O. K. Tikhomirov THE PSYCHOLOGICAL CONSEQUENCES OF COMPUTERIZATION* Editor’s Introduction In this paper Tikhomirov argues that we have often misconstrued the connection between computerization and human activity. The way he tries to rectify this is by going back to Vygotsky’s notations about the importance of sign systems in mediating higher mental functions. Before getting into the details of how he analyzes the role of the computer, Tikhomirov reviews some of the other approaches that have been used. First he deals with the “theory of substitution”. According to this theory, the computer’s role is to replace the human in intellectual spheres. Although Tikhomirov agrees that in the case of certain types of problems the computer can have the same input and output as a human being, he rejects the notation that this means that a program of a computer’s work is a theory of human thinking. Next he deals with the “theory of supplementation”, which argues that the computer’s role is to increase the volume and speed of human processing of information. According to Tikhomirov, this approach relies on our ability to make a formal analysis of human mental processes in terms of information theory. The main objection he raises against the theory of supplementation is that it assumes that the inclusion of a computer in human activity does nothing more than provide a purely quantitative extension of existing activity. Tikhomirov argues that both of these approaches are misguided because they fail to understand the essential role of mediation in human activity. He claims that if we view computerization in terms of the Soviet idea of mediation, we will begin to see it in a completely new light – that is, he maintains that the real question is not how the computer can replace mental processes or how it can make a purely quantitative addition to already existing psychological processes, but rather that computer programs should be viewed as a new kind of sign system that can mediate human activity. As Vygotsky and his followers have pointed out, when a new form of mediation is introduced into activity, it does not simply expand the capacity of the existing activity but often also causes a qualitatively new stage to emerge. Just as Vygotsky argued that a new form of mediation (speech) gives rise to a qualitatively new stage of thinking in ontogenesis, in this paper Tikhomirov is arguing that a new form of mediation (computerization) gives rise to a qualitatively new stage of thinking in history. (He also connects it with ontogenesis.) In building his argument Tikhomirov uses some of the tenets of the theory of activity introduced by Vygotsky. For example, he relies on the historical approach (one form of genetic explanation) and the notion of the tool as the most important component of human activity; and he utilizes several of the concepts and distinctions Vygotsky proposed in connection with his analysis of sign systems, the distinction between sense [smysl] and meaning [znachenie], among others. J. V. W. * From O. K. Tikhomirov (Ed.), [Man and computer]. Moscow: Moscow University Press, 1972. 1 Among the new theoretical problems that have confronted psychology in the course of the scientific-technological revolution is study of the psychological consequences of computers. Does the computer affect the development of human mental activity? If so, how? In order to answer these questions, we must compare how human beings and computers solve the same problem. Such an analysis allows us to establish whether or not human activity is reproduced in the computer. First, let us deal with the computers that have already been created; with regard to future models, I shall limit myself to evaluating concrete schemes for perfecting functional potential. In the last few years, the analogy between thought (and the behavior of organisms in general) and the working principles of computers has come to be used widely. Special significance has been given to so-called heuristic programs (Simon, Newell). The very term heuristic is a reflection of a certain stage in the development of the theory of programming problems for the computer. It designates any principle or system that contributes to decreasing the mean number of steps needed to make a decision. Heuristics are mechanisms that guide a search so that it becomes more selective, and therefore efficient. It is important to point out that this interpretation does not correspond to the broader meaning of the term heuristic, which is attributed to Papp and means “the skill to solve problems”, This latter meaning, which had very indefinite content, was used until the sciences that study human thinking were differentiated, and referred to devices of analysis and synthesis. The possibility of solving with a computer problems that earlier had been solved by humans led several scholars to the conclusions that: 1. A program of a computer’s work is a theory of human thinking. 2. The possibility of reproducing some functions on a machine is the criterion of the correctness or incorrectness of a truly psychological explanation of activity. 3. A negative response to the traditional question “Can a machine think?” is “unscientific and dogmatic,” since a comparison of the behavior of machine and human being often reveals identical results. On the basis of these ideas, the influence of computers on intellectual activity is viewed as follows. The computer replaces the human being or substitutes for him/her in all spheres of intellectual work. This is the theory of substitution. However, in order to check the validity of this theory of the influence of computers on human mental processes (a theory that begins with the assumption that the heuristic program reproduces human creative thinking), we must analyze how closely human processes for directing a search for the solution of a problem correspond to those used by computer in performing the same task. As has been shown in our laboratory, these processes are not the same. A large part of the control mechanisms of search in humans in general are not represented in existing heuristic programs for computers. When computer heuristics do resemble human ones, they are significantly simpler and are comparable in some essential way. Computer reproduction of some externally observed results of human activity has been carried out without reproducing human heuristics. On the basis of the data collected in the course of experimental psychological investigations, we can state that the idea of substitution does not express the real relations between human thought and the work of the computer. It does not accurately represent how the latter influences the development of the former. One can hardly determine how computers influence the development of human mental processes without considering what human thought is and what important 2 historical stages in the development of thinking can be identified up to the time of the appearance of computers. We approach the problem in this way in order to examine the question of computerization in a broader historical perspective – the perspective of the development of human culture. In recent years the information theory of thought has become very popular. We think it necessary to contrast this theory with the psychological theory of thought (they are sometimes incorrectly viewed as being identical). The former theory is often formulated as a description of thought at the level of elementary information processes and is concerned primarily with the characteristics of information processes. The information theory of thought consists of the following ideas. Any behavior, including thought, can and must be studied relatively independently of the study of its neurophysiological, biochemical, or other foundations. Although the differences between the brain and the computer are evident, there are important functional similarities. The idea that complex processes of thought consist of elementary processes of symbol manipulation is the main premise of the explanation of human thought at the revel of information processing, in general, these elementary processes are described as follows: Read the symbol, write the symbol, copy the symbol, erase the symbol, and compare two symbols. It is not difficult to see that the “elementary informational processes” or “elementary processes of symbol manipulation” are nothing but the elementary operations in the operation of calculator. Thus, the desire to study thought “at the level of elementary informational processes” is actually interpreted as a demand to explain human thought exclusively within a system of concepts that describe the operation of a calculating machine. The fundamental concepts within this framework are: (1) information, (2) the processing of information, and (3) the information model. Information is in essence a system of signs or symbols. The processing of information deals with the various types of processing of these symbols according to given rules (“symbol manipulation,” as some authors refer to this). The information model (or “the space of the problem,” as opposed to the environment of the problem) is the information about the problem represented or collected (in the form of a coded description) in the memory of the system that solves the problems. A complicated, but final and fully defined, set of rules for processing information was seen as forming the basis for the behavior of thinking human. This became, as it were, the position differentiating “scientific” approaches from “nonscientific” ones (i.e., those tolerating “mysticism”). If we accept the information theory of thought (i.e., the description of thought as analogous to the work of the computer), there can be one answer to the question of how computers affect human thought: computers supplement human thought in the processing of information, increasing the volume and speed of such processing. This point of view can be labeled the theory of supplementation. Within the framework of the theory of supplementation, the relations between the functioning of humans and the computer, if combined into one system, are the relations of two parts of one whole – the “processing of information.” With the aid of the computer, humans process more information faster, and perhaps, more accurately. A purely quantitative increase in their resources takes place. Inasmuch as the theory of supplementation is directly connected with the information theory of thought, it is necessary to examine it in more detail. Psychologically, what does thought mean? Does the informational approach describe actual processes of human thinking, or has it abstracted from it those characteristics that are most essential? We shall not find the answers to these questions through “modeling” mental processes, but through the theoretical and experimental analysis of thought processes. 3 Psychologically, thought often emerges as the activity of solving a problem. The problem is usually defined in terms of a goal in a certain situation. However, the goal is not always “given” initially. Even if it is externally imposed, it is often quite undefined and allows for complex interpretation. Therefore, the formulation and attainment of goals are among the most important manifestations of thinking activity. On the other hand, the conditions is which a goal is formulated are not always “defined”. It is still necessary to distinguish them from the general situation of activity on the basis of orientation or an analysis of the situation. The problem of a given goal in defined conditions still must be analyzed. Consequently, thinking is not the simple solution of problems: it also involves formulating them. What enters into the conditions of a problem? In other words, with what do humans who are solving a problem have to concern themselves? If we examine cases of so-called visual-active thought, it may be real objects or things, and/or it may be people. In the case of verbal thinking, it may be symbols. Is it sufficient to say of human verbal thought that it “operates with signs” in order to express the essential aspect of thinking? No, this is not sufficient. Following Vygotsky, in the analysis of verbal thinking I shall distinguish the sign itself, its referent, and its meaning. In “operating with signs,” human operate with meanings. In the final analysis, they operate with objects of the real world through meanings. Thus, if we describe human thought only as manipulation by means of signs, we are extracting and focusing on a single, isolated aspect of the thinking activity of a real person. This is precisely what the information theory of thought does. Actual objects or named objects that enter into the problem have important characteristics such as values. Actions with these objects (i.e., transformations of a situation) also have different values. There are different sources of the formation of the values of the same element of a situation and different interrelations among these values. The formal representation of the conditions of a problem (for example, in the form of a graph or list of signs), which reflects some reality, is artificially isolated from such accompanying objective situational characteristics. The conditions of the problem may include factors such as (1) the correlation of various values of elements and means for transforming the situation, and (2) the intention of the formulator of the problem. These characteristics, which are lost in a formal representation, not only exist but even determine (sometimes more than anything else) the flow of activity in solving the problem. Thus, the psychological and informational characteristics of a problem obviously are not the same. Mental activity often results in humans’ generating signs (for example, in identifying the actions that lead to a goal). However, these signs have a definite meaning (for example, embody the principle of the actions) and value. For someone solving a problem the meaning of a sign must be formulated, and the value emerges as an appraisal. Psychological examination of human intellectual activity must include an analysis of the following areas: (1) the operational sense* of the person solving the problem, (2) the sense of concrete attempts at solving the problem, (3) the sense of reinvestigating the problem, (4) the sense of various elements of the situation as opposed to their objective meaning, (5) the emergence of the senses of the same set of elements of the situation and of the situation as a whole at different stages of problem solving, (6) correlation of the non-verbal and verbal senses of the various types of formation in the course of problem solving, (7) the processes of the interaction of sense formation, (8) Tikhomirov’s notion of “sense” [smysl] is basically the same as that of A. N. Leont’ev and V. P. Zinchenko. See the footnote on this term in the paper in the volume by Zinchenko & Gordon. – J.V.W. * 4 the role of sense formation in the organization of research activity in the definition of its range (selectivity) and directionality, (9) the process of the emergence and satisfaction of search requirements, (10) change in the subjective value or meaningfulness of the same set of elements of a situation and actions expressed in the change of their emotional coloring (with constant motivation), (11) the role of a changing scale of subjective values in the organization of the flow of a search, and (12) the formation and dynamics of the personal sense of the situation of a problem and its role in the organization of activity in problem solving. In human mental problem-solving, such real functional forms as sense (operational and personal) and the values of the objects for the problem solver are not simply neutral with regard to the informational characteristics of the material; rather, they take part in the processes of directing the activity of problem solving in an a important way. It is this great importance that above all creates the qualitative distinctiveness of mental activity in comparison with information processing. This is what differentiates the psychological and the informational theories of thought. Thus, we cannot accept the theory of supplementation in our discussion of the problem of the influence of computers on the development of human intellectual activity, inasmuch as the informational approach on which it is based does not express the actual structure of human mental activity. It is impossible to discuss the problem of the influence of computers on the development of human mental processes without talking into account research on artificial intelligence. In cybernetics it is usually the case that not only narrow, specialized problems but also general, theoretical ones are dealt with. Analyzing how these problems are interpreted is a required step not only for evaluating the contribution of cybernetics to the establishment of a world view but also for predicting the development of certain areas of this science. In the course of its development, cybernetics, which Weiner understood as a theory of guidance and connections in the living organism and machine, was divided into a number of areas: self-regulating systems, the modeling of human thought, and artificial intelligence. Currently, the third area has become the leading one in a number of countries. This has happened because, over the course of its development, the first area encountered significant difficulties and the second and third in fact were combined. Artificial intelligence is not simply a theme for novels. It is a scientific trend also requires thorough analysis from the psychological point of view. I shall look at the interpretation of this trend that can be found in the works of Minsky, MsCarthy, Simon, and other investigators who have developed positions on the issue. The most widespread definition is the following: Artificial intelligence is the science whose goal is to develop methods that will enable machines to solve problems that world require intelligence if they were solved by humans. At the same time, the expression artificial intelligence is often used to signify human functional possibilities: a machine is intelligent if it solves human problems. Initially, in the theory of programming problems for computers there was a distinction, even opposition, between two scientific courses – “artificial intelligence” and “modeling mental processes.” The differentiation was along the following lines: The first course involved programming problems for the computer without regard to how these problems were solved by humans; the second course proposed programming with an attempt to duplicate human means of solving problems in machine programs. After a time, the boundaries between these two areas became practically nonexistent. However, the term artificial intelligence is significantly more popular than the second, modeling mental processes. Investigators 5 concerned with modeling have expressed their dissatisfaction with the term simulation since it implies an imitation (i.e., a purely external similarity of two objects that does not reflect the scientific aspirations of the authors). On the other hand, investigators concerned with artificial intelligence have emphasized that machine programs eventually must be based on an account of human problem-solving. Thus artificial intelligence now is interpreted differently than it was five to tem years ago. Making machine methods of problem solving approximate human ones is the strategic goat of artificial intelligence research. It is often said that are no longer any restrictions on the possibilities for duplicating human abilities by machine programs. A strategy that has been thus formulated touches on a series of fundamental problems. It is thought, for instance, that a positive answer to the traditional question “Can a machine think?” must be linked with a materialist world-view (Borko). Neisser expresses a similar idea: in his opinion the analogy between humans and computers is based on materialism. It is sometimes argued that the creation of artificial intelligence will influence human beings’ ideas about themselves and destroy the illusion of their “uniqueness.” With regard to the formation of a world view, this influence will be even more significant than understanding “the place of our planet in the galaxy” or the laws of “evolution from more primitive forms of life.” Other authors write about “undermining human beings’ egocentric conception of themselves.” At the same time, we are already beginning to see predictions of the peculiarities of the “world view” of future machines. In Minsky’s opinion, when intellectual machines are constructed, we must not be surprised to find that they, like humans, will have false ideas about mind matter, consciousness, free will, etc. It is sometimes thought that the creation of intelligent machines will shed light on the eternal mind-body problem and the role of humans in the universe. In Slagle’s opinion, the existence of intelligent machines will support the “mechanistic conception” that humans are only machines, and the answer to the psychophysiological problem supposedly will be that only the body exists. Thus, the position of mechanistic materialism is sometimes explicitly formulated as the methodological foundation of research in artificial intelligence. No distinction is made, however, between mechanistic and dialectic forms of materialism. The second is ignored, and the first is proposed as the sole form of materialism. What can one say about the influence of computers on the development of human mental processes from a mechanistic point of view? If “only the body exists” and humans are “only machines,” then at best one can imagine the “synthesis of two machines “ or the substitution of one machine for another. Consequently, in research on artificial intelligence we again run into theories of supplementation and substitution. The creation of artificial intelligence in past years relates to the future. To appraise the probability of achieving the strategic goal of this program means simultaneously appraising the theories of supplementation and substitution in another context. When humans solve certain problems, what, exactly, is most important? Which component or mechanism must one, as it were, try to transfer to machine programs? Specialists in artificial intelligence answer as follows: We must give the computer the system f concepts necessary for the solution of problems of a given class. We must give machines more “semantic information.” The interest in semantic problems is, to a significant degree, based on critical reinterpretation of unsuccessful experiences in machine translation. In Minsky’s opinion, the machine needs to acquire knowledge on the order of 100,000 bytes in order for in to be able to act adequately in relation to a simple situation. A million bytes, 6 along with the proper corresponding organization, would be sufficient for “powerful” intelligence. If these arguments do not appear adequate, it is suggested that these figures be multiplied by tem In analyzing an artificial intelligence program it is necessary to clarify what meaning is given to such terms (often used in philosophy and psychology) as thought, knowledge, and solution of a problem. Usually, in the context of research in artificial intelligence, these terms emerge in the following way. Knowledge is the ability to answer questions. If some system answers a question, this serves as an indicator that it possesses “knowledge.” This is the so-called empirical definition of knowledge. The way to identify the knowledge necessary for someone to solve a given class of problems is to observe one-self in the process of self-instruction. Human thinking is the process of realizing an algorithm. The source of information about human thinking with which the work of the computer is compared is often simply the definitions taken from a dictionary (a dictionary of definitions). All four positions have a direct relation to sciences that study human thought, to psychology in particular. Therefore, it is advisable to examine to what degree the concrete scientific results of psychological research are taken into account by investigators in artificial intelligence. A distinction between formal and descriptive knowledge has existed in psychology for a long time. Let us assume that a student knows three questions that will be on examination and learns the answers mechanically, “by heart.” A good teacher would hardly be satisfied with the success of the pupil. Psychologically, knowledge is the reflection of some essential relations among surrounding objects. It is a system of generalizations. When a person learns “mechanically,” he/she ascertains only the connection between the question and the answer (it is another matter that, in its pure form, this phenomenon is seldom seen). When some information is meaningfully acquired, it always is included in some system of a person’s past experience. Thus, in “artificial intelligence” knowledge is treated formally and has only an external similarity to genuine human knowledge. The method that the advocates of this scientific trend propose for revealing the nature of the human knowledge used in problem solving (through observations of the processes of self-instruction) seems to be extremely limited. The fact is that in human actions some components are consciously realized and some are not, including generalizations. It has long been known that studies of how human beings perform actions with objects reveal that they use practical generalizations that they do not fully realize at the conscious level but that play an active role in the process of solving certain problems. If we fail to take these generalizations into account, the answer to the question of what kind of knowledge a person uses in solving a given class of problems would be incomplete in an important way. As pointed out above, the psychological investigation of thought reveals a somewhat more complex organization of search processes in human problem-solving that the analogous processes in computer. These investigations, in particular, show that the actual process of search in human problem-solving is not based on an algorithm. The algorithm, as best, emerges as thee product of human creative activation. In this area the appeal is only to dictionaries, in which case comparison of human thought with the work of the computer appears outmoded. It is interesting that critical analyses of attempts to identify the features of thought that are unique to humans (as opposed to computers) have unearthed only one factor. Slagle has “refuted” only one thesis: “The computer can do only what it is told to do, and people can do more.” In his opinion, in some sense humans also can do only what they are “told.” For example, heredity supposedly tells humans what to do and 7 determines how they derive experience from the environment. On the basis of this reasoning, the author arrives at the conclusion that those arguments are wrong that favor the notion that the machine in principle cannot be as intelligent as humans. It is well known, however, that even at the level of animals’ so-called instinctive behavior, hereditary factors do not predetermine their orientation toward the surrounding environment in a simple way. Consequently, Slagle’s argument is not strictly correct. The defense of machine thought we have discussed contains an assumption that can be formulated as follows: If a technical system solves the same problems a human does by using thought, and if the system uses it for problem solving, this means that the technical system also possesses thought. There is good reason, however, to consider this initial (intuitive) assumption simply incorrect. The fact is that the scientific concept of thought is concerned not only with the nature of the results but also with the procedural side of human cognitive activity. Therefore, one can say that a technical system possesses thought only when it not only solves the same problems that humans solve but solves them in the same way. To label as thought the work of the machine that only solves the same problems as humans do has no more foundation than to label an airplane, in a scientific (for instance, biological), not everyday, context, a bird. Our analysis shows that the initial assumptions of the scientific trend that has come to be known as artificial intelligence consist of a fundamental simplification of ideas about thinking and knowledge and about the correlation of thought and knowledge in humans. This situation gives reason to doubt whether the advocates of the present trend (i.e., those who want to create an automaton that reproduces all human thinking abilities) have proved their point. One can also conclude that there are no foundations for thinking that a dialectical materialistic world-view, being the result of the development of a scientific system, must yield to a mechanical picture of the world. Given this, what can we say about “artificial intelligence”? We must regard “artificial intelligence” as a theory of programming problems for computers. The very label is no more than a metaphor reflecting the romanticism about which, for instance, Minsky and Papert write in their book Perceptrons (1969, p. 10). Psychology is now developing in its own ways. We shall examine how the problem with which we are concerned can be posed if defined the framework of Soviet psychology. Traditional Soviet psychology has evolved on the basis of a historical approach to the development of human mental processes. Vygotsky played a considerable role in establishing this principle. In analyzing practical activity, psychologists emphasize the tool as the most important component of human activity. This component creates the qualitative uniqueness of human activity in comparison with animal behavior. The tool is not simply added on to human activity: rather, it transforms it. For example, the simplest action with a tool – chopping wood – produces a result that could not have been achieved without the use of an axe. Yet the axe itself did not produce this result. Action with a tool implies a combination of activation and human creative adaptation. Tools themselves appear as supplementary organs created by humans. The mediated nature of human activity clearly plays a leading role in the analysis of practical activity. One of Vygotsky’s central theses is that mental processes change in human beings as their processes of practical activity change (i.e., the mental processes become mediated). The sign (language, mathematical sign, mnemotechnic means, etc.) emerges as such a mediational link. Language is the most important form of the sign. In using auxiliary means and signs (for example, in making a notch in stick in order to remember), humans produce changes in external things; but these changes subsequently have an affect on their internal mental processes. By changing the milieu, humans can 8 master their own behavior and direct their mental processes. Mediated mental processes initially emerge as functions distributed new forms of interfunctional relations. The emergence of: (1) logical thought as opposed to situational thought that is unmediated, (2) mediated memory as opposed unmediated memory, and (3) voluntary attention as opposed to involuntary attention are all examples of the development of higher mental functions. Writing is mankind’s artificial memory, which gives human memory an immense advantage over animal memory. With the help of speech, humans master thought, since the logical processing of their perception is conduced through a verbal formulation. We approach the analysis of the role of computers by relying on these traditions of the historical approach to human activity. For us, the computer and other machines are organs of the human brain created by the human hand. If, at the stage of the creation of engines, automobiles served as tools in human activity for carrying out work that required great expenditures of energy, at the stage of the development of computers, the latter became the tools of human intellectual activity. Mental activity has its own mediated structure, but the means are novel. As a result, the question of the influence or computers on the development of human mental processes must be reformulated as follows: How does mediation of mental processes by the computer differ from mediation by signs? Do the new means introduce new changes into the very structure of intellectual processes? In other words, can we distinguish a new stage in the development of human higher mental processes? As a result of using computers, a transformation of human activity occurs, and new forms of activity emerge. These changes are one expression of the scientifictechnological revolution. The distribution of bibliographic information and computation in a bank, the planning of new machines and the adoption of complex decision in a system of management, medical diagnosis and the control of airplane movement, scientific research, instruction, and the creations of art are all constructed in new ways. The problem of the creation of “human-computer” system is very real. The creation of such systems involves many scientific problems: technical, logicomathematical, sociological, and psychological. I shall examine the psychological ones in more detail. What is the specific nature of human activity in “human-computer” systems as opposed to other forms of activity? In accordance with the definition generally used, “human-computer” systems cannot function without at least one human and one program stored in the computer’s memory. Consequently, we are talking about forms of human activity that cannot take place without the participation of the computer. *”Human-computer” systems generally are not concerned with using computers for control processes in which a human is included only for reasons of safety or when a human becomes involved only when deficiencies in system functioning arise.) One of the characteristic features of human activity in the systems we are examining is the immediate receipt of information about one or another result of actions. In their time, Anokhin and Bernshtein described the principle of reverse afferentation or the mechanism of sensory correction as a necessary link in the regulation of activation. Later, the mechanism of feedback as a universal principle was formulated in Weiner’s works. The reorganization of this regulation in human intellectual activity is one of the ways in which computers have come to be studied. This reorganization of mechanisms of feedback makes the process more controlled. We shall compare the process of regulating human activity through normal verbal commands with the process when aided by computer. The similarity here is obvious, but there is an important difference: possibilities for immediate feedback are much greater in the second case. In addition, the computer can appraise and provide information 9 about intermediate results of human activity that are not perceived by an outside observer (for example, changes in states revealed by an electroencephalogram). Thus, with regard to the problem of regulation we can say that not only in the computer a new means of mediation of human activity but the very reorganization of this activity means described by Vygotsky are used. In concrete terms, what is involved in the change in human intellectual activity when a person is working in a “human-computer” system? Usually this change is described in terms of liberation from technical, performance components. We shall examine in more detail what the performance component of mental activity means from the psychological point of view. Let us take the situation of work in a “human-computer” system. The computer has an algorithm for solving chess problems consisting of two moves. The need to discover the solution for a specific problem arises in the person. The solution consists of enumerating or listing the succession of moves that would inevitably lead to a checkmate (i.e., it consists of a plan of actions). It is sufficient to give the computer the goal and a description of the conditions of the concrete problem in order to obtain the plan of actions necessary to attain the goal. Can one say that the computer has freed the person from an algorithm? No, one cannot say this, since this algorithm was not known to the person. It means that from the psychological point of view, this case involves freeing the person from the necessity of going through the actual search for the solution of the problem. As psychological studies have shown, even the solution of a chess problem often emerges as a genuinely creative activity that includes the complex mechanisms of the regulation of a search, which we described above. Consequently, in this example the human is freed not from “mechanistic work,” but from “creative work.” On a broader plane we can say that regard to problems for which the algorithm either is not known to humans or is so complex that impossible or irrational to use it, the algorithm can free humans from creative forms of search. At first glance our conclusions support the theory of substitution, in which the work of the computer replaces creative activity. Such substitution can actually take place, but it is relevant for only a certain class of problems, namely, those for which an algorithm can be, and actually has been, worked out. With regard to this class of problems, the psychological structure of the activity of the user and that of the programmer will be essentially different. The user who obtains an answer to a problem from the computer may not know completely the algorithm for its solution. This algorithm, which was developed by another person, is used after its creation without being mastered (i.e., it remains in computer memory). An algorithm is a fully formalized procedure for the solution of a given class of problem worked out by the programmer. Consequently, the user’s activity is mediated by this formalized procedure. It has a character external to the mediated activity. At first glance this activity is quite analogous to any assignment given to another person. For instance, instead of solving a chess problem myself, I request a colleague to do it. The difference is revealed when we say that colleague does not solve this problem in accordance with an algorithm. It means that we simply have a transmission of the solution from one person to another, not a transformation into a formalized process. We can justifiably state that users’ activity that is mediated by a procedure they have not mastered is a new form of human activity. Now let us examine the psychological features of the activity of programmers who earlier worked out an algorithm for the solution of chess problems and introduced it into the memory of the computer. They were confronted with a new problem 10 consisting of two moves. The solution of the problem can be achieved by two means: either without turning to the computer, in which case it will not have the nature of a fully formalized process, or by turning to the computer, in which case it will use a fully formalized procedure worked out earlier. Can one say that the user was freed a fully formalized process (in the second case?) No, because the problem is not solved even by programmers in a purely formal way if they do it without the help of the computer. Programmers do not themselves from a formalized process, but specially carry out formalization in order to relieve other people and themselves of the necessity of solving a certain class of problem. This means a machine can perform “mechanical” work not because human mechanical work is transferred to it, but because work formerly done in a nonmechanical way is transformed into mechanical work. Viewing computerization as the liberation of human activity from its technical aspects is correct only in some cases – namely, when an activity is formalized in the course of the development of human activity itself and in the division of labor, i.e., the activity begins to be composed of monotonous, repeated actions that are executed in accordance with rigidly fixed rules. Of course, it is never completely “mechanical” in the precise meaning of the word; but the meaningfulness of the activity can stop being essential from the point of view of achieving the unmediated results of the actions accomplished. In this case computerization can mean transmitting to the machine those elements that begin to be formalized in human activity itself. The cases of greatest interest are not those in which the computer takes over the solution of some problems solved earlier by humans (no matter how), but those in which a problem is solved jointly by humans and computer (i.e., “human-computer” systems proper). From our point of view, this type of system (not those of “artificial intelligence”) is of primary interest for the future of computerization. We know that creative thought is impossible without the use of previously prepared knowledge, which is often stored in the “artificial human memory” (reference books, encyclopedias, magazines, books, etc.). At the same time, the accumulation of information means that the search for information in external memory often turns into an independent task, which is sometimes so complex that it distracts the problem solver from the solution of the fundamental problem. Often such search activity turns out to be impossible, which is why we hear that it is sometimes easier to make a discovery anew by oneself than to check whether or not others have already made it. The use of computers for storing information is a new stage in the development of what Vygotsky called the “artificial memory of the human race.” Effective use of computers for the search for information in this memory reorganizes human activity in the sense that it makes it possible to focus on solving truly creative problems. Thus, we are confronted not with the disappearance of thought, but with the reorganization of human activity and the appearance of new forms of mediation in which the computer as a tool of mental activity transforms this very activity. I suggest that the theory of reorganization reflects the real facts of historical development rather than the theories of substitution and supplementation. The influence of computers on mental activity must be examined not only in terms of the historical development of human activity but also in ontogenetic and functional terms. Elaboration of the theory of ontogenetic development led to the formulation of the position that acquisition of the experience of society is the most characteristic trait of the processes of human ontogenetic development. With the appearance of the computer, the very form of storing the experience of society (the “electronic brain” vs. the library) is changed, as is the process of acquisition of knowledge when teacher – 11 student relations begin to be mediated by computer. Moreover, the process of acquiring knowledge is changed (i.e., it is now possible to reduce the number of formal procedures to be acquired thanks to the use of computers). This gives us a basis for stating that as a result of computerization, a new stage in the ontogenetic development of thinking has also developed. Psychological research has shown that the solution of a complex cognitive problem by humans is a process of functional development, or a process of the succession of different stages and mechanisms that realize this activity – for example, intuitive guesses and strictly logical verification of these guesses, the feeling of being close to a solution and the fully developed logical analysis of the solution. The computer drastically changes (or can change) this process of functional development. By translating the formal components of activity in the solution of a problem into the form of an external mediating chain, the computer makes it possible to reveal and develop the intuitive component of thought and the chain of the generation of hypotheses, since the complexity of the task of verifying these hypotheses often, as it were, overwhelms the intuitive components of thought. The results of the analysis that has been conducted allow us to state that even on a functional plane, changes occur in the intellectual processes of a person solving complex problems in conjunction with computer. In elaborating the theory of higher mental functions, Vygotsky drew parallel between the historical and the ontogenetic development of activity. The child acquires signs that were developed earlier in human history. In different instances, external mediation was interpreted by Vygotsky as a stage in the path to internal mediated activity. For example, speech, which evolves into inner speech, is the foundation of discursive thought, voluntary memory, attention, etc. A new stage of mediation (by computers) is not a stage in the path to internal mediation: it is the further development of external mediation or interpsychological functioning (according to Vygotsky), which exerts an influence on the development of intrapsychological functioning. Here we see yet another feature of the new forms of mediated human mental activity. In this paper I have given special attention to showing how the computer changes the structure of human intellectual activity. Memory, the storage of information, and its search (or reproduction) are reorganized. Communication is changed, since human communication with the computer, especially in the period when languages that are similar to natural language are being created, is a new form of communication. Human relations are mediated through use of computers. Of course, the computer creates only the possibility for human activity to acquire a more complete structure. Such possibilities are realized only when certain technical, psychological, and social conditions are found. The technical conditions is that the computer must be adequate; the psychological condition is that the computer must be adapted to human activity, and the human must adapt to the conditions of working with a computer. The main conditions are social – what the goals are for which the computer is used in a given social system. How society formulates the problem of advancing the creative content of its citizens’ labor is a necessary condition for the full use of the computer’s possibilities. From our point of view, analysis of the reorganization of human intellectual, mnemonic, communicative, and creative activity as a result of the use of computers and optimization of these reorganizations, including both expansion of the computer’s functional possibilities and coordination of human activity with the use of tools, must constitute the content of a new branch of psychology, which I propose to label the psychology of computerization. The development of this new branch of psychology will permit more complete use of the possibilities of the development of human activities, 12 which are revealed thanks to computers, and will permit us to avoid secondary negative consequences of technical progress in this area. Reference Minsky, M. & Papert, S. Perceptrons. Cambridge, Mass.: M.I.T. Press, 1969. This Paper was published in Wertsch, J. V. (Ed.). The Concept of Activity in Soviet Psychology. New York: M.E. Sharpe Inc. pp. 256 – 278, 1981 13 THE CONCEPT OF ACTIVITY IN SOVIET PSYCHOLOGY Translated and edited by James V. Wertsch Preface It is now just twenty years since I began studying Russian while a graduate student in psychology. It was considered a strange thing to do. What did I think I would learn? My student colleagues and my professors wondered if the rewards would be worth the effort. And the effort turned out to be considerable. Just think, for example, about getting into a theoretical discussion with Alexei Leont’ev. (In fact, you could not avoid such a discussion; Leont’ev was the theorist of Soviet Psychology, famous for his complex formulations). He might try to tell you that “...international activities emerge out of practical activity developed in human society based on labor, [and] international mental activities are formed in the course of each person’s ontogenesis in each new generation” (A. N. Leont’ev, this volume). Now, how does an American psychologist trained to study learning processes in rats and humans as part of a general theory of stimulus response learning interpret something like that? Not very fully, if my experience at the beginning of my career was any indication. As I have discussed elsewhere (Cole, 1979), my continued exposure to Soviet psychology gradually began to have an impact on my thinking, especially when I found that I had to act as an interpreter in addition to being a translator (a distinction that is often difficult to maintain). Citations such as the one from Leont’ev ceased to be “Greek to me” and started to make sense. Not all of it and not always, but some of it sometimes. There was certainly something there worth delving into. One of the most important ideas I had to understand in order for Soviet psychology to make much sense for was the concept of deyatel’nost, “activity,” as it called in this book. What does it mean, and why write a book about it? A large Soviet dictionary isn’t very helpful. Activity is called “work, a task of some kind,” or “the work of some kind of organ.” Examples include references to “pedagogical activity,” “business activity,” or “the heart’s activity” – not exactly the most obvious units to use for a psychology of learning. Ironically, our own Webster’s dictionary offers some additional ideas that begin to hint at what the Soviets are talking about. It refers to “any process actually or potentially involving mental function” and – even more relevant – “an organizational unit for performing a specific function.” When we put these two aspects of Webster’s idea of “activity” together, we get “an organizational unit for performing a mental function.” That definitely puts us on the right track. Activity is a unit of analysis that includes both the individual and his/her culturally defined environment. Now if we go back and reread the quotation from Leont’ev with which I began this discussion, we can begin to appreciate Leont’ev’s meaning. He begins by saying that internal organizational units for performing a mental function arise out of practical, external, organizational units for performing that function. If we keep on with interpretative process, we arrive at a theory so ambitious in Scope that it exceeds anything in American academic psychology. Vygotsky often wrote about the “crisis” in psychology, a term then very much in fashion among Continental Psychologists in the early decades of this century. They were arguing about the successor to the crumbled empire of Wundtian 14 psychology, established at the beginning of the discipline in 1879. The concept of activity is central to the Soviet strategy for overcoming that crisis. As Luria (1979) explains very clearly, Vygotsky was arguing against the views of his contemporaries (except insofar as they represented, in his opinion, mistaken strategies for overcoming a crisis that had its roots in Wundt’s methods, which in turn embodied Wundt’s kind of theory). Vygotsky was returning to the argument between Dilthey and Wundt about the very possibility of an experimental psychology. He argued (correctly, I believe) that when Wundt gave up on the laboratory study of higher psychological functions, he created a methodological dualism that would limit psychology so long as it was accepted. An “explanatory psychology” would live alongside a “descriptive psychology,” and the two would happily deal with their own phenomena, fight for hegemony, and never arrive at a principled basis for whole psychology built upon a single set of principles with a single, overarching theory. Surveying the current scene in psychology, who could say that Vygotsky was wrong? It is one thing to criticize and to moan about crises; it is another to resolve them. A linchpin in the Soviet resolution of that crisis is provided by the concept of deyatel’nost discussed in these pages. As Leont’ev says, it breaks down the distinction between the external world and the world of internal phenomena. But it produces new problems of its own. From an “activity perspective” the psychological experiment can no longer be set up entirely to model philosophical speculation, as in Wundt’s day: it must model the phenomena of everyday, “practical activity.” Among other distinctions that such a perspective erases is that between basic and applied research, for applied settings are the locus practical activity. And, of course, by definition the individual and those around him/her engaged in the same activity are the basic unit of analysis of psychology. Social psychology and cognitive psychology are one area of inquiry. Manmachine systems are simply an extreme version of human – environment interactions in which instead of people to interact with, one has the products of human activity. It is heady stuff, the Soviet version of “general systems theory.” Like any approach that attempts to provide an alternative world view, it is difficult to grasp in its entirety and difficult to interpret in its details. American psychology, driven by the limitations of the computer as an analogy for human thinking, is beginning to deal. 15