27th Feb 2012 HPS250H Lecture 7 Lakatos
27th Feb 2012 HPS250H Lecture 7 Lakatos (ie. Kuhn’s viewpoint on the issue of: “how can we obtain absolute knowledge”?) CARRYING ON FROM LAST LECTURE: Kuhn’s “incommensurability problem”: - premise one: translatability requirement In order to be comparable, two competing theories must employ mutually translatable taxonomies. - premise two incommensurability Competing theories can employ untranslatable taxonomies, i.e. they can be incommensurable. - conclusion reached incomparability It is not always possible to compare two competing theories. ------------------------------------------------------------------------------------Pierre Duhem Scientific experiments are completely interpenetrated by theoretical interpretation, to the point where it becomes impossible to express fact in isolation from theory. Thus, it is impossible to falsify an individual scientific proposition. an example: with the theory: - A thread breaks when it is loaded with a weight that exceeds the thread’s tensile strength. This theory can “avoid” faisification by many ways! ie. the experiment results can contradict the theory’s predictions because of many reasons: - eg. the theory could be false - eg. the experimenter reports another conclusion - eg. the tensile strength of the thread was actually more than reported - eg. the weight was reported as 1 kg, but was actually 2 kg - etc (see ppt slides)
There are hundreds of ways to save a theory from falsification. In order to make a proposition testable, we must add many auxiliary hypotheses about working of the instruments, about initial conditions, etc. - Non-Separability An individual theoretical proposition has no empirical consequences and cannot be tested in isolation. - Non-Falsifiability An individual theoretical proposition is not falsifiable, i.e. there are many ways to save it from falsification. So, Kuhn = problem of incommensurability - Competing theories can employ untranslatable taxonomies, i.e. they can be incommensurable The dialog across paradigms is problematic, because of incommensurability. Pierre Duhem= problem of non- separability - An individual theoretical proposition has no empirical consequences and cannot be tested in isolation. How can we evaluate our competing theories in these circumstances? What happens to the idea of scientific progress? Lakatos’s theory (of research programmes) bypasses both problems above! Lakatos: The challenge is to show how the growth of knowledge is possible in spite of non-separability and incommensurability. My basic idea is that the unit of appraisal is not an individual theory, but a modification of a theory, i.e. a transition from one theory to another, or a research programme, as I call it. Hard Core ≡ Fundamental propositions that define the programme. Within the programme they are never abandoned, but protected at all costs.
Protective Belt ≡
Auxiliary hypotheses which are modified in order to shield the hard core from falsification and to make it applicable to new phenomena. Concerning the hard core, But these laws alone are not sufficient to make specific predictions about observable processes. Concerning the protective belt: - In order to make specific predictions, a research programme needs some auxiliary hypotheses. - The protective belt of a programme can include individual hypotheses as well as full-fledged theories. - The protective belt is modified in order to explain some anomaly or to extend the scope of the theory. - When we add some auxiliary hypotheses we thus obtain - a specific theory within a research programme.
A research programme can be actually tested only when the propositions of the hard core are combined with the propositions of the protective belt. ------------------------------------------------------------------------------------------------progressive research programmes regressive research programmes -
The initial state of the research programme. When we modify the protective belt, we obtain a new theory within the same research programme
Progressive modification ≡ A modification that satisfies all of the following requirements: It increases the empirical content (makes predictions more detailed or precise, or predicts novel facts); Some of this excess content has been corroborated; It is in organic unity with the rest of the programme. This is an extremely ambiguous idea. Roughly, the idea is to ensure a high degree of continuity in the programme and protect it against threats of fragmentation, disintegration.
Regressive modification ≡ Regressive modifications are also called ad hoc. Ad hoc modification ≡ A modification of any of the following types: Ad hoc1 modification It does not increase the empirical content of the programme; Ad hoc2 modification It does increase the content, but none of the excess content has been corroborated; Lakatos: These above two requirements are Popper’s. Ad hoc3 modification It introduces a hypothesis foreign to the spirit of the programme. Lakatos: To ensure the growth of scientific knowledge, we must avoid the three ad hoc stratagems. It is these modifications of the protective belt that are subject to rational appraisal. The task is to find out whether a particular transition was progressive or regressive an example of the underlined points above: Newtonian Research Programme predicting the orbit of the moon
“track record” - is how we compare competing research programmes (and see which one is better) - For every research programme, we can examine its historical track record. eg. The track record of this programme shows that it has been mostly progressing. The track record of this programme shows that it has been gradually degenerating. Therefore we can conclude that, at the moment, Research Programme 1 has a better record compared to its competitor Research Programme 2. --------------------------------------------------------------------------------------------------
Therefore, according to Lakatos: The “translatability requirement” premise is false: - because g The “incommensurability” premise can be accepted but incommensurability is not an obstacle to scientific progress! -------------------------------------------------------------------------------------------------Two new premises and conclusion: translatability requirement In order to be comparable, two competing theories must employ mutually translatable taxonomies. Lakatos: When comparing two research programmes, we do not collate their empirical contents, but their progressiveness. - The translatability requirement is not the case. Untranslatable taxonomies are not an obstacle for theory comparison. It is possible to compare two competing theories only when their predictions are expressible in a common language.
incommensurability Competing theories can employ untranslatable taxonomies, i.e. they can be incommensurable.
Lakatos: Thus, this assumption is false. We can compare research programmes even when they are not mutually translatable. Lakatos: Even if we were to accept incommensurability Incommensurability is not an obstacle. - The incommensurability thesis is not accurate. - Although competing theories often employ incommensurable taxonomies, we always find ways to express their predictions in a common language. Predictions of competing theories are expressible in a common language even when the theories themselves are incommensurable. - From the two bolded premises above, is deduced the conclusion: Comparability It is always possible to compare competing theories. and not: incomparability
It is not always possible to compare two competing theories (programmes)
Therefore, incommensurability has never been an obstacle for theory comparison.
There are hundreds of ways to save a theory from falsification. In order to make a proposition testable, we must add many auxiliary hypotheses about working of the instruments, about initial conditions, etc. - Non-Separability An individual theoretical proposition has no empirical consequences and cannot be tested in isolation. - Non-Falsifiability An individual theoretical proposition is not falsifiable, i.e. there are many ways to save it from falsification. So, Kuhn = problem of incommensurability - Competing theories can employ untranslatable taxonomies, i.e. they can be incommensurable The dialog across paradigms is problematic, because of incommensurability. Pierre Duhem= problem of non- separability - An individual theoretical proposition has no empirical consequences and cannot be tested in isolation. How can we evaluate our competing theories in these circumstances? What happens to the idea of scientific progress? Lakatos’s theory (of research programmes) bypasses both problems above! Lakatos: The challenge is to show how the growth of knowledge is possible in spite of non-separability and incommensurability. My basic idea is that the unit of appraisal is not an individual theory, but a modification of a theory, i.e. a transition from one theory to another, or a research programme, as I call it. Hard Core ≡ Fundamental propositions that define the programme. Within the programme they are never abandoned, but protected at all costs.
Protective Belt ≡
Auxiliary hypotheses which are modified in order to shield the hard core from falsification and to make it applicable to new phenomena. Concerning the hard core, But these laws alone are not sufficient to make specific predictions about observable processes. Concerning the protective belt: - In order to make specific predictions, a research programme needs some auxiliary hypotheses. - The protective belt of a programme can include individual hypotheses as well as full-fledged theories. - The protective belt is modified in order to explain some anomaly or to extend the scope of the theory. - When we add some auxiliary hypotheses we thus obtain - a specific theory within a research programme.
A research programme can be actually tested only when the propositions of the hard core are combined with the propositions of the protective belt. ------------------------------------------------------------------------------------------------progressive research programmes regressive research programmes -
The initial state of the research programme. When we modify the protective belt, we obtain a new theory within the same research programme
Progressive modification ≡ A modification that satisfies all of the following requirements: It increases the empirical content (makes predictions more detailed or precise, or predicts novel facts); Some of this excess content has been corroborated; It is in organic unity with the rest of the programme. This is an extremely ambiguous idea. Roughly, the idea is to ensure a high degree of continuity in the programme and protect it against threats of fragmentation, disintegration.
Regressive modification ≡ Regressive modifications are also called ad hoc. Ad hoc modification ≡ A modification of any of the following types: Ad hoc1 modification It does not increase the empirical content of the programme; Ad hoc2 modification It does increase the content, but none of the excess content has been corroborated; Lakatos: These above two requirements are Popper’s. Ad hoc3 modification It introduces a hypothesis foreign to the spirit of the programme. Lakatos: To ensure the growth of scientific knowledge, we must avoid the three ad hoc stratagems. It is these modifications of the protective belt that are subject to rational appraisal. The task is to find out whether a particular transition was progressive or regressive an example of the underlined points above: Newtonian Research Programme predicting the orbit of the moon
“track record” - is how we compare competing research programmes (and see which one is better) - For every research programme, we can examine its historical track record. eg. The track record of this programme shows that it has been mostly progressing. The track record of this programme shows that it has been gradually degenerating. Therefore we can conclude that, at the moment, Research Programme 1 has a better record compared to its competitor Research Programme 2. --------------------------------------------------------------------------------------------------
Therefore, according to Lakatos: The “translatability requirement” premise is false: - because g The “incommensurability” premise can be accepted but incommensurability is not an obstacle to scientific progress! -------------------------------------------------------------------------------------------------Two new premises and conclusion: translatability requirement In order to be comparable, two competing theories must employ mutually translatable taxonomies. Lakatos: When comparing two research programmes, we do not collate their empirical contents, but their progressiveness. - The translatability requirement is not the case. Untranslatable taxonomies are not an obstacle for theory comparison. It is possible to compare two competing theories only when their predictions are expressible in a common language.
incommensurability Competing theories can employ untranslatable taxonomies, i.e. they can be incommensurable.
Lakatos: Thus, this assumption is false. We can compare research programmes even when they are not mutually translatable. Lakatos: Even if we were to accept incommensurability Incommensurability is not an obstacle. - The incommensurability thesis is not accurate. - Although competing theories often employ incommensurable taxonomies, we always find ways to express their predictions in a common language. Predictions of competing theories are expressible in a common language even when the theories themselves are incommensurable. - From the two bolded premises above, is deduced the conclusion: Comparability It is always possible to compare competing theories. and not: incomparability
It is not always possible to compare two competing theories (programmes)
Therefore, incommensurability has never been an obstacle for theory comparison.
