What is the ‘covering law’ model of explanation? Essay
What is the ‘covering law’ model of explanation?
Carl Hempel’s “covering law” model of explanation states essentially that an explanation for an event can be drawn from a set of general laws or, in the case of the social sciences, universal hypotheses. Hempel claims the study of history is not generally associated with the search for general laws governing historical events.
However, history is a discipline within which the theory of “covering law”—with some slight modifications—can function. Hempel’s model is one of deductive reasoning in which two sets of information are paired to develop a hypothesis: one set includes all the facts of an event (time, place, actions, etc) while the second includes applicable “empirical laws” (laws governing the variables in situations similar to those in the first set).
Hempel claims that the “covering law” is relevant to social sciences because, like in the natural sciences, because both have similar strengths and weaknesses with regard to their ability, or lack of ability, to “grasp the unique individuality” of their objects. Hempel is adamant about the use of empirical laws, and cautions, especially in the social sciences, against using romantic ideals such as “destiny” and “mission in history” in place of scientific explanations.
To determine the suitability of an explanation, Hempel states that the explanation must satisfactorily pass a series of tests. These tests include empirical tests of “the sentences which state the determining conditions” and “the universal hypotheses on which the explanation rests” in addition to “an investigation of whether the explanation is logically conclusive in the sense that the sentence describing the events to be explained follows the statements” of the two sets of information (45).
Reference: Hempel, Carl. “The Function of General Laws in History. ” 2. Is the model applicable to the study of human phenomena? While Carl Hempel asserts that his “covering law” model of explanation is applicable to social sciences, the reality of this claim is called into question by several other authors. F. A. Hayek’s initial criticism of Hempel’s model is that is can only serve a reactionary function, because it relies on the identification of patterns to initiate inquiry.
Thus, predicting outcomes is nearly impossible without first being able to recognize problems and all relevant variables. In the case of history the model might have some limited use, but only when both the causes and effects have already been identified. Even in the case of history then the applicability of the “covering law” model is limited by the mind’s ability to locate patterns. This would not be such an insurmountable challenge if the patterns were themselves easily discernable.
Indeed, the principle fault Hayek finds with the “covering law” model is that while it can be applied to social sciences theoretically, it is unable to cope with the complexity of issues and variables inherent in “the more complex phenomena of life, mind, and of society. ” Hayek determines a patterns level of complexity by seeking the “minimum number of elements of which an instance of the pattern must consists in order to exhibit al the characteristic attributes of the class of patterns in question appears to provide an unambiguous criterion” (56).
That is essentially, how much of the pattern must be present to determine the pattern’s existence with certainty. Hayek goes on to assert that in the study of human phenomenon, “individual events regularly depend on so many concrete circumstances” that it is impossible to determine them all especially as many of the interactions and their results are not observable. This leads to his closing discussion of relativism. Even if it were possible to determine all of the circumstances that give rise to a specific condition, it is impossible to conclude the exact interaction of the circumstances.
As Hayek asserts, “while we know that all those values are relative to something, we do not know to what they are relative”(64). In absence of this knowledge, the actual cause and effect relationship between a given set of circumstances and a condition may seem to exist, but is impossible to prove unambiguously. While Michael Scriven begins by enumerating the ways in which physical and social sciences (though he shies away from such terminology) are similar, he quickly turns to examining their differences.
Foremost of these is his claim that “practical problems of prediction, or explanation at any level…are more like to be insoluble in the study of behavior” (72). In general, Scriven agrees with Hayek’s argument about the problem of data collection in establishing explanations for human behavior. Yet, Scriven extends the problem beyond asking about how one might find the formulas that unlock patterns to stating that in the study of human behavior there is no reason to believe that a formula or explanation must exist.
Again returning to similarities among all sciences, Scriven claims that “exact predictions and faultless explanations” will be difficult, if not impossible, to establish in the social sciences because the physical sciences (which all authors seem at least to some extent to concede are simpler applications for the “covering law” model) rarely produce such conclusive results. Further, Scriven compares the successes of physical scientists such as Galileo and Dalton to the possibilities and realities of social scientists, and finds the latter to be far less promising given the number of variables with which social scientists must content.
This in conjunction with the lack of precise laws already in existence make it unlikely that social scientists will ever be able to experience success in the location of fundamental basic laws that govern human behavior. Even when groundbreaking work such as that of Freud is undertaken, it is “nonquantitative” and uncovers rules of abnormal behavior rather than any which might be generally applicable.
In his essay, “Psychology as Philosophy’” Donald Davidson’s analysis of the applicability of the “covering law” in the social sciences is more subtle than that of Hayek and Scriven, yet in the end they all come to the same conclusion that Hempel’s model is inappropriate for the explanation of human phenomenon. Davidson seeks to evaluate the “covering law” by examining “the arguments against the possibility of deterministic laws of behavior” and his discussion applies specifically to psychology. Davidson’s tone is often more conciliatory than the other two authors, but his analysis is in many ways more precise.
He states that in order for the “covering law” to be useful it must have the ability to be predictive. These predications would most likely be based on human desires and beliefs, and in order to predict actions on these bases it is necessary to have “a quantitative calculus that brings all relevant beliefs and desires into the picture,” which would be impossible (81). Citing a study that sought to quantify “relations between actions, and treats wants and thoughts as theoretical constructs,” Donaldson locates yet another problematic variable: time.
Even if one could determine all the beliefs and desires that a subject would factor into a decision, the process of thinking is dynamic and will change over time. Donaldson’s issue with Hempel’s model as applied to the social sciences can be summarized by stating that the “covering law” presupposes a level of rationality on the part of humans, and that actions are based solely on desires and beliefs. The problem here is that psychology is not a closed system, that is it is effected by outside variables for which social scientists cannot control putting it in stark contrast to the methodologies employed by the physical sciences.
References: Hayek, F. A. “The theory of Complex Phenomena. ” Scriven, Michael. “A possible distinction between traditional scientific discipline and the study of human behavior. ” Davidson, Donald. “Psychology and Philosophy” 3. Is that model of explanation opposed to the critical interpretation of human behavior? In contrast to Hempel, Hayek, Scriven, and Davidson, philosophers R. G. Collingwood and Willam Dray call into question the entire notion of using physical science methodologies to approach problems in the social sciences.
According to their analyses, the “covering law” model is a misguided attempt at understanding human nature (as opposed to “nature” in general and the “animal nature of humans”). Dray, more direct and harsh in his critique than Collingwood, even goes so far as to deem it “inept” as a means for explaining human behavior. They thus propose a very different approach to the study of the social sciences predicated on the methodologies employed by the discipline of history as rather than those of the physical sciences.
To this end, they also seek within their essays to define the goals and techniques of historical inquiry and place these in opposition to those of the physical sciences. Collingwood claims that physics found success in the 17th and 18th centuries because it located the correct method for the types of issues the discipline sought to address. Given it’s success and the resulting contemporary notion that all reality was in someway physical, it is understandable that individuals seeking to explore the science of human nature thought it advisable to adopt a similar methodology.
This, according to Collingwood, was a key error that sent social sciences off in a direction that would never lead them to similar success. Collingwood asserts his contention that the physical and social sciences are, at their cores different by using what he sees to be the strongest, methodologically, of the respective sciences: physics and history. Whereas he sees the methodologies of physics as a false hope for the social sciences, he advises that they should instead look to history as a model of the most successful methodological discipline within the social sciences.
Physics, he says provides the “right way of investigating nature,” while history provides the “right way of investigating mind” (168). Dray makes a similar distinction, but does it in a different way. He claims that in analyzing an action, historians must attempt to determine the reasons for that action. It is his view that the historian’s goal is “to show that what was done was the thing to have done for the reasons given, rather than merely the thing that is done on such occasions” (176).
This is in contrast to the physical scientist who would be more interested in determining what is generally don on such occasions, thus locating a general law or, in Hempel’s terminology, a universal hypothesis. Collingwood and Dray both understand history to be the study of human actions with the goal of locating the thoughts that inform them. Here Collingwood makes a distinction similar to some of the earlier authors in saying that the physical scientists look to “goes beyond an event, observes it relation to others, and thus brings it under a general formula or law of nature.
” Where his definitions diverge is in his assertion that while scientists study phenomenon, “events of history are never mere phenomena,” and that the task of the historian is to look “not at, but through [these events] to discern the thought within them” (168). The difference here is subtle, and is not helped by Collingwood’s use of terms such as “event” in different contexts, for he later claims that history if not the study of the “processes of events but of processes of actions…consisting of processes of thought, and what the historian is looking for is these processes of thought” (169).
One extremely interesting aspect of Collingwood and Dray’s conceptions of history is his claim that historians in order to understand the actions and thoughts of individuals in the past must attempt to think the thoughts of those people. While this is not necessarily in conflict with the “covering law” its extreme reliance on subjectivity does recall some potential reasons why the social sciences sometimes receive less respect than the physical sciences.
It also hints at insights as to why social scientists might thus look to develop more at least outwardly objective methodologies as a means of legitimizing their fields in the eyes of a world that values objectivity so highly. This discussion rises from Collingwood argument that things (people, institutions, etc) must be studied as changing entities that can be seen “only as a phase in a process leading from a very different past to a very different future” (167). This of course opens the issue of evolution, a process that his extremely historical, and yet has been the domain of the natural sciences.
Collingwood uses this example to recognize the potential grey area between history of human nature and nature itself. He seeks to rectify this by differentiating between two types of human actions: those that are purely in response to one’s “animal nature…impulses and appetites” are deemed “nonhistorical” and left to the realm of physical science, and those that are “social customs…within which these appetites fin satisfaction in ways sanctioned by convention and morality” (170).
In his final paragraph, Collingwood claims that natural processes could be considered historical if something like intelligent design were the case—that is, if natural processes could be explained as a result of some larger thought or plan analogous to the types that humans create and act upon. References: Collingwood, R. G. “Human Nature and Human History. ” Dray, William. “Rationale of Actions. ” 4. Can that model account for the phenomenon of reflexive predictions?
The phenomenon of reflexive predictions provides a means by which to test the “covering law” model defended and refuted in the readings. Reflexive predictions, as defined by Geroge D. Romanos are those prediction that by their very existence alter the outcome of the event whose outcome they seek to predict. Reflexive predictions can either be self-fulfilling (if the predicted outcome is achieved) or self-frustrating (if the prediction is proven false).
In his essay, Romanos discusses a debate among philosophers about the existence of reflexive predictions in the natural sciences. As all of the authors he discusses seem to be in agreement that reflexive predictions occur in the social sciences, the debate center around whether they are solely a social science phenomenon as can be found within physical science as well. This debate then touches on many of the issues presented by the other authors. Specifically, it addresses the inherent differences and similarities between the nature of the physical and social sciences.
In his essay, Romanos cites a debate about whether a machine changes its course do to an alteration in its instructions can be considered analogous to a person acting in a certain way because of a prediction. In the machine analogy, the instructing machine had noted an error in calculation and thus, in the context of the argument, made a prediction that it would be wrong, and altered the original instructions to correct the error and complete the task successfully.
The essential point that Romanos makes is that humans may receive predictions, and choose to alter their behavior accordingly, but they do so because they believe the prediction to be true (or false). Whereas, a machine does not act because it believes the instruction, but rather because they are instructions, and it has no choice to whether nor not to follow them. This distinction is similar to that made by Collingwood and Dray when they assert that history is the study of human actions that are based on beliefs and desires. To them as to Romanos, merely following orders does not fall within the scope of the social sciences.
The machine example can thus be likened to the “animal nature” that Collingwood discusses as physical and nonhistorical. After much consideration, however, Romanos throws out this idea that for a prediction to be reflexive it must stem from an “acting-on-beliefs model. ” He thus reestablishes the possibility of reflexive predictions in the physical sciences, though he concedes that the “acting-on-beliefs model” is likely commonly accepted because this model has held true for “most of the cases (if not all)” where reflexive predictions have been located.
One requirement established by other philosophers is that the dissemination of a prediction must be a factor in establishing it as reflexive. This distinction is important, because without the dissemination, the prediction can merely be seen as true of false, but no causal relationship can be determined between the prediction and the outcome if the prediction were wholly unknown to the actors be they human, animal, or machine. Romanos argues that the while term “dissemination” may imply communication on a human level, this is not necessarily the case.
In the example of machines, the issuance of a second set of instruction can qualify as dissemination of information that implies a prediction. This too obviously has implications for the physical sciences, because the use of this term has resulted in the assumption that reflexive predictions could not take place, and yet again Romanos attempts to demonstrate that this is not the case. For Romanos, an important point about reflexive predictions is that they express a prediction in physical terms, thus making them a physical phenomenon.
According to his argument, it is inappropriate to apply a nonphysical standard—the “acting-on-beliefs model”— to satisfy a physical phenomenon. Doing so is a conflation of the two very separate ideas. For him, the only necessary condition in this area is that there be a causal relationship between the prediction and the outcome. The reason for the relationship is therefore not essential to it mere existence. This then is the second major point that Romanos disposes of, and like the first, his analysis leaves open the possibility of the existence reflexive predictions in the physical sciences.
Romanos notes that “acting-on-beliefs” is just one model for reflexive predictions. This model, he argues prejudices reflexive predictions toward the social sciences, as it requires “social actors ‘acting-on’ certain beliefs” (156). As in other cases, he concedes that this may be the most prevalent model, and yet it is not necessarily the only one that exists, and assuming its universality is a fallacy based in the fact that the most common and obvious reflexive predictions satisfy this model and then assuming that all other possibilities must as well.
Romanos’ project is not to prove that reflexive predictions are possible within the physical sciences, but rather to disprove the claims that attempt to support its impossibility. On this he may be considered successful, but in relation to the “covering law” model, his result can be seen as proportionate to the goals of his essay. He does not, at any point, prove that the social sciences are necessarily distinct methodologically from the physical sciences. Instead, he dismantles some very specific notions that assume they must be so. In essence, Romanos seeks to prove that the two are differentiated only theoretically.
Thus, for him, the fact that only examples from the social sciences have been found adequate to the definition of reflexive predictions does not mean that those are the only possibilities. In his view, the prediction that the reflexive prediction cannot exist in the physical sciences is potentially a reflexive one itself (though it can be noted that this example too falls within the social sciences). In noting that this is impossible, thinkers have constructed a definition that attempts to exclude the existence of potential examples from the physical sciences. Reference: Romanos, George D. “Reflexive Predictions. ”