Scientific realism is the view that science aims to provide us, in its theories, with a literally true account of the nature of the objective world. Anti-realism rejects this, and claims that science provides, at most, a convenient model that accounts for the phenomena in the observable world. One anti-realist position, namely the constructive empiricism of Bas van Fraassen (1980), states that theories aim to be empirically adequate, and the acceptance of a theory involves no more than the belief that it is empirically adequate. The following are the arguments against scientific realism.
Underdetermination
The formation of scientific theories involves two main steps. First is the gathering of empirical data acquired through observation, and second is the construction of the theory that accounts for this data via an inferential process. The second step is non-deductive. Hence, the theory is not entailed by the data. Rather, it goes beyond the data to account for and unify it.
As noted by van Fraassen in The Scientific Image (1980), this allows the underdetermination of theory by data. Since the empirical data does not entail the theory, there can be several different competing theories that each accounts for it. Furthermore, since each of these theories are based on and account for the same data, the scientist cannot select one theory from a set of theories from the data alone. The theories are empirically equivalent, and so extra premises are required to select a theory in favour of its rivals. These are known as superempirical virtues.
In The Scientific Image, van Fraassen argues that these superempirical virtues that are used to select a theory from its empirically equivalent rivals do not reflect the truth of the theory, but rather its pragmatic usefulness and æsthetic appeal. Simplicity is a good example of such a superempirical virtue. The principle of Ockham's razor states that a theory should not multiply properties unnecessarily, and so simple theories are favoured over their empircally equivalent but complex rivals. However, this is for pragmatic and æsthetic reasons, rather than a reflection of the theory's truth-conduciveness. Simplicity makes the theory more comprehensible, convenient, and satisfying to the human mind. In fact, the claim that simplicity is an indicator of the truth of a theory has no justification unless one already knows that the world is simple.
The Pessimistic Meta-Induction
This argument is championed by Larry Laudan in his paper "A Confutation of Convergent Realism" (1981). He states that for a theory to be true, it must be empirically successful, and its terms must refer to actual phenomena in the objective world. If the theory is not empirically successful, it cannot be true, since it contradicts our experience of the world. Similarly, if a theory's terms do not refer to actual phenomena, it cannot be true, since it directly contradicts the nature of the world.
Laudan presents an extensive list of past theories which were once empirically successful. Furthermore, the terms of these theories, in their time, were considered to refer to actual phenomena, but are now considered not to refer. In other words, these theories were once considered to be true, but are now considered false. Among these theories are the celestial sphere theory, the phlogiston theory, and the luminiferous æther theory.
Using induction, Laudan argues that since these previously accepted and once empirically successful theories were subsequently considered false, current and future accepted and empirically successful theories will also subsequently be considered false. In other words, modern theories may at present have empirical and predictive success, but so did previous theories in their day. These previous theories have since been rejected, and so by induction it is likely that modern theories will too be rejected. This principle extends not only to current theories, but also to all future theories. Therefore, since all current and future theories will subsequently be rejected, realism is false.
Verisimilitude
The pessimistic meta-induction suggests that the empirical success of a theory does not guarantee that its central terms refer to actual phenomena, and so does not guarantee its truth. One of the realist objections to this argument refers to the notion of approximate truth, or verisimilitude. Although the empirical success of a theory does not guarantee its absolute truth, it does indicate that the theory is at least approximately true. Furthermore, although past empirically successful and hence approximately true theories have been rejected, they have been rejected in favour of new theories which are even closer to the truth. Thus, scientific knowledge is cumulative in the sense that newer theories are more approximately true than their predecessors.
The anti-realist argument against this claim is that the concept of verisimilitude is arbitrary and meaningless. As Laudan points out, there has been no successful definition of what it means for a statement to be approximately true. For example, consider a glass of water. One person states that it is a bottle of water, while another states that it is a glass of whiskey. Which statement is closer to the truth? The first is closer with respect to the type of liquid, whereas the second is closer with respect to the type of container. Nevertheless, both statements are strictly false. This suggests that the notion of verisimilitude is arbitrary, and so it is meaningless to claim that one theory is closer to the truth than another.
This argument holds even for numbers. For example, is 8 closer to 7, 4, or 3? One may argue that it is closest to 7 because of the smallest numerical difference. However, one may also argue that it is closer to 4 because 8 is a multiple of 4, or that it is closer to 3 because both the figures 8 and 3 contain curved lines. Nevertheless, none of these numbers equals 8.
What this critique of verisimilitude illustrates is that statements which are considered to be approximately true are actually strictly false, and to claim that one false theory is closer to the truth than another is arbitrary and meaningless. Therefore, scientific knowledge is not cumulative, and realism is false.
The Incommensurability Thesis
In The Structure of Scientific Revolutions (1962), Thomas Kuhn argues that the evolution of scientific theory does not follow a linear process of the accumulation of knowledge, but involves non-linear paradigm shifts as a result of changing intellectual circumstances. Kuhn distinguishes two types of scientific activity. Firstly, normal science involves routine problem-solving within an accepted framework of meanings, concepts, and interpretations, known as a paradigm. Secondly, revolutionary science occurs when the accumulation of anomalous data casts doubt on the success of the current paradigm, and involves a shift to a new paradigm when a sufficient number of the scientific community are persuaded that the new paradigm is preferable to the old. Examples of paradigm shifts include the shift from Newtonian mechanics to Einsteinian relativity, the shift from Ptolemaic cosmology to Copernican cosmology, the shift from the phlogiston theory of combustion to Lavoiser's theory of chemical reactions, and the shift from classical physics to quantum mechanics.
Kuhn argues that because any two paradigms are so radically different from each other, there is no meaningful way to compare them objectively. In other words, different paradigms have different meanings of terms and concepts, different interpretations of data, and different values. Therefore, there are no common measures by which they can be objectively compared. This is the principle of incommensurability.
For example, although the concept of mass features in both Newtonian mechanics and Einsteinian relativity, the term has a different meaning in each paradigm. Therefore, the two paradigms cannot be compared in a meaningful way with respect to the term mass. Furthermore, consider the fact that all observation is influenced by the background framework under which it is observed. Because of the theory-ladenness of observation, empirical data is interpreted differently in each paradigm. Therefore, empirical data cannot be used to favour one paradigm over another. Finally, different paradigms have different criteria and values for what constitutes a good theory. Therefore, any two paradigms cannot be compared with each other objectively, because they possess different values for comparison.
Because one paradigm cannot objectively be considered better or worse than another paradigm, it follows that scientific knowledge is not cumulative. Different paradigms have different criteria for the validity of theories, and so there is no objectively meaningful and independent means by which one theory can be considered more or less valid than another. Therefore, realism is false.
Conclusions
The anti-realist arguments considered here claim that scientific theories are not representative of objective truth, but provide a means of accounting for phenomena that we observe in a convenient and comprehensible way. The virtues that influence our decisions in a choice between competing theories are not truth-conducive, but pragmatic, æsthetic, and social. Nevertheless, this does not make them any less valid reasons for choosing a theory, for the theory’s reception, understanding, and application are greatly aided by these virtues. Theories, after all, despite their incapability to provide us with a literal account of objective truth, serve to facilitate our judgements about the world, and so, it is a given requirement that they be pragmatic.
References
Underdetermination
The formation of scientific theories involves two main steps. First is the gathering of empirical data acquired through observation, and second is the construction of the theory that accounts for this data via an inferential process. The second step is non-deductive. Hence, the theory is not entailed by the data. Rather, it goes beyond the data to account for and unify it.
As noted by van Fraassen in The Scientific Image (1980), this allows the underdetermination of theory by data. Since the empirical data does not entail the theory, there can be several different competing theories that each accounts for it. Furthermore, since each of these theories are based on and account for the same data, the scientist cannot select one theory from a set of theories from the data alone. The theories are empirically equivalent, and so extra premises are required to select a theory in favour of its rivals. These are known as superempirical virtues.
In The Scientific Image, van Fraassen argues that these superempirical virtues that are used to select a theory from its empirically equivalent rivals do not reflect the truth of the theory, but rather its pragmatic usefulness and æsthetic appeal. Simplicity is a good example of such a superempirical virtue. The principle of Ockham's razor states that a theory should not multiply properties unnecessarily, and so simple theories are favoured over their empircally equivalent but complex rivals. However, this is for pragmatic and æsthetic reasons, rather than a reflection of the theory's truth-conduciveness. Simplicity makes the theory more comprehensible, convenient, and satisfying to the human mind. In fact, the claim that simplicity is an indicator of the truth of a theory has no justification unless one already knows that the world is simple.
The Pessimistic Meta-Induction
This argument is championed by Larry Laudan in his paper "A Confutation of Convergent Realism" (1981). He states that for a theory to be true, it must be empirically successful, and its terms must refer to actual phenomena in the objective world. If the theory is not empirically successful, it cannot be true, since it contradicts our experience of the world. Similarly, if a theory's terms do not refer to actual phenomena, it cannot be true, since it directly contradicts the nature of the world.
Laudan presents an extensive list of past theories which were once empirically successful. Furthermore, the terms of these theories, in their time, were considered to refer to actual phenomena, but are now considered not to refer. In other words, these theories were once considered to be true, but are now considered false. Among these theories are the celestial sphere theory, the phlogiston theory, and the luminiferous æther theory.
Using induction, Laudan argues that since these previously accepted and once empirically successful theories were subsequently considered false, current and future accepted and empirically successful theories will also subsequently be considered false. In other words, modern theories may at present have empirical and predictive success, but so did previous theories in their day. These previous theories have since been rejected, and so by induction it is likely that modern theories will too be rejected. This principle extends not only to current theories, but also to all future theories. Therefore, since all current and future theories will subsequently be rejected, realism is false.
Verisimilitude
The pessimistic meta-induction suggests that the empirical success of a theory does not guarantee that its central terms refer to actual phenomena, and so does not guarantee its truth. One of the realist objections to this argument refers to the notion of approximate truth, or verisimilitude. Although the empirical success of a theory does not guarantee its absolute truth, it does indicate that the theory is at least approximately true. Furthermore, although past empirically successful and hence approximately true theories have been rejected, they have been rejected in favour of new theories which are even closer to the truth. Thus, scientific knowledge is cumulative in the sense that newer theories are more approximately true than their predecessors.
The anti-realist argument against this claim is that the concept of verisimilitude is arbitrary and meaningless. As Laudan points out, there has been no successful definition of what it means for a statement to be approximately true. For example, consider a glass of water. One person states that it is a bottle of water, while another states that it is a glass of whiskey. Which statement is closer to the truth? The first is closer with respect to the type of liquid, whereas the second is closer with respect to the type of container. Nevertheless, both statements are strictly false. This suggests that the notion of verisimilitude is arbitrary, and so it is meaningless to claim that one theory is closer to the truth than another.
This argument holds even for numbers. For example, is 8 closer to 7, 4, or 3? One may argue that it is closest to 7 because of the smallest numerical difference. However, one may also argue that it is closer to 4 because 8 is a multiple of 4, or that it is closer to 3 because both the figures 8 and 3 contain curved lines. Nevertheless, none of these numbers equals 8.
What this critique of verisimilitude illustrates is that statements which are considered to be approximately true are actually strictly false, and to claim that one false theory is closer to the truth than another is arbitrary and meaningless. Therefore, scientific knowledge is not cumulative, and realism is false.
The Incommensurability Thesis
In The Structure of Scientific Revolutions (1962), Thomas Kuhn argues that the evolution of scientific theory does not follow a linear process of the accumulation of knowledge, but involves non-linear paradigm shifts as a result of changing intellectual circumstances. Kuhn distinguishes two types of scientific activity. Firstly, normal science involves routine problem-solving within an accepted framework of meanings, concepts, and interpretations, known as a paradigm. Secondly, revolutionary science occurs when the accumulation of anomalous data casts doubt on the success of the current paradigm, and involves a shift to a new paradigm when a sufficient number of the scientific community are persuaded that the new paradigm is preferable to the old. Examples of paradigm shifts include the shift from Newtonian mechanics to Einsteinian relativity, the shift from Ptolemaic cosmology to Copernican cosmology, the shift from the phlogiston theory of combustion to Lavoiser's theory of chemical reactions, and the shift from classical physics to quantum mechanics.
Kuhn argues that because any two paradigms are so radically different from each other, there is no meaningful way to compare them objectively. In other words, different paradigms have different meanings of terms and concepts, different interpretations of data, and different values. Therefore, there are no common measures by which they can be objectively compared. This is the principle of incommensurability.
For example, although the concept of mass features in both Newtonian mechanics and Einsteinian relativity, the term has a different meaning in each paradigm. Therefore, the two paradigms cannot be compared in a meaningful way with respect to the term mass. Furthermore, consider the fact that all observation is influenced by the background framework under which it is observed. Because of the theory-ladenness of observation, empirical data is interpreted differently in each paradigm. Therefore, empirical data cannot be used to favour one paradigm over another. Finally, different paradigms have different criteria and values for what constitutes a good theory. Therefore, any two paradigms cannot be compared with each other objectively, because they possess different values for comparison.
Because one paradigm cannot objectively be considered better or worse than another paradigm, it follows that scientific knowledge is not cumulative. Different paradigms have different criteria for the validity of theories, and so there is no objectively meaningful and independent means by which one theory can be considered more or less valid than another. Therefore, realism is false.
Conclusions
The anti-realist arguments considered here claim that scientific theories are not representative of objective truth, but provide a means of accounting for phenomena that we observe in a convenient and comprehensible way. The virtues that influence our decisions in a choice between competing theories are not truth-conducive, but pragmatic, æsthetic, and social. Nevertheless, this does not make them any less valid reasons for choosing a theory, for the theory’s reception, understanding, and application are greatly aided by these virtues. Theories, after all, despite their incapability to provide us with a literal account of objective truth, serve to facilitate our judgements about the world, and so, it is a given requirement that they be pragmatic.
References
- Kuhn TS (1962). The Structure of Scientific Revolutions. Chicago: University of Chicago Press.
- Laudan L (1981). "A Confutation of Convergent Realism". Philosophy of Science, 48:1, 19-49.
- van Fraassen BC (1980). The Scientific Image. Oxford: Clarendon Press.
0 comments:
Post a Comment