Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.psy.msu.ru/people/danilova/danilova1982.pdf Дата изменения: Tue Dec 2 14:26:26 2014 Дата индексирования: Sun Apr 10 01:27:29 2016 Кодировка:

Activation Dynamics in Learning R. Sinz & M. R. Rosenzweig (eds.): Psychophysiology 1980 VEB Gustav Fischer Verlag Jena (GDR) and Elsevier Biomedical Press, Amsterdam (Netherlands) 1982

407

The Activation Dynamics in the Learning Process and its Reflection in VEP
N. N. Danilova
Department of Psychology, Moscow State University, Moscow, USSR

Sum mary
The dynamics of learning to identify the target interval among similar ones was investigated in two groups of subjects (5 persons in each) with different strength of nervous processes according to J. Strelau's questionnaire. It was proved that in the group of strong subjects a higher level of identification efficiency estimated accorded to the sensitivity index (d') correlates positively with positive shifting in VHP of the occipital lead. As for the group of weak subjects, a lower level of identification efficiency of the target interval corresponds to a higher level of decision criterion and a higher negative amplitude of N 150 in VEP of the vertex lead.

Introduction

A large number of cognitive operations connected with sensory information processing such as perception, retrieval of information from memory, decision making, recording of information in memory etc. are expressed in a change of average evoked potentials (Dustman & Beck, 1963; Kooi & Bagchi, 1964; Kooi et al. 1972; Sutton et al. 1965; Eason et al. 1966; Karlin, 1970; Ivaniz-ki, 1974; Dyomina, 1975; Hillyard et al. 1971; Paul & Sutton, 1972, 1973; Picton & Hillyard, 1974; Squires et al. 1975; Naatanen, 1975, Naatanen & Michie, 1979). There are some reasons to believe that cognitive operation is controlled by its own separate nonspecific system of activation and under certain conditions can be regulated independently of the others.
In this investi gati on we studied acti vati on pr ocesses connected with memor y mechanis m and with the formation in memor y of the target stimulus model and its reflection in visual evoked potentials (VEP) in particular. To perform it we used the method of identification of a target stimulus from a number of similar ones. Interstimulus intervals in the range fr om 1 t o 3 sec. were used as stimuli.

As a rule the subjects have no well formed models of interstimulus intervals of this range. This made it possible to follow the dynamics of target model formation in memory and its reflection in VEP. Time intervals were formed by means of rhythmic light flash blocks at four frequencies: 0.35; ' 0.5; 0.6; 0.8 c/sec. The target block was presented up to 10 times before the beginning of the experiment. During the experiment the subject did not get any feed back confirming his correct choice. The experiments were performed with two groups of subjects differing in the strength of their nervous system according to J. Strelau's questionnaire. We supposed that the efficiency of learning to differentiate between the time intervals would depend on individual characteristics. Task performance efficiency was evaluated in two ways: by routine method-by a number of errors as such and in the terms of signal detection theory: with the help of the sensitivity index (d') and decision-making criterion of the observer. We suppose that the sensitivity index shows accuracy and stability of the target model in memory (Klatzky, 1975).


408 Methods

N.N.Danilova

The experiments were carried out in two groups of subjects (5 persons in each) with "strong" and "weak" nervous systems. Two experiments with a week interval between them were conducted with each subject. The first experiment consisted of three sessions: session I included 4 types of light flashes in blocks that were presented at random with the same probability and one of the four frequencies was 0.35; 0.5; 0.6; 0.8 c/sec. Each block consisted of 10 flashes, being presented over 10-15 sec intervals. In the session II the instruction was given to differentiate the blocks by the frequency of the flashes and react with the hand movements to the blocks equal to the target, presented before session II. In session III another instruction to react with the hand movements to blocks different from the target was used. The second experiment was similar to the first one but without session I and the target presentation. It included sessions IV and V. Occipital and vertex EEG with monopolar leads, EMG of right hand flexors and ECG were recorded. The EP averaging was done by the LP 4050 analyser, the number of summations being equal to 100. We averaged the EP separately for each of the 4 light flash frequencies in each session. The learning efficiency was estimated by the number of correct responses and by d' and Zj/io according to detection theory.

Results
All the subject s encountered great diffi culties when the y attempt ed t o identif y the tar get among the presented blocks of light flashes. In fact, none of them was able to cope with the task unerringl y, either in the first, or in the second experi ment.

Figure 1. Dynamics of sensitivity index (d') and decision criterion (Zj/io) in the process of training to differentiate and to identify time intervals in the group of strong (continuous line) and weak (dotted line) subjects. The abscissa axis-number of the session of the experiment.


Activation Dynamics in Learning

409

Nevertheless, both groups of subjects learning to differentiate time intervals. The group of the strong subjects is characterized by a greater speed of learning and a higher efficiency in performing the task in the beginning and as well as in the end of the experiment. The mean number of errors for the group of strong subjects was significantely lower than the mean number of the errors in the weak subjects. The estimation of task performance efficiency, according to the sensitivity index (d') and decision making criterion gave the following results: The dynamics of a change of d' and Zj/io for strong and weak subjects in the process of learning is depicted on Fig. 1. Fig. 1 shows that the degree of efficiency of task performance in a weak group (dotted line) according to d' is very low and becomes 1.38 times as much towards the end of training. As for a strong group (continuous line) during the experiment d' becomes 3.14 times as much. This testifies to the productivity of learning process of these subjects. In the strong group two processes take place simultaneously: the increase of d' and gradual decrease of decision criterion (Zj/io). That means that in the strong group the increase of sensitivity index is determined by both the growth of the number of hits and the decrease in the number of false alarms. The results in the weak group are absolutely different. The sensitivity index does not change during the experiments. The decision criterion used by the weak subjects is always higher than in the strong group. It varies from session to session, reaching the lower level only in the second experiment (IV, V session). In the weak group the higher level of decision criterion indicated a great number of false alarms, typical of weak subjects. Only in the weak group the high positive correlation is observed between the number of hits and the number of false alarms (r = 0.71, p<0.001). Thus in the weak group the dynamics of changes in the number of hits is determined mainly by shifting the criterion of decision but not by the sensitivity index. So, weak and strong groups differed not only in the level of efficiency of task performance mea sured by the number of errors and sensitivity index, but also in different strategies of performance according to decision criterion.

Figure 2. Dynamics of the VHP transformations for the occipital (OJ and vertex (C2) leads of the "strong" subject observed during two experiments on presenting the target stimulus (0.5 c/sec). Roman numerals show the number of s ession. The duration of the VEP-512 ms ec. Numb er of summation-100. Horizontal lines --- baseline calibration signal-5v.


410

N.N.Danilova

In the weak group the amplitude of negative component (N150) in vertex VEP becomes higher with the growth of the number of false alarms (z = --Q.52, p = 0.018). Correct and wrong identification of target stimulus is accompanied by the growth of the amplitude of negative component (N150) (r = --0.52, p = 0.0.3). Meanwhile decrease of negativity of-negative component N150 is characteristic of making decisions about mismatching with target. Thus, the direction of N150 shifting referred to baseline is determined by the character of decision made concerning the identification of the stimulus namaly by evolution of the stimulus as the target.
No component in vertex and occipital VEP is correlated with sensitivity index in the weak group. In the strong group correlation between sensitivity index and amplitude of negative component (N150) in occipital VEP was discovered. The growth of sensitivity index causes a shifting of the negative component (N150) in the positive direction (r = 0.35, p = 0.1).

Comparing certain changes of VHP and infor mation pr ocessing i n the strong and weak gr oups we got the following data:

The positive correlation may be observed between the positive components (P200) and the whole number of correct and wrong decisions about the identification of the target stimulus. In this case, the identification of target stimulus is accompanied by the increase of the amplitude of P200. It is importent to mention that making an opposite decision brings about the decrease of the amplitude P200.
It is of no i mportance whether the de Fi g. 2 s hows t he d ynami cs of VEP One can see that positi ve shifting in oc msec is the most characteristic feature differentiate time inter vals. ci si on is cor rect or not. of t he str ong subject in the pr ocess es of t wo experi ments. cipital VEP occuring before 150 msec and lasting after 200 of VEP changes taking place in the pr ocess of learning t o

Discussion

So, weak and strong subjects perform the task of differentiating between time intervals differently. While performing a task weak subjects faced a lot of difficulties from the start. They have an indistinct, vague target model of interval unlike strong subjects. Almost all the intervals were considered by them to be the target one. This stage of task performance resembles the generalization stage of a conditioned reflex. It is characterized by a high level of activation and orienting reflex. A great number of false alarms indicate this fact, which is seen by some authors as the result of the growing level of arousal. Under such conditions, decision making is characterized by a high level of uncertainty. In the process of training the information making the target model more accurate is recorded in memory, but it is very often false. As a result, the process of target model formation of the interval, in fact, has no chance to develop. A subj ect facing some hazards has to choose other strategies and is to use stricter criterion of decision making. That phenomenon was observed in the group of weak subjects in the second experiment. Thus, a subject starting to act more cautiously does not identify all the intervals as target ones. New strategies change the correlation between the number of hits and false alarms in favour of the former. Objectively, it gives better opportunity for more accurate target model formation in memory. As a result of sessions IV and V, the weak subjects showed the growth of sensitivity index (d') which is accompanied by the decrease of decision criterion (Zj/io). A high correlation of the amplitude of negative component (N150) in the vertex VEP with the number of false alarms makes it possible to regard it primarily as the index of a subject's activation. However, the amplitude of the negative component (N150) depends also on the specific


Activation Dynamics in Learning

411

character of the decision made. The growth of negativity is the result of referring the interval to the target model. The decision about mismatching the target model causes the decrease of negativity. Consequently, negative shifting and the growth of the amplitude of negative component (N150) are the result of identification of the target stimulus when a subject makes a decision on identifications with a high level of uncertainty, which brings about the negative emotional reaction. That fact was mentioned by Naatanen & Michie (1979) who wrote that negative component (N1) invariably comprises a stress element. A higher level of efficiency is typical of strong subjects during task performance. It is proved by the fact that strong subjects have a higher initial level of sensitivity index and that this parameter grows in the process of training. Strong subjects have a more accurate and distinct target model of interval; during the experiments it becomes more concrete gradually. The process of learning in the group of strong subjects is characterized by positive shifting in occipital VEP. Positive shifting is found in the place of N150 and P200. It manifests itself in decreasing negativity of N150 and in the growth of positivity of P200 referred to baseline. According to the data obtained during the experiments in the group of the strong subjects, positive shifting of the negative component correlates positively with the sensitivity index and the number of hits. Meanwhile, the growth of positivity of positive component (P200) was observed only in the case of identification of the target interval, including all the correct and wrong decisions. All the above-mentioned facts make it possible to consider positive shifting in the VEP to be the index operation of recording in memory the information about the target interval. Thus the more a subject is confident of the decision made the earlier the process of information recording begins. So the amplitude of negative component (N150) showing the level of activation of a subject is substituted for or disguised by its positive shifting. In these cases, when a subject doubts and is not sure of the decision made, the operation of information recording in memory is performed with greater delay-about 20 msec and even more. This operation is characterized by the growth of the amplitude of the positive component (P200). Thus, the peculiarities of the process of learning to differentiate time intervals are reflected in specific modifications of the VEP phases in the group of the weak and strong subjects. It is supposed that positive shifting in the VEP is the index of activation of memory operations dealing with recording of the processed sensory information.

References
Dustman, R. E. & E. C. Beck: Long-term stability of visual evoked potentials in man. Science. 142,1480-1481 (1963). Eason, R. G., M. R. Harter & C. T. White: Effects of attention and arousal on visual evoked cortical potentials and reaction time in man. Physiol. Behavior. 4, 283-289 (1969). Hillyard, S., K. Squ ir es, I. Bau er, & P. Lindsay: Evok ed p ot entia l cor relates of audit or y s ignal d et ect ion. Science, 772,1357-1360 (1971). Karlin, L.: Cognition, preparation and sensory-evoked potentials. Psychol. Bull. 70, 122-136 (1970). Kooi, K. A. & B. K. Bagchi: Visual evoked responses in man: normative data. Ann. N. Y. Acad. Science. 722 (1), 254-269 (1964). Kooi, K. A., B. K. Bagchi & R. N. Jordan: Observations on photocally evoked occipital and vertex waves during sleep in man. Ann. N. Y. Acad. Science. 112 (1), 270-280 (1972). Naatanen, R. : Select ive attent ion and evok ed p otent ia ls in humans-a cr it ica l r eview. Biol. Psychol.2, 237307(1975). Naatanen, R. & P.T. Michie: Early selective-attention effects on the evoked potential: a critical review and reinterpretation. Biol. Psychol.S, 81-136 (1979). Paul, D. D. & S. Sutton: Evoked potential correlates of response criterior in auditory signal detection. Science 777, 362-364 (1972).


412

N. N. Dan ilo va

Paul, D . D . & S.Sutton: Evoked potential correlates of psychophysical judgments: the threshold problem. A new reply to Clark, Butler, and Rosner. Beha v.Biol. 9, 421^33 (1973). Picton, T. W. & S. A. Hillyard: Human auditory evoked potentials. II. Effects of attention. EEG clin. Neurophysiol.36, 191-199 (1974). Poon, L. W., L. W.Thompson & G. R. March: Average evoked potential changes as a function of processing complexity. Psychophysiol. 13, 43^9 (1976). Squires, K. C., N. K. Squires & S. A. Hillyard: Vertex evoked potentials in a rating-scale detection task: relation to signal probability: Behav. Biol. 13, 21-34 (1975). Su tt on , S. , M. Br a r en & J. Lub in : E vok ed p ot en t ia l c orr ela t es of sti mu lu s un cert a in t y. Sc i enc e 15 0, 1187-1188.