Den kvalificerande erfarenheten Lärande vid störningar i automatiserad produktion

Abstract

Technical development and new organizational solutions have given rise to changed demands on operators as handlers of disturbances in modern industrial production. This thesis is concerned with learning in relation to the management of disturbances at automated machines. Its main purpose is to contribute to theoretical development in the arena of individual learning by, on the basis of empirically founded analyses, understanding, conceptualizing and illustrating the learning of the individual in relation to a specific work task. The task is one that is both complex and complicated, and is strongly linked to modern production technology. It arises suddenly, or at least apparently so, and imposes immediate demands for action. It is not determined in advance : with regard to time, place or accomplishment. What is deviant or normal in the task is dependent on whose perspective it is viewed from. This thesis can be regarded as both a journey of discovery and a generator of theory: the study of learning in relation to a specific work task is designed to provide a basis for understanding and discussing learning-related inferences at a more general level. A subordinate purpose is to describe how operators in modern manufacturing relate to and handle disturbances to production. Accidents as a point of departure Over a decade of conducting accident research (Backström and Döös, 1996; Döös and Backström, 1993), I have learned that undisturbed production is the exception rather than the rule. Dealing with deviations from how production is envisaged to function is a quite normal work task, and thereby forms part of the everyday working life of the machine operator. However, in occupational-accident research, attention is seldom explicitly directed at the normal situation. For natural reasons, albeit not out of necessity, the concentration is on deviations. The concepts of disturbance and deviation are closely related. They occupy a central position in the field of accident and safety research in that, in many contexts, disturbances/deviations have been demonstrated to be significant to accident occurrence (Faverge, 1967; Kjellén, 1983). Disturbance to production is an early occurrence in the chain of events leading to an automation accident (Backström and Döös, 1997, in press). In focusing on disturbances as normal and everyday phenomena, the intention here is also to make a contribution to accident research. My study of learning in relation to production disturbances represents a different kind of initial analysis than that which has been customarily performed in the field of occupational-accident research. From previous research, there has emerged an image of operators as actors dealing with symptoms and not causes; the operator has appeared as a "fixer" of immediate problems, rather than a devisor of long-term solutions to production problems. Such an image prompts questions concerning how operators handle disturbances to production, and how learning in relation to this task can be understood. This gave rise to ideas concerned with a learning-related qualitative dimension to the handling of production disturbances : a dimension where unskilled casual fixing "on the spur of the moment" contrasts with the qualified devising of durable solutions. In this context, I speculated on the problematic tension that I had perceived between accident research on the one hand and qualification research on the other. We, in the accident-research field, had tended to regard disturbance as a negative risk that should be avoided to the greatest extent possible. Indeed, some accident researchers have seen both variation and autonomy at work as safety risks (Saari and Lahtela, 197). By contrast, qualification researchers have regarded disturbances as providing opportunities for interesting and personally developing problem-solution, i.e. as offering a task that enhances operators' knowhow and enriches their job content (Nemitz, 1983). Desires for extended occupational tasks and richer job content appeared to come into conflict with requirements of safety. Development work on the risk-analysis method known as Riv (performed in this specific context by myself and a colleague in Sweden) showed how operators were involved in the identification of hazards and in finding permanent solutions to disturbance problems. Enriching work tasks and work for the promotion of safety seemed to go hand in hand. Production disturbances could be handled in a manner that promoted development and learning for operators, while, at the same time, safety remained in focus (Backström and Döös, Accepted, Döös and Backström, 1993). A recent paper presents empirical support for there being a positive relationship between safety initiatives and complexity/autonomy in work tasks (Simard and Marchand, Accepted). Theoretical frame of reference Through this study, I have attempted to investigate and demonstrate how everyday learning is built up in connection with a specific work task : seen from the horizon of operators, in their eyes, and in terms of how they think. The kind of learning under study is experiential by nature : "in which the learner is directly in touch with the realities being studied" (Keeton and Tate, 1978). The study is founded on the perspective of contextual didactics, within which constructivism, action theory and the concept of affordance constitute the essential points of departure (Löfberg, 1994; Piaget, 1947 and 1981, 1974 and 1980; Reed, 1993). This is a perspective in accordance with which the individual constructs knowledge on the basis of his interpretation of the conditions afforded by the surrounding environment. In other words, the environment, as also the task and the situation, cannot be regarded as purely and objectively true, predetermined and fixed, but is perceived and given meaning by each individual. Action, both as doing and as reflection, emerges as a precondition for task-related learning to take place. It links the individual and the situation in a specific context. From the theory originally emanating from Piaget (1947 and 1981, 1974 and 1980) and further developed by neo-Piagetians such as Basseches (1984,1986) and Kegan (1982,1994), I have utilized the fundamental ideas of the individual as an active constructor of knowledge and expertise, and of there being two stages to the adaptation process (assimilation and accommodation); further, I have made use of neo-Piagetian ideas concerning cognitive structures and their development during adulthood (especially in relation to work), the subject-object balance, of embeddedness in the self-evident, and of intrinsic stage imbalance. For me, conceptualization of these stages has offered an opportunity to view the learning process. Kolb (1984) also builds further on Piaget, as too on Dewey and Lewin. Like Kolb, I see knowledge as a result of the combination of apprehending experience and of transforming it. In his or her learning process, a person moves, in varying degrees, from actor to observer, and from specific involvement to general analytical seclusion. Concrete experience, reflective observation, abstract conceptualization, and active experimentation constitute, on Kolb's view of the learning process, a cyclical movement between doing and thinking. The four adaptive learning modes make up two dimensions: prehension and transformation. The prehension dimension is represented by the dialectic between the concrete and the abstract. The transformation dimension represents two ways in which experience is re-molded: by inner reflection (intention), and by actively dealing with the external world (extension). Intention and extension can just as well be applied to concrete apprehension of the world via the senses as to symbolic comprehension. In Kolb's view, the two dimensions make an equal contribution to the learning process. From action theory I have taken, above all, the model of hierarchically ordered levels of regulations of action (Frese and Zapf, 1994; Rasmussen, 1982). These are important for understanding the preconditions for and results of learning. In some parts of action theory, a distinction is made between the result of an action and its consequences, and I have found this fruitful. Given the increasing cognitive demands that are imposed on operators in modern industrial production, it becomes necessary to develop a "theoretical" conceptualizing approach : one that is abstracted from the immediate, concrete apprehension-experience, i.e. that which is perceived directly by the senses and has been traditionally linked to the practice of an occupation. To an increasing extent, it is essential for operators (and others) to construct their own generalizations and concepts in order to be able to manage occupational practice, and continually to learn and develop. Tools, in the form of concepts, are needed, both to conceptualize and to enable the posing of relevant task-related questions. Method The study was conducted at a large company within the Swedish engineering industry. It covered two production lines for the manufacture of vehicle-transmission shafts and the 14 operators who worked there. Each line contained ten to twelve computer-controlled, automated items of machining equipment. Data were gathered in case-study form (Westlander, 1991; Yin, 1989). The methodological approach is qualitative, and data have been acquired in stages by means of abductive inference (Fann, 1970). In the selection of methods, the idea of triangulation has been employed. Data have been gathered using interviews and observations, supplemented by questionnaires, disturbance-reporting and shop-floor conversations. Some statistical information has also been utilized. I do not regard data analysis as a separate, narrowly demarcated task to be performed after field work and data collection are completed. Rather, on the approach to the research process I chose to adopt, analysis was embarked upon even in the course of data collection, continued during data processing, and was completed during final writing. The final stage constituted a more concentrated and pure form of analyzing. The generalizing that I wished to undertake, in accordance with the main purpose of the thesis, is not statistical : to make an inference from a sample to a population : but analytical (Firestone, 1993, Yin, 1989). In analytical generalization there is a striving to test whether results can provide a supplementary input of empirical data into a certain theoretical domain, and thereby act as an aid to theory extension. The kind of theory that I am referring to, as mentioned above, is one that is based on an approach that treats learning from a perspective that is constructivist, action-oriented and contextually didactic, and characterized by the view that the encounter with situation and task in a specific context is of significance. With regard to the thesis's subordinate, more descriptive aim, generalizing is more to be regarded as what has been called "case-to-case transfer" (Firestone, 1993), where thorough description of context enables the reader to determine whether results obtained in any one context may be utilizable in others. Results and conceptual discussions with an empirical foundation The results of this thesis are reported in part in a purely empirical section under three main headings: "Operators define and handle production disturbances"; "Proceeding in small incremental steps" ; and "Knowhow grows in its context". In part also, they are presented by means of a conceptual discussion with an empirical foundation under the headings: "Conceptual network and dimensional stance"; "Making the commonplace available : learning as a process of contextualization". The reasoning presented under these latter two headings should be regarded as contributions made "along the way", arrived at on empirical grounds and proposed as possible ways of understanding and conceptualizing learning in relation to a specific and complex work task. Operators define and handle production disturbances. How far a person, as an operator, develops the task of disturbance handling seems to be related to his stance towards the normal and the deviant. When an operator envisages his basic task as that of managing ongoing production, of maintaining normality, the work task comes across as essentially different and considerably more limited than when it is conceived in terms of dealing with and remedying the faults and problems that arise. In the latter case, the deviant can be said to have been integrated into the normal, which can be supposed to be of particular relevance in the kind of production in which machine operators are involved. Production halts, problems and disturbances are commonplace. How the task is envisaged by operators is dependent upon how the situation appears to them. Operators state that they have to weigh matters up and make their own judgments. Decisions on how to handle a fault are by no means self-evident. Ad-hoc fixing "for the time being" is something performed by all operators. Situations that prompt this type of action include running behind schedule, needing to get production going quickly, and the appearance of unclear faults, those that arise irregularly, do not cause much bother, or disappear anyway on re-rigging. Situations that prompt permanent solutions are related to recurring troublesome faults : those that are intolerable, or those giving rise to a risk of machine malfunction or poor machining quality. It is part of a skilled operator's occupational expertise to know in which situations it is alright to effect a temporary solution and keep production running. To treat every fault as the same, in the sense that production must be halted and the fault remedied, is not regarded as reasonable among machine operators. A perspective that operators have in common is expressed in the following chain of actions and decisions: fix it oneself; get help from/consult workmate(s) on the line; (consult team leader); finally, summon maintenance personnel. The contents of each link in the chain, however, are not commonly shared; rather, they vary considerably. That there were major differences between what was referred to in the chain emerged in the course of data collection and interactive analysis. Thus, it is in this chain and in the expression "It all depends" that dissimilarities start to emerge : dissimilarities that are clearly linked to both the person and the situation. It is in this context that the operator's way of relating to normality and deviation both obtain and reveal their significance. Proceeding in small incremental steps & Knowhow grows in its context. Learning in an occupational task takes place in small steps where the everyday experience is needed as a basis for being attentive to, understanding, and remedying disturbances. What becomes ingrained has an important function. Quality in particular experiences is of importance for learning in the task. Qualifying disturbance handling is characterized by ramification, cyclical completeness, and having questions of one's own at hand. Among that which is essential to learning is to seek faults with the normal as a point of departure and to "converse" with the machine, thereby being "in dialogue" with one's own expertise. Such conversations can continue for shorter or longer periods, be brought to a conclusion or interrupted, never finished, or taken up later Disturbances to production offer occasions for communication, collaboration and collective learning : sometimes in the acute situation itself, sometimes by virtue of being subjects of later conversation. Shift work on automated lines creates partially contradictory preconditions for joint work; on the one hand, linkage is created and there are common tasks; on the other, there are interruptions to experiences and the taking of responsibility. The learning that takes place in the occupational task is linked to both utility and use, which at a minimum level lies in keeping the production line in operation. Operators require knowhow for the line to function, and they need experience to be able to control its operation. The value of knowhow lies in use. Use is a superordinate driving force in learning. For several reasons, it proves that regular experiences are of key importance. Everyday knowledge must constantly be maintained. Learning in the work task naturally has consequences in terms of how the task is performed. Knowhow can be regarded as the product of what one has learnt. Through study of the immediate performance of a work task, a link can therefore be made to the learning process. It becomes possible to understand how operators generalize experiences into something that goes beyond simply running or handling the machine, i.e. to move from operation to what might be called "direction and control". There are two key types of generalization: first, generalized understanding, making it possible to apply knowledge in new contexts; second, a generalized method can be applied. Individually designed methods and work principles can, for example, be employed to break down the problem into its component parts, to proceed on the basis of what exists, not to be locked-in by the rule, to isolate a fault, or to go back to an earlier point in the process so as to be able to move forwards again. Attentiveness is central in several respects to both qualified and unqualified disturbance handling. This is a matter of various aspects of discerning details in everyday activities. It can apply to immediate concentration on what one is working with, to knowing what one should be looking for, or to an openness or alertness to indications that something is not quite as it should be. That sense-based experience is also used in the task emerges in relation to attentiveness. Questioning and attentiveness are closely related. In my view, they also presuppose each other. The questions of the individual are decisive to learning. The pedagogically relevant question has been traced to dissonance. It is there where it seems to have its roots. By this I mean that it finds expression in puzzlement and consequent wondering arising in response to the operator perceiving something to be deviant. What has happened is not truly compatible with what has been previously assimilated. And if experiences, e.g. in the form of dialogue with the machine, continue to be out of line for a while, this provides an opportunity for, or invitation to, learning on the part of the operator. Genuine wondering on the part of the person prompts a search for answers among his own experiences, a search for links to contexts that have explanatory value. It is here, in my view, that the ramifications of previous experiences are put to use. When methods and work principles are generalized, the genuine question is at the fore. Thought networks and dimensional stances & Making the commonplace available. Learning can be understood as a contextualization process in which everyday experience is transformed by means of situations appearing as normal, typical or exceptional. The building of knowhow is based on discerning and utilizing everyday details. The commonplace becomes available when details and events appear as normal, typical or exceptional. As conceivable structural concepts to operationalize the individual's encounter with practice and learning in a specific, situation-related and complex work task, thought networks and dimensional stances are proposed. The concepts have their foundation in empirical observations, but also have points of contact with other concepts in their field (such as cognitive structures, schemata, patterns, and mental models). I regard thought networks as a dynamic conceptual concept. These are mobile structures, but not within a learning process from A to B; there are many movements that go in different directions, which jointly build up an integrated contemporaneous context, and also knowhow within this context. Thought networks are situationally related. I regard them as controllers of what one experiences; they characterize both the situation and specific paths of action, thereby linking factors in the situation with personal action alternatives. To learn a new task involves, among other things, building up cognitive structures, thought networks, in small steps. In the case of operators, part of such a network consists of production disturbances. Such a network is a kind of living reconstruction which the operator builds upon and then re-shapes. How, in which direction, and with what qualities are dependent on the experiences of the operator and aspects of these. It is via conceptual networks that situationally related principles enabling the taking of an overview are built up. The person involved in the process of learning and developing adopts a succession of thought networks. They are open to change by virtue of the particular questions posed and through the actions that the work task demands. The question makes a contribution to learning every time an already devised construction is not simply applied or re-used. In this situation, previous experiences of similar situations are re-constructed in thoughts through the using of experiences. When, by contrast, learning comes into abeyance, already existent structures are employed in a mechanical manner. Through frequent use and as a result of the operator employing them without reconstruction, they become a kind of petrifying network. In accordance with this, data have indicated the appearance of two kinds of thought networks: mobile, living reconstructions; and fixed constructions, petrified networks, punched out historically and applied to a situation without the presence of thought. The latter appear as fixed or static. Stances are built up from a number of dimensions, and have consequences for the tasks which the operator regards as his own : how he defines and looks upon his work. Crudely speaking, the extremes can be described as follows: one dimensional stance has non-reflective routine action as its consequence; the other promotes reflection and problem-solving. How then can a stance that results in non-reflective routine action be characterized? There is a tendency to dichotomized thinking where opportunities and alternatives are not realized, and thought proceeds in "either-or" terms. The thinking process rapidly comes to an end, and no genuine questions are posed by the individual. Rather than obtaining explanations where various explanations are linked together to form insights, the operator is left with rigid assertions. The explanations he has available "come up short", so to speak, or at least a couple of steps before comprehension is possible. Petrified thought networks are applied, rather than something new being constructed from the conditions prevailing in the situation in question. To be able to reflect, there is a need for one's thoughts to be rooted in something extremely concrete and specific, i.e. in a certain situation or a certain machine. Generalizations can be hard to make. Even the opposite applies in that a petrified statement is hard to relate back to previous examples and stories. The genesis of the assertion has been lost. In any case, few stories of disturbances are told. The future is not included in the work task, and one is governed by external demands rather than personal goals and requirements. Not much preventive work is done; rather, what becomes acute is taken care of. The task is to keep production running, not to devote oneself to improvement or prevention. Thinking is confined to one's own group of machines, and there is an anxiety to achieve an acceptable production outcome. When the machines run as they should, there is time to pass. Then, a fault can come across as a welcome break in the routine. Work and leisure time are quite separate; thinking about job problems when free is inconceivable, and never happens. A reflective problem-solving stance results in a quite different kind of action. This incorporates both qualified and qualifying dimensions, in that process thinking takes place, and alternative opportunities and solutions, are recognized. The operator ponders, wonders and poses questions over why things are as they are and how they might be. He knows the background to and reasons why he takes the actions he does in relation to the work task, and has related explanations and insights. External changes are not so essential; changes are effected by virtue of his own mode of working. Personal demands and goals govern action, and systematic strategies and individual approaches are developed and employed. There is free movement between concrete examples from which conceptions take root and general, overall thinking not related to particular examples. Quality ambitions exceed formal tolerance requirements, and both thought and action can apply to proximal and distal conditions in the workshop. During the times that the machines run as they should, time is devoted to other tasks : those which, so to say, "lie in waiting" but still need to be done. Faults are not seen as disruptions to routines, but can be regarded as interesting problems. Tasks designed to promote improvements are a part of the work; the operator experiments, and develops systems for control and quality assurance of his own work performance. There are many stories of disturbances, and the boundary between work and free-time is less strictly demarcated. Pondering over job-related problems during leisure time is not an impossibility; nor is it too much of a burden to come to work a few hours early if needed, or to leave late if a problem has to be resolved. Thought networks and dimensional stances are not regarded as permanent and unchangeable : rather as stable, albeit to varying degrees and extents. They make sense through their relations to each other. While the thought network expresses the person's relation to a situation, his or her stance is more linked to the performance and definition of the work task on a general level. The implication of this is that the dimensional stance will determine which thought network is employed, and thereby how the task will be performed. In this way, over time, the thought network shapes the dimensional stance. Concluding remarks For a study of learning to highlight the details of the commonplace and be based on everyday small-scale experiences is relatively rare in the literature. My study shows how knowhow is built up, and demonstrates the complexity of learning when studied in a context, i.e. in its real surroundings. The significance of small-scale learning, how the quality in previous experiences is passed on to others, and how operators can be seen to generalize their understanding within the work task can be regarded as important contributions of the study. So too are the discussion of living as opposed to petrified thought networks and their relation to dimensional stances, and also how context can be made to emerge from everyday reality and used in the process of learning. By means of puzzlement and questioning, the learning process of operators was shown to be promoted, albeit slowly, in an almost abductive spirit. Possibly, this outcome can be seen as a practical exemplification of what Pierce means by abduction as a form of inference (Fann, 1970). Whether this entails that the result applies even beyond the work task studied here, however, can be regarded as an empirical question, upon which further studies may shed light. In conclusion, a discussion is conducted on grading of the concept of accommodation and on the weight to be attached to cyclical completeness (Kolb, 1984); there is a question mark in relation to whether the "cycle" is genuinely cyclical by nature. Learning from experience at work can be regarded as a field in which much research remains to be conducted. Continued studies from a contextual-didactic, constructivist perspective might usefully be performed so as to extend my work to encompass learning in relation to other specific work tasks. Their aim would be better to understand task-intrinsic learning, taking a point of departure in the significance of the encounter with the task and the situation in a specific context. Of relevance, for example, would be to build on current knowledge as part of a cumulative process, in which work tasks, selected according to some strategic principle of similarity and dissimilarity, are zoomed in upon. The selection of tasks should be made on the basis of strategic concepts that may be envisaged to be of significance.