattern and when he failed to do so. Some of the above mentioned neuronal populations had NIA that changed differently whether or not the patient recognized the pattern, whereas in other adjacent brain areas there occurred neuronal populations which had statistically identical late components of evoked reactions whatever the result of the trial. Thus differences in the long-latency components and count rate slow shifts of evoked reactions, depending on the subjective stimulus estimation, are characteristic of some neuronal groups only, which confirms our concept that it is the system determining the optimal use of the brain’s potentialities and not individual areas or the whole brain that maintains the mental activity (Bechtereva, 1983). The detection of the signal common to both variants of trials for recognition suggests that the short-latency components, which always occur whether or not the pattern is recognized, are actually related with signal detection. On the basis of our investigations we may assume that the long-latency components and the count rate slow shifts can be correlated with mental processes of comparing the input information with the memory engram and decision making. For this assumption to be varified in conditions different from the first test, investigations were carried out when patterns with or without semantic meaning were presented not on the recognition threshold but during the period sufficient for their recognition. The analysis of PSTHs estimated for recognition of semantically meaningful and meaningless stimuli suggests that the same components of the
evoked NIA as in the previous investigations may be observable. None of the studied neuronal populations showed any differences in the mean count rates in the background fragments that preceded familiar and unfamiliar patterns. Neither did we record any significant differences in the mean count rates of the neuronal populations in the NIA record fragments corresponding to shortlatency components. The latter fact is ample evidence that short-latency components are not stimulus meaning bound, but they reflect physical characteristics of the signal which in our test were statistically identical for both familiar and unfamiliar patterns. In some neuronal populations of the investigated brain structures the long-latency components and slow count rates were found to differ considerably when familiar and unfamiliar patterns were presented. The differences were in the long-latency components and count rate slow shifts essentially increasing when the patient had a familiar pattern presented to him. In addition to the above assumption concerning the functional meaning of the evoked NIA late components, the data of these investigations suggests that the late components are not stimulus bound but are related to the stimulus semantic characteristics and its subjective estimation by the patient. Three orthogonal components: N 100, P 300 and the potential slow positive shifts were separated in the course of investigating ER recorded from the scalp of healthy subjects by the principal components method. The N 100 components was also shown to be related to the stimulus detection, whereas the P 300 component and the potential slow positive shifts were related to recognition processes and the decision making (Parasuramen et al. 1982). Thus data concerning the NIA count rate dynamics in the subcortical formations of the human brain during the visual stimulus recognition show good agreement with the investigations of the evoked potentials recorded from the scalp of healthy subjects in the test for audial stimulus recognition. REFERENCES Anichkov, A.D. (1977) Stereotaksitcheskij apparat dlija wedenija dolgosrotchnych mnoshestvennych vnutrimozgovych elecktrodov. Fuiol. tcheleueka, 3: 372-375.
324
Bates, Y.A.V. (1961) Depth electrodes in the human subject. Basic technical, interpretative and ethical considerations. Int. Congr. Ser., 31: 62-64. Bechtereva, N.P. (1971) Neirophysiologitcheskie aspektl psichitcheskoi dejntelnosty tcheloveka, Nauka, Leningrad, 119 PP. Bechtereva, N.P. (1978) The Neurophysiological Aspects of Human Mental Activity, 2nd end., Oxford Univ. Press, New York, 181 pp. Bechtereva, N.P. (1980) Zdorovyi i Bolnoi Morg Tcheloveka, Nauka, Leningrad, 208 pp. Bechtereva, N.P. (1983) Preface to the book by Gogolitsin, Yu.L. and Kropotov, Yu.D. Issledovanre Tchastoty Razjadov Neironov Mozga Tcheloveka, Nauka, Leningrad, pp. 3-7. Bechtereva, N.P., Gratchov, K.V., Orlova, A.N. and Jatsuk, S.L. (1963) Ispolzovanie mnoshestvennych electrodov, vzivlennych v podkorkovie obrazovanija golovnogo mozga tcheloveka dlija lechenia giperkinezov. J. nevropatol. psichiatr., 1: 3-8. Bechtereva, N.P., Bundzen, P.V. and Gogolitsin, Yu.L. (1977) Mozgoue Cody Psichitcheskoi DeJatelnosty, Nauka, Leningrad, 165 pp. Bechterva, N.P., Gogolitsin, Yu.L., Ilyukhina, V.A. and Pakhomov, S.V. (1983) Dynamic neurophysiological correlates of mental processes. Int. J. Psychophyslol., 1: 49-63. Bickford, R.G., Dodge, H.W., Peterson, M.C. and Sem-Jocobsen, C.W. (1953) A new method of recording from subcortical regions of the human brain. Electroenceph. clin. Neurophysiol., 5: 464. Brown, W.S., Marsh, J.T. and Smith, J.C. (1973) Contextual meaning effects on speech-evoked potentials. Behav. Biol., 9: 755-161. Danko, SC. and Kaminskij, Yu.L.. (1982) Sistema Technitcheskych Sredstv Neirofiziologitcheskych IssledovanQ Tcheloveka, Nauka, Leningrad, 135 pp.
Gogolitsin, Yu.L. and Kropotov, Yu.D. (1983) Issledooanre Tchastoty Razrjadov Neironov Morga Tcheloveka, Nauka, Leningrad, 120 pp. Gretchin, V.B. (1972) Nalichnij kislorod golovnogo mozga tcheloveka. In Klinitcheskaja newophyslologia, Nauka, Leningrad, pp. 494-530. Glezer, V.D., Leushina, L.I., Nevskaya, A.A. and Prazdnikova, N.V. (1974) Studies on visual pattern recognition in man and animals. Vuion Res., 14: 555-583. Halgren, E., Babb, T.L. and Crandall, P.H. (1978) Activity of human hippocampal formation and amygdala neurons during memory testing. Electroenceph. clin. Neurophyslol., 45: 585-601. Ilyukhina, V.A. (1977) Medlennye Bioelektritcheskie Protsessy Goloonogo Morga Tcheloueka, Nauka, Leningrad, 184 pp. John, E.R. (1972) Switchboard versus statistical theories of learning and memory. Science, 177: 850. John, E.R. and Schwarts, E.L. (1978) The neurophysiology of information processing and cognition. Ann. Rev. Psyrhol., 29: l-29. Neisser, U. (1966) Cognitive Psychology, Appleton Century Crofts, New York, 351 pp. Nickerson, R.S. (1972) Binary-classification reaction time. A review of some studies of human information-processing capabilities. J. Psychon. Monogr. Suppl., 4: 275-318. Parasuraman, R., Richer, F. and Beatty, J. (1982) Detection and recognition: concurrent processes in perception. Percept. Psychophys., 31: p. l-12. Shvyrkov, V.B. (1978) Neirofiziologitcheskoe Izutchenie Systemnich Mechanirmoo Pooedenia, Nauka, Moscow, 239 pp. Smith, E.E. (1968) Choice reaction time: an analysis of the major theoretical position. Psychol. Bull., 70: 77-110. Walter, W.G., Crow, H.J. (1961) Depth recording from human brain Int. Congr. Ser. 37: p. 64-65.
Journal of Psychophysiology
NEUROPHYSIOLOGICAL
BECHTEREVA
CORRELATES
N.P. and KROPOTOV
YU.D.
Department of Human Neurophysiology, Leningrad 197022 (U.S.S.R.)
Institute
(Accepted
November
317
1 (1984) 317-324
OF VISUAL STIMULUS
of Experimental
Medicine,
U.S.S.R.
RECOGNITION
Academy
IN MAN
of Medical Sciences, Pavlov’s Street, 12,
26th, 1983)
Key words: neurophysiological correlates components - human
- visual stimulus
recognition
~ neuronal
activity
- implanted
electrodes
- early and late
This study investigated the components of evoked impulse activity of neurons and neuronal populations (NIA) in the human brain. Subjects were 5 parkinsonian patients, two patients with skull trauma and an epileptic, diagnozed and treated with implanted electrodes. NIA was recorded during the following psychological tests: (1) identification of letters and digits presented at near-threshold exposures; (2) recognition of polygonal shapes with and without semantic meaning. Peri-stimulus time histograms (PSTHs) for the cases of recognition and non-recognition in the first test and for cases of presentation of familiar and unfamiliar patterns in the second test were computed and compared with each other. PSTH components in the stimulus-response interval were classified into 3 groups: the earliest components with the latency 60-200 ms; the late components with latency 300-400 ms; and slow count rate shifts revealed 300-500 ms after stimulus presentation. No significant differences were found between the short-latency components for cases of recognition and non-recognition in the first test and for cases of presentation of familiar and unfamiliar patterns in the second test, while late components depended upon subjective estimation by the patient of the stimulus. Early components are supposed to be related to physical characteristics of the stimulus, while the late components with semantic meaning.
INTRODUCTION The last few decades have witnessed an increasing interest in the investigation of human cognitive processes, which has stimulated a great number of experimental studies in this field. Two approaches to the problem appear to be most workable at present: the psychological, based on recording different behavioural parameters (the time of the reaction, the number of correct responses, etc.) in the trials for visual and audial stimulus recognition (Neisser, 1966; Smith, 1968; Nickerson, 1972; Glezer et al., 1974), and the neurophysiological, based on analyzing bioelectric processes of the human brain. Briefly, recognition can be defined as a mental process of associating the presented object with some class (category) familiar to the subject from previous experience and fixed in his memory. 0167-8760/83/$03.00
0 1983 Elsevier Science Publishers
B.V.
Until the 1960s the scalp-recorded event-related potential (ERP) method was the only source of information concerning the neurophysiology of human cognitive processes. In spite of the progress made in studying the event-related potentials in the visual and audial stimulus recognition tests (Brown et al., 1973; John and Schwarts, 1978; Parasuraman et al., 1982; and others) the data obtained fail to provide sufficient insight into the neuronal cognitive mechanisms of the brain due to the obscure nature of the EEG and ERP genesis. The application of long-term implanted electrodes in the 1960s for diagnostics and therapy of brain disorders has opened up new opportunities for studying neurophysiological correlates of various mental activities and recognition processes in particular (Bikford et al., 1953; Bates, 1961; Walter and Crow, 1961; Bechtereva et al., 1963; Bechtereva, 1971). Taking advantage of the direct con-
318
tact with the brain, we used electrode implantation for diagnostics and treatment to record neuron and neuronal population impulse activity (NIA) of the human brain during the mental process without interruption of treatment, account being taken of all the medical and ethical considerations previously described (Bechtereva et al., 1983). In the course of the investigations the brain areas were located maintaining tests for mental operations (short-term memory in particular) (Bechtereva, 1978; Bechtereva et al., 1977; Gretchin, 1972; Ilyukhina, 1977; Halgren et al., 1978). The present study further researches this problem and aims to investigate the components of neuron evoked impulse activity of the human brain responsible for visual stimulus recognition.
METHODS Investigations were made of 5 parkinsonian patients, two patients with scull trauma and an epileptic, their age ranged from 25 to 50 years, who had electrodes implanted for diagnostics and therapy into different nuclei of the thalamus, strio-pallidar system, premotor and temporal areas of the cortex. The electrodes were implanted by neurosurgeons Khilko, Ruchkov and Gurchin who used the computer stereotaxic method developed by Anichkov (1977). The electrodes were made of gold wire 0.1 mm in diameter; throughout their length except for the active surface on the electrode tip, they were insulated by neutral plastic. The electrodes were gathered in bundles of 6-10 in a staircase like way so that their active surfaces were spaced at a distance of 2-4 mm. The active surface ranged from 0.01 to 0.15 mm*, the mean value of the electrode impedance in cell structures made up 15 k0 when recorded by bipolar lead at 10 kHz. All the patients had normal vision with no disorders of visual memory. The investigations began 2 or 3 weeks after the electrodes were implanted. During the tests the patients were sociable, felt quite well and responded with interest to psychological visual stimulus recognition trials, displaying no negative emotional attitudes to the investigation. Seated in a specially darkened room, the patients per-
formed two different psychological tests. The tests of the first type were made to separate the NIA components responsible for visual stimulus recognition. For this purpose the stimulus exposure time was so short that approximately in 50% of cases the patients failed to recognize visual stimuli due to fluctuations in their functional state though they could detect them. The method used was the following: the light matrix for visual stimulus presentation was placed at a distance of 1.5 m at eye level in front of the patients. There were a rectangular matrix of 70 X 50 mm of red LEDs, a microphone amplifier and a decipher receiving controlling signals from the computer ‘Plurimat-S’ (Intertechnique, France). The decipher presented one of 16 stimuli (10 digits, 4 letters, a regular dot pattern and a dot in the center of the screen). Our stimuli were digits 6, 8, 9 and letters E, A, N; these were chosen so that the intensity of their light flow was identical. The patient was asked to try and recognize the stimuli presented to him on the matrix. All the tests, consisting of 128 similar trials, were carried out in the morning. At the beginning of the trial a patient had one of the above stimuli presented to him, which, for convenience, will be hereafter referred to as the recognizable stimulus. Another stimulus, a regular dot pattern used as a signal for verbal response of the patient to check the correct performance of the trial, followed 1.5 s after a short presentation of the recognizable stimulus. If the visual stimulus was recognized the patient was asked to name the figure or letter presented to him, and to say ‘no’ if it was not recognized. In the rare cases (l-2%) when the patient failed to notice the stimulus he was supposed to answer ‘I did not notice’ and such trials were not taken into account in further analysis. Before the beginning of the test all the patterns were presented to the patient with a long exposure time (150 ms). Further the exposure time was gradually shortened so that finally the patient failed to recognize half of the visual stimuli presented to him. On this basis the test exposure time varied individually within the range of 500-2000 ps. After the preliminary trials were made and the patient had adapted to the surroundings we recorded the NIA during the
319
Fig. 1 illustrates the polygonal shapes used in the investigations. Approximately 50% of the whole number of the polygowith familiar nal shapes (for example: l-5) were associated patterns, while others (for example: 6-10) were identified as unfamiliar.
psychological recognition trials and the data were fed into the computer. Psychological tests of the second type were used to differentiate NIA evoked components related with the semantic features of visual stimuli. For this purpose the patient had various polygonal shapes presented to him (white lines against a dark background - the negatives are presented in Fig. l), which in approximately half of the cases could be associated with familiar patterns. Physical characteristics of the stimuli (illumination, colour, number of lines and angles) were on average identical for the group of semantically significant and insignificant stimuli. The shapes were presented by means of a 3-channel tachistoscope. A white screen with projected shapes of 50 x 50 cm was placed 1.5 m from the patient at his viewing level. The tests were carried out in the morning and consisted of 80-160 similar trials. In each trial one of the above shapes referred to as recognizable was presented during 150 ms. 1.8 s after the presentation of recognizable shape, another shape, a checkerboard pattern was presented as a signal for the patient’s vocal response. The patient was asked to watch the shape and if it was familiar he was to say ‘yes’ on presentation of the checkerboard pattern; in the reverse case he was expected to say ‘no’. The NIA of the human brain was amplified by means of the g-channel polyneurograph with input impedance 100 MO and the input noise level not exceeding 25 PV from peak-to-peak (Danko,
Kaminskij, 1982). A reference electrode, a silver-silver chloride plate with the active surface of 15 cm2, was placed on the patient’s forehead. We succeeded in recording NIA, with amplitude range from 80 to 300 PV peak-to-peak, on the average in 20% of the total number of 312 brain areas where the electrodes were implanted. The total number of neuronal populations was 62. The NIA was fed into the computer memory through 8 channels by amplitude threshbld discriminators. Standard impulses from the threshold discriminators were applied to separate lines of input register of the ‘Plurimat-S’ special data acquisition interface. The contents of the register was transferred into the computer memory at the rate of 2 kHz, which means that the moment of spike generation was determined with the accuracy up to 500 ps. The equipment used for the NIA computer processing is dealt with at length in the book by Gogolitsin and Kropotov (1983). The data were not fed continuously into the computer, but in parts, each trial lasting 4 s. Each 4-s record included a background fragment when the patient performed no activity preceding the stimulus presentation. In each test of every patient the psychological trials were divided into two groups depending on the patient’s individual response (‘recognized - not recognized’ in the first round of tests and ‘familiar pattern - unfamiliar pattern’ in the second). In each group of trials the k-neuronal population in the i-trial and within the j-time interval (bin) was characterized by value flJ”- the number of discharges in the given neuron population in the appropriate trial and the appropriate bin (in our tests the bin was equal to 64, 128, 150 and 256 ms (Fig. 2)). The mean values and variances of fi estimated for each neuronal population were the object of our analysis. These values were estimated in all the trials of the same group by averaging. Reliable count rate reactions in the neuronal populations in response to a trial were revealed in the following way. The first to be determined were the statistic characteristics of the background activity - the mean value and count rate variance. Then comparison was made for each bin using the t-test between the mean count rate of the given population in this bin and the mean value of the back-
320
n
&
R NO
I
Fig. 2 illustrates the procedure for computing the PSTHs and differences between mean count rates of evoked responses for familiar (YES) and unfamiliar (NO) visual patterns. On the top the presentations of polygonal shape (S) and checkerboard pattern (R) are indicated. l-3-evoked count rates in single trials. PSTH, per&stimulus-time histogram. For one chosen (blackened) bin count rate histograms are shown (N, number of trials; F, count rate in single trial and in the given bin).
ground fragment thus estimating the reaction validity. To compare the NIA reactions for two different groups of the trials comparison was made in each bin by the t-test of the count rate mean values calculated for these two groups. The validity standard was taken to be P < 0.01.
RESULTS Fig. 3 shows variants of mean count rate dynamics obtained for different neuronal populations in different patients when visual stimuli were presented in the tests on the recognition threshold (cases of recognition). In 90% of the neuronal populations the evoked reactions were found to be multiphase and were observed on the peri-stimulus time histograms (PSTHs) as peak with a pronounced extreme and as slow shift of the count rate. In the interval between the stimulus presentation and a regular dot pattern characteristic PSTH
I
Fig. 3. Examples of the count rate responses to the familiar patterns in 6 different neuronal populations. Bin = 150 ms. The interrupted vertical lines correspond to the presentations of the recognizable stimulus and checkerboard pattern. The horizontal line, background count rate level. The interrupted horizontal line corresponds to significant deviations (P < 0.01) from mean count rate of the background fragment. F, mean background count rate. Abbreviations: D, right hemisphere; S, left hemisphere; VL, ventrolateral nucleus of the thalamus; Cd, caudate nucleus; LP, posterior lateral nucleus of the thalamus: PI. globus pallidum. On the bottom, short-, long-latency components and slow count rate shift are schematically shown.
components could be classified into 3 groups: the earliest components with the latency 60-200 ms and duration of about 200 ms; the late components with the latency 300-400 ms and duration 300-600 ms and slow count rate shifts revealed 300-500 ms after the stimulus presentation and persisted until the patient’s verbal response (Fig. 4). Besides these components of the count rate, evoked changes during the patient’s verbal response were recorded in 70% of the neuronal populations (Fig. 3). Fifteen percent of the investigated neuronal populations (located in the posterior and anterior ventral nuclei, lateral posterior of the thalamus, corpus callosum and caudate nucleus) were shown by PSTHs to have components with short latency, their amplitude ranging from 10 to 30% of the mean count rate of the background fragment.
321
. ... . 1..
““ii’“”
““‘i”“’
. ..
Fig. 4. Examples of the PSTHs in the stimulus-response interval for recognizable (top) and non-recognizable (bottom) stimuli in 8 different neuronal populations. Arrows indicate bins with significant (P -c 0.01) differences between mean count rates of evoked responses to recognizable and non-recognizable stimuli. Abbreviations: AV, anterior ventral nucleus; VP, posterior ventral nucleus; CM, centrum median of the thalamus, CC, corpus callosum. Other notations as in Fig. 3.
PSTH stimulus-response interval had long-latency components with the amplitude 20-40% of the background fragment mean count rate in 40% of the investigated neuronal populations (located in the anterior ventral, posterior lateral, reticular nuclei, centrum median of the thalamus and caudate nucleus). Slow count rate shifts with the maximum amplitude ranging from 20 to 50% of the background fragment count rate were observed in 45% of the investigated neuronal populations (in anterior ventral, posterior lateral, posterior ventral, ventralateral nuclei, centrum median of the thalamus, caudate nucleus and corpus callosum). No significant differences were found in any neuronal populations under study between the mean count rate in the background fragments followed by presentations of recognizable and non-recognizable stimuli. Neither did we record any significant differences between short-latency components of the evoked responses to recognition and non-recognition.
Significant differences between mean count rate of the neuronal populations for cases of recognition and non-recognition were recorded 300-500 ms after the stimulus presentation in 25% of the investigated neuronal populations (in the posterior ventral, posterior lateral, centrum median of the thalamic nucleus, in the corpus callosum and caudate nucleus). Half of the above neuronal populations were found to differ significantly within 300-800 ms after the stimulus presentation. The amplitude of the long-latency components of the evoked reactions in these neuronal populations appeared to be larger in cases of recognition as compared with cases of non-recognition (histograms 2-5 in Fig. 4). In the other half of the above mentioned neuronal populations significant differences between mean count rate for cases of recognition and nonrecognition were observed (histogram 7 in Fig. 4) between count rate slow shifts, the amplitude being larger in case of recognition as compared to nonrecognition. It is noteworthy that in the evoked NIA of some brain areas from one to three components could be observed. Thus besides the responsebound component, the PSTH showed a shortlatency component and a slow shift of the count rate in the neuronal population of the posterior lateral nucleus of the thalamic nucleus (histogram 1 in Fig. 4); in the neuronal populations of ventrolateral nucleus of the thalamic nucleus we observed simultaneously short and long-latency components (histogram 2 in Fig. 4). However only one of the above components was observable in the evoked activity of other neuronal populations (histogram 7 in Fig. 4). The same patient was found to have both neuronal populations with differences and without any for cases of recognition and non-recognition (histogram 8 in Fig. 4). As was mentioned in Methods the patients had polygonal shapes with and without meaning presented to them with the view to studying neuronal reactions reflecting semantic characteristic of visual stimuli. PSTH obtained in different patients for different neuronal populations had the same characteristic features as those elicited in the first series of investigations when the visual stimuli were presented on the perception threshold. Sig-
322
nificant differences between mean count rates of the neuronal populations for recognition of familiar and unfamiliar shapes were observed only 300-500 ms after the stimulus presentation. They were observable in 20% of all the investigated neuronal populations in the ventrolateral nucleus of the thalamus, in the globus pallidum, the caudate nucleus and the corpus callosum, in whose evoked activity long-latency components or count rate slow shifts could be recorded. The differences were in the amplitude of the long-latency components in these neuronal populations which were found to be larger on presentation of familiar shapes (histograms 2, 3 in Fig. 5), as for the count rate slow shifts, they were either recorded 300-600 ms later in case of unfamiliar shapes or their amplitude appeared to be smaller (histograms 5, 7 in Fig. 5). It is noteworthy that although long-latency components could be different for some neuronal populations on presentation of familiar and unfamiliar shapes, their phases remained unchanged (histograms 3, Fig. 5). Besides the neuronal populations in whose evoked activity we recorded significant differences between the late components
.
...iiiiiiii “”
..I
Fig. 5. Examples of the PSTHs in the stimulus-response interval for familiar (top) and unfamililar (bottom) shapes in 8 different neuronal populations. Abbreviations: T COT,temporal cortex. Other notations as in Figs. 3 and 4.
of evoked reactions for familiar and unfamiliar shapes, the same patient was found to have neuronal populations whose long-latency components and the count rate slow shifts of the evoked reactions did not differ for two varients of trials (histograms 1, 4, 6, 8 in Fig. 5).
DISCUSSION Multiphase count rate changes during the presentation of visual stimuli were shown by our investigations to evolve in the neuronal populations of different nuclei of the thalamus and the striopallidar system, i.e. in the structures not involved in primary visual analyzer areas. Count rate changes correlating in time with the patient’s verbal response were also recorded in those neuronal populations. These data confirm our concept concerning the polyfunctional nature of many neuronal populations in the human brain based on studying the oxygen dynamics, impedance and some other indices of the brain’s activity during various psychological trials (Bechtereva, 1971; 1978; 1980). In analyzing the data, of special interest were similar responses in different distant populations of the brain obtained from the same and different patients, which proves that there must be some typical neuronal mechanisms maintaining the performance of the psychological trials. Similar data indicative of the similarity of reactions in the neuronal populations of different brain structures in animals were obtained in studying the impulse activity and evoked potential during various behavioural acts (John, 1972; Shvyrkov, 1978). Short- and long-latency components as well as count rate slow shifts evolved in the evoked reactions of the neuronal populations of the human brain when visual stimuli were presented. These components could be recorded simultaneously in the NIA evoked changes of the same brain area. We also revealed neuronal populations with only one of the above components recorded in the count rate dynamics. None of the investigated neuronal populations had different shortlatency components for cases of recognition and non-recognition. However, the long-latency components and the count rate slow shifts, at least for
323
some neuronal populations, did differ for these two cases of recognition trials. The difference was as follows: when the patient recognized the pattern, the amplitude of the long-latency components and the count rate slow shifts appeared to be larger than in the reverse case when the pattern was not recognized by the patient. Note should be made that none of the neuronal populations revealed any significant differences between the count rates in the background segments. Thus the original functional states of the brain which served as background for the visual stimulus presentation did not differ, according to the count rate dynamics data, when the patient recognized the pattern and when he failed to do so. Some of the above mentioned neuronal populations had NIA that changed differently whether or not the patient recognized the pattern, whereas in other adjacent brain areas there occurred neuronal populations which had statistically identical late components of evoked reactions whatever the result of the trial. Thus differences in the long-latency components and count rate slow shifts of evoked reactions, depending on the subjective stimulus estimation, are characteristic of some neuronal groups only, which confirms our concept that it is the system determining the optimal use of the brain’s potentialities and not individual areas or the whole brain that maintains the mental activity (Bechtereva, 1983). The detection of the signal common to both variants of trials for recognition suggests that the short-latency components, which always occur whether or not the pattern is recognized, are actually related with signal detection. On the basis of our investigations we may assume that the long-latency components and the count rate slow shifts can be correlated with mental processes of comparing the input information with the memory engram and decision making. For this assumption to be varified in conditions different from the first test, investigations were carried out when patterns with or without semantic meaning were presented not on the recognition threshold but during the period sufficient for their recognition. The analysis of PSTHs estimated for recognition of semantically meaningful and meaningless stimuli suggests that the same components of the
evoked NIA as in the previous investigations may be observable. None of the studied neuronal populations showed any differences in the mean count rates in the background fragments that preceded familiar and unfamiliar patterns. Neither did we record any significant differences in the mean count rates of the neuronal populations in the NIA record fragments corresponding to shortlatency components. The latter fact is ample evidence that short-latency components are not stimulus meaning bound, but they reflect physical characteristics of the signal which in our test were statistically identical for both familiar and unfamiliar patterns. In some neuronal populations of the investigated brain structures the long-latency components and slow count rates were found to differ considerably when familiar and unfamiliar patterns were presented. The differences were in the long-latency components and count rate slow shifts essentially increasing when the patient had a familiar pattern presented to him. In addition to the above assumption concerning the functional meaning of the evoked NIA late components, the data of these investigations suggests that the late components are not stimulus bound but are related to the stimulus semantic characteristics and its subjective estimation by the patient. Three orthogonal components: N 100, P 300 and the potential slow positive shifts were separated in the course of investigating ER recorded from the scalp of healthy subjects by the principal components method. The N 100 components was also shown to be related to the stimulus detection, whereas the P 300 component and the potential slow positive shifts were related to recognition processes and the decision making (Parasuramen et al. 1982). Thus data concerning the NIA count rate dynamics in the subcortical formations of the human brain during the visual stimulus recognition show good agreement with the investigations of the evoked potentials recorded from the scalp of healthy subjects in the test for audial stimulus recognition. REFERENCES Anichkov, A.D. (1977) Stereotaksitcheskij apparat dlija wedenija dolgosrotchnych mnoshestvennych vnutrimozgovych elecktrodov. Fuiol. tcheleueka, 3: 372-375.
324
Bates, Y.A.V. (1961) Depth electrodes in the human subject. Basic technical, interpretative and ethical considerations. Int. Congr. Ser., 31: 62-64. Bechtereva, N.P. (1971) Neirophysiologitcheskie aspektl psichitcheskoi dejntelnosty tcheloveka, Nauka, Leningrad, 119 PP. Bechtereva, N.P. (1978) The Neurophysiological Aspects of Human Mental Activity, 2nd end., Oxford Univ. Press, New York, 181 pp. Bechtereva, N.P. (1980) Zdorovyi i Bolnoi Morg Tcheloveka, Nauka, Leningrad, 208 pp. Bechtereva, N.P. (1983) Preface to the book by Gogolitsin, Yu.L. and Kropotov, Yu.D. Issledovanre Tchastoty Razjadov Neironov Mozga Tcheloveka, Nauka, Leningrad, pp. 3-7. Bechtereva, N.P., Gratchov, K.V., Orlova, A.N. and Jatsuk, S.L. (1963) Ispolzovanie mnoshestvennych electrodov, vzivlennych v podkorkovie obrazovanija golovnogo mozga tcheloveka dlija lechenia giperkinezov. J. nevropatol. psichiatr., 1: 3-8. Bechtereva, N.P., Bundzen, P.V. and Gogolitsin, Yu.L. (1977) Mozgoue Cody Psichitcheskoi DeJatelnosty, Nauka, Leningrad, 165 pp. Bechterva, N.P., Gogolitsin, Yu.L., Ilyukhina, V.A. and Pakhomov, S.V. (1983) Dynamic neurophysiological correlates of mental processes. Int. J. Psychophyslol., 1: 49-63. Bickford, R.G., Dodge, H.W., Peterson, M.C. and Sem-Jocobsen, C.W. (1953) A new method of recording from subcortical regions of the human brain. Electroenceph. clin. Neurophysiol., 5: 464. Brown, W.S., Marsh, J.T. and Smith, J.C. (1973) Contextual meaning effects on speech-evoked potentials. Behav. Biol., 9: 755-161. Danko, SC. and Kaminskij, Yu.L.. (1982) Sistema Technitcheskych Sredstv Neirofiziologitcheskych IssledovanQ Tcheloveka, Nauka, Leningrad, 135 pp.
Gogolitsin, Yu.L. and Kropotov, Yu.D. (1983) Issledooanre Tchastoty Razrjadov Neironov Morga Tcheloveka, Nauka, Leningrad, 120 pp. Gretchin, V.B. (1972) Nalichnij kislorod golovnogo mozga tcheloveka. In Klinitcheskaja newophyslologia, Nauka, Leningrad, pp. 494-530. Glezer, V.D., Leushina, L.I., Nevskaya, A.A. and Prazdnikova, N.V. (1974) Studies on visual pattern recognition in man and animals. Vuion Res., 14: 555-583. Halgren, E., Babb, T.L. and Crandall, P.H. (1978) Activity of human hippocampal formation and amygdala neurons during memory testing. Electroenceph. clin. Neurophyslol., 45: 585-601. Ilyukhina, V.A. (1977) Medlennye Bioelektritcheskie Protsessy Goloonogo Morga Tcheloueka, Nauka, Leningrad, 184 pp. John, E.R. (1972) Switchboard versus statistical theories of learning and memory. Science, 177: 850. John, E.R. and Schwarts, E.L. (1978) The neurophysiology of information processing and cognition. Ann. Rev. Psyrhol., 29: l-29. Neisser, U. (1966) Cognitive Psychology, Appleton Century Crofts, New York, 351 pp. Nickerson, R.S. (1972) Binary-classification reaction time. A review of some studies of human information-processing capabilities. J. Psychon. Monogr. Suppl., 4: 275-318. Parasuraman, R., Richer, F. and Beatty, J. (1982) Detection and recognition: concurrent processes in perception. Percept. Psychophys., 31: p. l-12. Shvyrkov, V.B. (1978) Neirofiziologitcheskoe Izutchenie Systemnich Mechanirmoo Pooedenia, Nauka, Moscow, 239 pp. Smith, E.E. (1968) Choice reaction time: an analysis of the major theoretical position. Psychol. Bull., 70: 77-110. Walter, W.G., Crow, H.J. (1961) Depth recording from human brain Int. Congr. Ser. 37: p. 64-65.