Falsification HM5501 Lecture 4

October 15, 2017

Karl_PopperWhat we want is some criterion which will allow us to distinguish science from any other discourse. In other words what makes science, science, as opposed to religion? What is specific to the method of science? Our simplest response to this question is that science deals with facts that are objective (out there in some way) and that religion has to do with belief and is subjective. We might want to say, then, that science is true, and religion is not. When we looked at this simple definition, however, the less certain and clear it seemed. For the idea that science is made up of many observations of facts that are then converted into theories breaks down in the problem of induction, which, in its most succinct form, is the impossibility of leaping from a singular judgement to a universal one. No amount of logical finessing will get you from a particular to a universal judgement. This would seem to imply that science is no more objective than religion, and that a theory is as much a belief as any faith, though a particularly strong one when it comes to habits of the human mind.[1]

Moreover, it was also clear that the ‘inductionist’ picture of science was not accurate at all, since facts are not just littered throughout the world such that we pick them up and notice common characteristics from which we then construct some universal law, as though coming up with a scientific theory were no different from observing black crows and coming up with the statement ‘All crows are black’. On the contrary, we already come to facts with a pre-existing theory, which determines which facts we take as relevant or not (or even which fact we can see). As Ladyman explained, Newton did not find the law of gravity in Kepler’s data, he already had to have it in order to interpret the data (Ladyman 2002, pp.55–6).

This reversal of the relation between theory and facts, that theory is first and facts second, is the basis of the next philosophy of science that we shall look at, Popper’s theory of falsification, and indeed rose out of the insurmountable problems of ‘inductionism’. His argument is that we should give up induction as the basis of science, but such a rejection would not lead to irrationalism. Rather we substitute for induction, deduction. But did we not argue already in the first lecture that deduction could not be the basis of science, since deduction is merely tautological? Deductive logic tells us nothing new about the world, but only analyses what we already know, whereas science actually tells us something about nature that we didn’t know before.

Deduction does not work as a basis of science only if we move from the singular to the universal, but if we go from the universal back to the singular then deduction does work. Indeed, this move from the universal back to the singular is exactly, Popper argues, how science operates. We do not start with facts and then make laws, rather we start with laws and then we attempt to test them with facts. The logical point is that we can’t go from observations to theories, even if the observations themselves are true, but it is possible the other way around. We can go from theories then back to observational statements to show that the theory is false. Thus to use Chalmers example, if someone was to see a white raven outside the lecture room today, then this would prove deductively that the statement ‘All Ravens are black’ is false. Such deductive arguments are known as modus tollens, which take the form if P, then Q. ⌐Q, therefore ⌐P (Chalmers 1999, p.61).

When we look at the history of science, this seems exactly what happens. Take for the example, Eddington’s proof of Einstein’s theory that gravity bends light. If the theory was correct then a star that was beyond the sun should be displaced from the direction of the observer so that we could see it. Normally the light from the sun would mean that these starts would not be visible to us, but would be if the light of the sun was blocked. Eddington managed to measure just such a displacement with the eclipse of the sun in 1919. For Popper, the point of this story is that he could have proved otherwise. In other words, Einstein’s theory could have been falsified, if there had not been any displacement.

The real difference between science and religion or any other discourse is not the theories or hypotheses put forward, but how they are tested. Popper is adamant that science is creative as any other human discourse and that the origin of this creativity is outside any logical explanation. That someone comes up with such an idea at such a time cannot be logically deduced. Thus, it would require a complex explanation, comprising of psychology, history, and sociology, for example, of how Galileo or Einstein came up with their ideas, and why not someone else, or at different time and place, but what we do know that what makes these creations scientific, as opposed to anything else is that they can be falsified (this is the difference between context of discovery and context of justification). In the opposite case, it does not seem possible to falsify a religion logically. I can always find a reason to believe something. Think for example of the classic problem of evil in theology. How do I justify the existence of God with evil in the world? It is perfectly possible to find such a reason, as Leibniz did, that this is the ‘best of all possible worlds’, and it is just our lack of human understanding that prevents us from seeing it so.

Here we might need to know a little of the story of Popper’s life. When he was young he was a communist and of course Marxism was treated as a science. He says that one day in went on a march with his friends and they were attacked by the police and some of them were killed. He was so shaken by this incident that he had to speak about to his political leaders. They told him that these deaths were necessary for the political emancipation of the workers as was explained by scientific Marxism. But what then would falsify Marxism, for they did not seem to be any instance, including the death of his friends that could not be explained by it.[2]

This is precisely the difference between a science and a pseudo-science (religion is only a pseudo-science when it takes itself to be answering scientific questions, otherwise it is perfectly meaningful for Popper): a pseudo-science has the answer to everything and can never not be true, whereas a science does not have the answer to everything and can always be false. It is this that demarcates, to use Popper’s word, empirical science, from anything else, and it is a question of method, rather than logical form, by which he means the positivist obsession with the correlation of statements with aspects of reality. Metaphysics and religion are only pseudo sciences when they pretend to be sciences. If they do not, then there is nothing intrinsically wrong with them. They are certainly not meaningless which is just derogatory word, rather than having any useful philosophical sense.

If what makes a scientific theory scientific is falsification, what exactly makes a falsification? Can any falsification be scientific? Such a broad generalisation does not seem to be correct because just to falsify something would not make it a scientific theory. I could falsify physics, by quoting Genesis but no one would think I was being scientific. The answer here is intersubjective testability. One cannot conceive of how it would be possible to set up an experiment that would test my falsification of physics that claimed God had created the universe in the way that it is described in Genesis. One can imagine, however how it might be possible to test the falsification of Newtonian science through the prediction made by Einstein, which is entirely what the example from Eddington proves, and it is perfectly possible that other scientists could conceive of such an experiment, whether in principal or in practice.[3]

Could a theory always secure itself by simple adding an ad hoc modification every time a falsification was produced? Thus, to use Chalmers’s example, we could take the generalisation that all bread was nutritious to be falsified by the death of all the members of French village who ate bread. We could then qualify our theory by saying that all bread is nutritious except when it is eaten by these members of the French village and we could do this every time any falsification was discovered. Such ad hoc modification would destroy any progress in scientific discovery. How then can we distinguish between an authentic and inauthentic ad hoc modification (Chalmers 1999, p.75). In this example, the modification cannot be falsified, so it does not tell us anything new about the world. It in fact tells us less than the original theory that all bread nourishes. So, an authentic modification must be one that is also falsifiable. If we had said instead that all bread nourishes except one that is contaminated by certain fungus called Claviceps purpurea, then this would be an authentic ad hoc modification, since it could be tested and falsified, and thus does tell us something new about the world.

This distinction between authentic and inauthentic ad hoc modifications of scientific theories, however, tells us that we should not over-estimate falsifications of theories. When we look at the history of science we can see that ad hoc modifications can confirm rather than deny a theory. Take the case of the discovery of Neptune. Irregularities in the orbit of Uranus predicted that there must be another planet that had not be observed. Rather than reject Newton’s theory, scientists argued that a planet must exist that would explain it. Thus, the fact that Neptune was found in 1846 confirmed Newton’s theory rather than falsified it. Rather than seeing science as just a series of falsifications, which lead from theory to the next, Aristotle to Newton to Einstein, we should see it as the confirmation of bold conjectures and the falsification of cautious ones. For what difference does it make to science if one falsifies conjectures such as the universe is made of porridge or confirms a cautious one? But how then do we determine what makes a bold conjecture? The only answer to this must be background theories themselves, for only in relation to them could we know what would be bold or timid. The background knowledge is therefore the cautious conjecture (what we take to be correct) and the bold conjecture flies in the face of what everyone thinks is the case. We can see, then, what the real fundamental difference between the falsificationist and inductionist is. The first takes the history of science seriously, and the second has no conception of the history of science at all. There is no background knowledge. Rather facts are accumulated as though there were no context at all and science existed in the eternal present.

Is falsification immune to criticism then? The answer must be unfortunately not. The real problem is still the relation to the theory and the observation. All we can say deductively is that if there is O, then the falsity of T follows if the O is not given, but it tells us nothing about the standard of the evidence itself. What if the evidence is incorrect? Perhaps when person who said that the raven was white and no idea what white was. Perhaps the photograph of the white raven was created in Photoshop, and no such evidence exists. Popper does not have a better story about the correctness of evidence than the positivist.

Moreover, when we look at science, it does not take the simple form of ‘All swans are White’. Rather, sciences are made up of complex collection of universal statements which are interrelated to one another. Now if a prediction tells us the theory is false it tells is that one of the premises might be wrong but not which one or even that our own experience might be the problem. It might not the theory that is out, but the ‘test situation’ itself, because we cannot isolate the premise which allows us to falsify the theory (this is known as the Duhem/Quine thesis). So to use Ladyman’s example, if we were to try and predict the path of a comet, the law of gravity would not be sufficient, so if the predication were incorrect we would not know that it was the theory of gravity that was being falsified or something else (Ladyman 2002, pp.77–8). You cannot isolate hypotheses in science and refute one and others, because these different hypotheses fit together to make a whole.

Even if such an isolation were possible, falsification does not seem to capture what scientists do, for when we look at the history of science we do not find one great conjecture following another, but that scientists stick to their theories despite the fact that they can be falsified or they adopt a new hypothesis even though all the known evidence at the time should have killed them off at birth. This is what we find when we look at the detail of the eventual transition from the Aristotelian to the Copernican view of the world as Feyerabend and Kuhn describe it. It certainly was not the simple falsification of the one by the other. Science works, to some extent, because scientists are dogmatic and not open to falsification. If that is the case, how is it possible to differentiate, or demarcate, science from any other dogma? Will we not have to use different criteria, which are more pragmatic and historical than methodological?

Works Cited

Chalmers, A.F., 1999. What is this Thing Called Science?, St. Lucia, Qld.: University of Queensland.

Ladyman, J., 2002. Understanding Philosophy of Science, London; New York: Routledge.

Popper, K.R., 2002. Unended quest, London; New York: Routledge.


[1] When we look at science as a method this is a problem. We might ask, however, if we think of science as an activity, whether it is such a problem since science is not a straightforward process of induction.

[2] The source of this story can be found in Popper’s autobiography (Popper 2002, pp.30–8).

[3] Does this open Popper to a more pragmatic account of science than an epistemological one? For if testability is inter-subjective how are we to describe it? Popper appears to want to separate questions of method from question of practice, but later criticisms will in turn want to question this distinction by asking whether it is really the case, when we look at the history of science, that scientists really are committed to the principle of falsifiability. This will be part of Kuhn’s critique of Popper.

Advertisements

Popper and Falsification – Lecture 4

October 28, 2015

Karl_PopperWhat we want is some criterion which will allow us to distinguish science from any other discourse. In other words what makes science, science, as opposed to religion? What is specific to the method of science? Our simplest response to this question is that science deals with facts that are objective (out there in some way) and that religion has to do with belief and is subjective. We might want to say, then, that science is true, and religion is not. When we looked at this simple definition, however, the less certain and clear it seemed. For the idea that science is made up of many observations of facts that are then converted into theories breaks down in the problem of induction, which, in its most succinct form, is the impossibility of leaping from a singular judgement to a universal one. No amount of logical finessing will get you from a particular to a universal. This would seem to imply that science is no more objective than religion, and that a theory is as much a belief as any faith.[1]

Moreover, it was also clear that the inductionist picture of science was not accurate at all, since facts are not just littered throughout the world such that we pick them up and notice common characteristics from which we then construct some universal law. On the contrary, we already come to facts with a pre-existing theory, which determines which facts we take as relevant or not (or even which fact we can see). As Ladyman explained, Newton did not find the law of gravity in Kepler’s data, he already had to have it in order to interpret the data (Ladyman 2002, pp.55–6).

This reversal of the relation between theory and facts, that theory is first and facts second, is the basis of the next philosophy of science that we shall look at, Popper’s theory of falsification, and indeed rose out of the insurmountable problems of ‘inductionism’. His argument is that we should give up induction as the basis of science, but such a rejection would not lead to irrationalism. Rather we substitute for induction, deduction. But did we not argue already in the first lecture that deduction could not be the basis of science, since deduction is merely tautological? Deductive logic tells us nothing new about the world, but only analyses what we already know, whereas as would say that science actually tells us something about nature that we didn’t know before.

Deduction does not work as a basis of science only if we move from the singular to the universal, but if we go from the universal back to the singular then deduction does work. Indeed, this move from the universal back to the singular is exactly, Popper argues, how science operates. We do not start with facts and then make laws, rather we start with laws and then we attempt to test them with facts. The logical point is that we can’t go from observations to theories, even if the observations themselves are true, but it is possible the other way around. We can go from theories then back to observational statements to show that the theory is false. Thus to use Chalmers example, if someone was to see a white raven outside the lecture room today, then this would prove deductively that the statement ‘All Ravens are black’ is false. Such deductive arguments are known as modus tollens, which take the form if P, then Q. ⌐Q, therefore ⌐P (Chalmers 1999, p.61).

When we look at the history of science, this seems exactly what happens. Take for the example, Eddington’s proof of Einstein’s theory that gravity bends light. If the theory was correct then a star that was beyond the sun should be displaced from the direction of the observer so that we could see it. Normally the light from the sun would mean that these starts would not be visible to us, but would be if the light of the sun was blocked. Eddington managed to measure just such a displacement with the eclipse of the sun in 1919. For Popper, the point of this story is that he could have proved otherwise. In other words, Einstein’s theory could have been falsified, if there had not been any displacement.

The real difference between science and religion or any other discourse is not the theories or hypotheses put forward, but how they are tested. Popper is adamant that science is creative as any other human discourse and that the origin of this creativity is outside any logical explanation. That someone comes up with such an idea at such a time cannot be rationally explained. Thus we don’t know how Galileo or Einstein came up with their ideas, and why not someone else, or at different time and place, but what we do know that what makes these creations scientific, as opposed to anything else is that they can be falsified (this is the difference between context of discovery and context of justification). In the opposite case, it does not seem possible to falsify a religion logically. I can always find a reason to believe something. Think for example of the classic problem of evil in theology. How do I justify the existence of God with evil in the world? It is perfectly possible to find such a reason, as Leibniz did that this is the ‘best of all possible worlds’, and it is just our lack of human understanding that prevents us from seeing it so.

Here we might need to know a little of the story of Popper’s life. When he was young he was a communist and of course Marxism was treated as a science. He says that one day in went on a march with his friends and they were attacked by the police and some of them were killed. He was so shaken by this incident that he had to speak about to his political leaders. They told him that these deaths were necessary for the political emancipation of the workers as was explained by scientific Marxism. But what then would falsify Marxism, for they did not seem to be any instance, including the death of his friends that could not be explained by it.[2]

This is precisely the difference between a science and a pseudo-science (religion is only a pseudo-science when it takes itself to be answering scientific questions, otherwise it is perfectly meaningful for Popper): a pseudo-science has the answer to everything and can never not be true, whereas a science does not have the answer to everything and can always be false. It is this that demarcates, to use Popper’s word, empirical science, from anything else and it is a question of method, rather than logical form, by which he means the positivist obsession with the correlation of statements with aspects of reality. Metaphysics and religion are only pseudo sciences when they pretend to be sciences. If they do not, then there is nothing intrinsically wrong with them. They are certainly not meaningless which is just derogatory word, rather than having any useful philosophical sense.

If what makes a scientific theory scientific is falsification, what exactly makes a falsification? Can any falsification be scientific? Such a broad generalisation does not seem to be correct because just to falsify something would not make it a scientific theory. I could falsify physics, by quoting Genesis but no one would think I was being scientific. The answer here is intersubjective testability. One cannot conceive of how it would be possible to set up an experiment that would test my falsification of physics that claimed God had created the universe in the way that it is described in Genesis. One can imagine, however how it might be possible to test the falsification of Newtonian science through the prediction made by Einstein, which is entirely what the example from Eddington proves, and it is perfectly possible that other scientists could conceive of such an experiment, whether in principal or in practice.[3]

Could a theory always secure itself by simple adding an ad hoc modification every time a falsification was produced? Thus, to use Chalmers’s example, we could take the generalisation that all bread was nutritious to be falsified by the death all the members of French village who ate bread. We could then qualify our theory by saying that all bread is nutritious except when it is eaten by these members of the French village and we could do this every time any falsification was discovered. Such ad hoc modification would completely destroy any progress in scientific discovery. How then can we distinguish between an authentic and inauthentic ad hoc modification (Chalmers 1999, p.75). In this example, the modification cannot be falsified, so it does not tell us anything new about the world. It in fact tells us less than the original theory that all bread nourishes. So an authentic modification must be one that is also falsifiable. If we had said instead that all bread nourishes except one that is contaminated by certain fungus called Claviceps purpurea, then this would be an authentic ad hoc modification, since it could be tested and falsified, and thus does tell us something new about the world.

This distinction between authentic and inauthentic ad hoc modifications of scientific theories, however, tells us that we should not over-estimate falsifications of theories. When we look at the history of science we can see that ad hoc modifications can confirm rather than deny a theory. Take the case of the discovery of Neptune. Irregularities in the orbit of Uranus predicted that there must be another planet that had not be observed. Rather than reject Newton’s theory, scientists argued that a planet must exist that would explain it. Thus, the fact that Neptune was found in 1846 confirmed Newton’s theory rather than falsified it. Rather than seeing science as just a series of falsifications which lead from theory to the next, Aristotelianism to Newtonism to Einstein, we should see it as the confirmation of bold conjectures and the falsification of cautious ones. For what difference does it make to science if one falsifies conjectures such as the universe is made of porridge or confirms a cautious one? But how then do we determine what makes a bold conjecture? The only answer to this must be background theories themselves, for only in relation to them could we know what would be bold or timid. The background knowledge is therefore the cautious conjecture (what we take to be correct) and the bold conjecture flies in the face of what everyone thinks is the case. We can see, then, what the real fundamental difference between the falsificationist and inductionist is. The first takes the history of science seriously, and the second has no conception of the history of science at all. There is no background knowledge. Rather facts are accumulated as though there were no context at all and science existed in the eternal present.

Is falsification immune to criticism then? The answer must be unfortunately not. The real problem is still the relation to the theory and the observation. All we can say deductively is that if there is O, then the falsity of T follows if the O is not given, but it tells us nothing about the standard of the evidence itself. What if the evidence is incorrect? Perhaps when person who said that the raven was white and no idea what white was. Perhaps the photograph of the white raven was created in Photoshop, and no such evidence exists. Popper does not have a better story about the correctness of evidence than the positivist.

Moreover, when we actually look at science, it does not take the simple form of ‘All swans are White’…. Rather, sciences are made up of complex collection of universal statements which are interrelated to one another. Now if a prediction tells us the theory is false it tells is that one of the premises might be wrong but not which one or even that our own experience might be the problem. It might not the theory that is out, but the ‘test situation’ itself, because we cannot isolate the premise which allows us to falsify the theory (this is known as the Duhem/Quine thesis). So to use Ladyman’s example, if we were to try and predict the path of a comet, the law of gravity would not be sufficient, so if the predication were incorrect we would not know that it was the theory of gravity that was being falsified or something else (Ladyman 2002, pp.77–8).

Even if such an isolation were possible, falsification does not seem to capture actually what science and scientists do, for when we look at the history of science we do not find one great conjecture following another, but that scientists stick to their theories despite the fact that they can be falsified or they adopt a new hypothesis even though all the known evidence at the time should have killed them off at birth. This is what we find when we look at the detail of the eventual transition from the Aristotelian to the Copernican view of the world as Feyerabend and Kuhn describe it. It is certainly was not the simple falsification of the one by the other. Science works, to some extent, because scientists are dogmatic and not open to falsification. If that is the case, how is it possible to differentiate, or demarcate, science from any other dogma? Will we not have to use different criteria?

Works Cited

Chalmers, A.F., 1999. What is this Thing Called Science?, St. Lucia, Qld.: University of Queensland.

Ladyman, J., 2002. Understanding Philosophy of Science, London; New York: Routledge.

Popper, K.R., 2002. Unended quest, London; New York: Routledge.


[1] When we look at science as a method this is a problem. We might ask, however, if we think of science as an activity, whether it is such a problem.

[2] The source of this story can be found in Popper’s autobiography (Popper 2002, pp.30–8).

[3] Does this open Popper to a more pragmatic account of science than an epistemological one? For if testability is inter-subjective how are we to describe it? Popper appears to want to separate questions of method from question of practice, but later criticisms will in turn want to question this distinction by asking whether it is really the case, when we look at the history of science, that scientists really are committed to the principle of falsifiability. This will be part of Kuhn’s critique of Popper.


Kuhn – Lecture 5

December 13, 2014

JastrowDuckRabbitScience does not begin with facts and then construct theories out of them. Nor does science begin with theories and then just find facts that would confirm them. Both these conceptions conceive of science as though it were a discourse that was completely context free. In the first case, facts are simply available as though they were waiting for interpretation of a specific kind, and in the second case, theories are simply open to facts as though there were no inertia or hindrance to the smooth progress of science from one theory to the next, each equally open to the possibility of falsification.

One of the first philosophers of science in the Anglo-American tradition to take the idea of context or background to scientific activity seriously was Thomas Kuhn.[1] Loosely characterised this approach might be called ‘historical’. What does it mean to treat science as though it were part of history rather than outside of it? It means first of all to take scientists seriously. It is to treat what they do the same way that we would analyse the thoughts and actions of French peasants or the 13th century or a military general in the 20th. First of all to record scientific achievements correctly (who thought of what at what time), and secondly to examine exactly how scientists come up with their theories in relation to the material they were investigating. What it certainly is not is the importation of philosophical theories from the outside (like verification or falsification) followed by squeezing the scientific activity to see whether it would fit these ideal models.

However much the logical positivists and Popper might differ, they both have the same idealised view of science: there is a sharp difference between theory and observation; knowledge is cumulative tending towards a true understanding of the universe; the language of science is precise and science is unified; science is either inductive or deductive; the key question of the philosophy of science is legitimacy and validity, rather than the contingency of discovery. Against all the suppositions Kuhn puts forward exactly the opposite: there is no sharp difference between observation and theory; knowledge is discontinuous; science is neither a tight deductive structure nor an inductive reading of facts; scientific concepts are not precise; nor is science unified; context is important and science is historical and temporal.

At the heart of the idealised picture of science is scientific progress. This is the view that science is leading to ever increasing knowledge about the universe and that finally one day we will have a theory of everything, and I suppose, science can come to end, because there will be no more questions that need to be answered. So first of all there are pre-scientific theories of the universe that we find in the religious and mythical texts (like Genesis), and then we get the first science, Aristotelianism (though this is a really a mixture of science and occult explanations), then Newtonism (which is the first science proper) and then finally in our times, Einsteinian science which is a response to the crisis that befell Newtonism. One imagines that sometime in the future, though one can never tell, there will be fourth science that will replace Einstein, but only because it contains more truth and is close to the universe as it really is than all the other theories that we have had. Such a view of the history of science, we might call ‘convergence’, since it views the series of scientific discoveries of converging on the true understanding of reality.

There are two problems with this image of science. One is temporal and the practical. First of all it has a conception of time, where the past is merely a stepping stone to the present but the past has no meaning in itself. For how can we measure the progress of science in this sense unless we imagine an end towards which it is moving and this end is supposed to be an advance on the past?[2] But how can we know that this advance is real unless we can stand outside of time and measure it? Is it not really the case that past is not the stepping stone to the future, but that we judge the past from the vantage point of the present, and in looking back, project a false teleology back into the past? In terms of the past itself, there were numerous possibilities and the present that we now occupy did not have to occur. Equally the present that we now stand in has infinite possibilities, so we cannot know what the future will be.

In terms of the practice of science, we also know that his temporal picture of progress is false. This is what Kuhn discovered when he did his own historical research. Rather than the history of science demonstrating that each scientific period progressed into the next one moving to ever greater level of truth and closing the distance between discourse and reality, we find that it is discontinuous and non-cumulative and that there is no reality out there, which we could know independently and through which we could measure the relative truths of each discourse because reality itself is a creation of discourse and not its external validation.

What does it mean to say that the history of science is discontinuous rather than continuous, non-cumulative rather than cumulative? Let’s go back to the image of progress where science moves smoothly from Aristotle, to Newton, to Einstein. What is left out in this description is the gaps or spaces between each scientific theory (or what Kuhn calls a paradigm, because it is more than just a theory) and it can leave this gaps out because the fantasy of some ultimate truth which is where reality and discourse are the same. As soon as we leave this fantasy behind, and realise that it too is a creation of a discourse (in this case metaphysics), then we can see that there is no transition of one to the other. Rather, they are separate or incommensurable. They belong to different worlds.

Again this is visible when we actually study the history of science, rather than project our own view of progress upon it. What we get instead of single continuous line is line of breaks: Aristotle, Newton and Einstein. What then causes these breaks? Why don’t we just go from one science to another in an endless progression towards the truth? The answer for Kuhn is to be found in history and not in the philosophical image of science as a universal method.

The new picture we have of science is now as follows: first we have pre-science – normal science – crisis or revolution – new normal science – new crisis (Chalmers 1999, p.108). When at first science begins to emerge we don’t have a collection of facts or theories that explain facts, rather we have a competition between many theories (Chalmers gives the example of the state of optics before Newton). Gradually different scientists will be attracted to the one explanation. What is important is that the reason for this attraction will not just be scientific or rarely just scientific. It will be a combination of difference elements some of which will be psychological, sociological and even metaphysical. As more and more scientists come on board, what is in a state of chaos will coagulate into a paradigm. Only at that point will normal science be possible (the kind of science that Popper and the logical positivists describe). But even a paradigm, which makes normal science possible, is not made up of merely theories and observations. Like Newtonian mechanics, it is constructed from fundamental laws and theoretical assumptions, standard techniques and methods of investigation, general rules about exceptions and application to reality and most importantly of all a kind of world view or metaphysics which will unify all of this together (in Newtonism, that we exist in an infinite deterministic universe).

Rather than anomalies, as Popper would have us believe, being antithetical to normal science, it can quite happily accept them as long as they don’t attack the fundamentals of the paradigm. Everyone can get happily to work devising their experiments and putting in their grants and anyone who goes against the status quo can be banished to the outer darkness. The paradigm is reinforced by the institutions themselves. If you don’t follow the paradigm you won’t get the grant money, and anyway the education of young scientists make sure that they follow the paradigm. This is clearly what Kuhn saw when he first looked into the history of science as a practicing scientist: young scientists were taught the idealised image of science that had nothing at all to do with the history of science at all.

So why do paradigms fall? Why are revolutions inevitable? This is because of the anomalies. Because no discourse can close the gap between itself and reality, there will always be the nagging doubt that something is not being explained by the paradigm. As more and more money and experiments are thrown at these anomalies, cracks begin to appear in the scientific establishment. Thus a normal science begins to take the form of the pre-science. Rather than scientists doing experiments, they start having ideas and hypothesis. Some might be said to be cranks and fools, but gradually they begin to attract other scientists. Again Kuhn is clear that the reason for this cannot be scientific or logical, because there is nothing in one paradigm that would justify the leap to another, for there is no commensurability that would link them together, such that one might say that one is truer than the other. The reasons are practical. As more and more are attracted to this new science, gradually a new paradigm is born and the whole process repeats itself. We get a new normal science, where again people can happily devise their experiments, apply for grants and get promotion. Until of course the cracks start appearing again.

Although this appears to be an accurate representation of what scientists do, there is a fundamental problem with it. If we are to give up the image of science as the progress towards a truth in which the distance between discourse and reality is progressive closed, for a discontinuous series of closed paradigms, then does this make scientific truth relative? We can distinguish normal science from pseudo-science because of how paradigms work (the difference between astronomy and astrology), but that does not make science itself any truer. Can we say that Einstein, for example is truer than Newton? We want to feel that this is the case, but Kuhn’s principle of incommensurability will not let us do so. The answer to this question, as we shall see when we read Kuhn’s The Structure of Scientific Revolutions in more detail, is that we might have to change what we mean by truth, rather than giving up truth altogether. It means that we have to think of truth as a practice or activity, rather than as a representation of a reality that stands outside of us waiting for our discourse to catch up with it.

Works Cited

Chalmers, A.F., 1999. What is this Thing Called Science?, St. Lucia, Qld.: University of Queensland.

Sharrock, W.W. & Read, R.J., 2002. Kuhn : Philosopher of Scientific Revolution, Cambridge: Polity.


[1] He might have been the first American philosopher to take this idea seriously. In France, this was the dominant view of science (Sharrock & Read 2002, p.1).

[2] It is science (think for example of evolution) itself that should make us suspect such teleological arguments.


Falsification – Lecture 4

November 30, 2014

Karl_PopperWhat we want is some criterion which will allow us to distinguish science from any other discourse. In other words what makes science, science, as opposed to religion? What is specific to the method of science? Our simplest response to this question is that science deals with facts that are objective (out there in some way) and that religion has to do with belief and is subjective. We might want to say, then, that science is true, and religion is not. When we looked at this simple definition, however, the less certain and clear it seemed. For the idea that science is made up of many observations of facts that are then converted into theories breaks down in the problem of induction, which, in its most succinct form, is the impossibility of leaping from a singular judgement to a universal one. No amount of logical finessing will get you from a particular to a universal. This would seem to imply that science is no more objective than religion, and that a theory is as much a belief as any faith.[1]

Moreover, it was also clear that the inductionist picture of science was not accurate at all, since facts are not just littered throughout the world such that we pick them up and notice common characteristics from which we then construct some universal law. On the contrary, we already come to facts with a pre-existing theory, which determines which fact we take as relevant or not (or even which fact we can see). As Ladyman explained, Newton did not find the law of gravity in Kepler’s data, he already had to have it in order to interpret the data (Ladyman 2002, pp.55–6).

This reversal of the relation between theory and facts, that theory is first and facts second, is the basis of the next philosophy of science that we shall look at, Popper’s theory of falsification, and indeed rose out of the insurmountable problems of ‘inductionism’. His argument is that we should give up induction as the basis of science, but such a rejection would not lead to irrationalism. Rather we substitute for induction, deduction. But did we not argue already in last lecture that deduction could not be the basis of science, since deduction is merely tautological? Deductive logic tells us nothing new about the world, but only analyses what we already know, whereas as would say that science actually tells us something about nature that we didn’t know before.

Deduction does not work as a basis of science only if we move from the singular to the universal, but if we go from the universal back to the singular then deduction does work. Indeed, this move from the universal back to the singular is exactly, Popper argues, how science operates. We do not start with facts and then make laws, rather we start with laws and then we attempt to test them with facts. The logical point is that we can’t go from observations to theories, even if the observations themselves are true, but it is possible the other way around. We can go from theories then back to observational statements to show that the theory is false. Thus to use Chalmers example, if someone was to see a white raven outside the lecture room today, then this would prove deductively that the statement ‘All Ravens are black’ is false. Such deductive arguments are known as modus tollens, which take the form if P, then Q. ⌐Q, therefore ⌐P (Chalmers 1999, p.61).

When we look at the history of science, this seems exactly what happens. Take for the example, Eddington’s proof of Einstein’s theory that gravity bends light. If the theory was correct then a star that was beyond the sun should be displaced from the direction of the observer so that we could see it. Normally the light from the sun would mean that these starts would not be visible to us, but would be if the light of the sun was blocked. Eddington managed to measure just such a displacement with the eclipse of the sun in 1919. For Popper, the point of this story is that he could have proved otherwise. In other words, Einstein’s theory could have been falsified, if there had not been any displacement.

The real difference between science and religion or any other discourse is not the theories or hypotheses that they put forward, but how they test them. Popper is adamant that science is creative as any other human discourse and that the origin of this creativity is outside any logical explanation. That someone comes up with such an idea at such a time cannot be rationally explained. Thus we don’t know how Galileo or Einstein came up with their ideas, and why not someone else, or at different time and place, but what we do know that what makes these creations scientific, as opposed to anything else is that they can be falsified (this is the difference between context of discovery and context of justification). In the opposite case, it does not seem possible to falsify a religion logically. I can always find a reason to believe something. Think for example of the classic problem of evil in theology. How do I justify the existence of God with evil in the world? It is perfectly possible to find such a reason, as Leibniz did that this is the ‘best of all possible worlds’, and it is just our lack of human understanding that prevents us from seeing it so.

Here we might need to know a little of the story of Popper’s life. When he was young he was a communist and of course Marxism was treated as a science. He says that one day in went on a march with his friends and they were attacked by the police and some of them were killed. He was so shaken by this incident that he had to speak about to his political leaders. They told him that these deaths were necessary for the political emancipation of the workers as was explained by scientific Marxism. But what then would falsify Marxism, for they did not seem to be any instance, including the death of his friends that could not be explained by it.[2]

This is precisely the difference between a science and a pseudo-science (religion is only a pseudo-science when it takes itself to be answering scientific questions, otherwise it is perfectly meaningful for Popper): a pseudo-science has the answer to everything and can never not be true, whereas a science does not have the answer to everything and can always be false. It is this that demarcates, to use Popper’s word, empirical science, from anything else and it is a question of method, rather than logical form, by which he means the positivist obsession with the correlation of statements with aspects of reality. Metaphysics and religion are only pseudo sciences when they pretend to be sciences. If they do not, then there is nothing intrinsically wrong with them. They are certainly not meaningless which is just derogatory word, rather than having any useful philosophical sense.

If what makes a scientific theory scientific is falsification, what exactly makes a falsification? Can any falsification be scientific? Such a broad generalisation does not seem to be correct because just to falsify something would not make it a scientific theory. I could falsify physics, by quoting Genesis but no one would think that I was being scientific. The answer here is intersubjective testability. One cannot conceive of how it would be possible to set up an experiment that would test my falsification of physics that claimed God had created the universe in the way that it is described in Genesis. One can imagine, however how it might be possible to test the falsification of Newtonian science through the prediction made by Einstein, which is entirely what the example from Eddington proves, and it is perfectly possible that other scientists could conceive of such an experiment, whether in principal or in practice.[3]

Could a theory always secure itself by simple adding an ad hoc modification every time a falsification was produced? Thus, to use Chalmers’s example, we could take the generalisation that all bread was nutritious to be falsified by the death all the members of French village who ate bread. We could then qualify our theory by saying that all bread is nutritious except when it is eaten by these members of the French village and we could do this every time any falsification was discovered. Such ad hoc modification would completely destroy any progress in scientific discovery. How then can we distinguish between an authentic and inauthentic ad hoc modification (Chalmers 1999, p.75). In this example, the modification cannot be falsified, so it does not tell us anything new about the world. It in fact tells us less than the original theory that all bread nourishes. So an authentic modification must be one that is also falsifiable. If we had said instead that all bread nourishes except one that is contaminated by certain fungus called Claviceps purpurea, then this would be an authentic ad hoc modification, since it could be tested and falsified, and thus does tell us something new about the world.

This distinction between authentic and inauthentic ad hoc modifications of scientific theories, however, tells us that we should not over-estimate falsifications of theories. When we look at the history of science we can see that ad hoc modifications can confirm rather than deny a theory. Take the case of the discovery of Neptune. Irregularities in the orbit of Uranus predicted that there must be another planet that had not be observed. Rather than reject Newton’s theory, scientists argued that a planet must exist that would explain it. Thus, the fact that Neptune was found in 1846 confirmed Newton’s theory rather than falsified it. Rather than seeing science as just a series of falsifications which lead from theory to the next, Aristotelianism to Newtonism to Einstein, we should see it as the confirmation of bold conjectures and the falsification of cautious ones. For what difference does it make to science if one falsifies conjectures such as the universe is made of porridge or confirms a cautious one? But how then do we determine what make a bold conjecture? The only answer to this must be background theories themselves, for only in relation to them could we know what would be bold or timid. The background knowledge is therefore the cautious conjecture (what we take to be correct) and the bold conjecture flies in the face of what everyone thinks is the case. We can see, then, what the real fundamental difference between the falsificationist and inductionist is. The first takes the history of science seriously, and the second has no conception of the history of science at all. There is no background knowledge. Rather facts are accumulated as though there were no context at all and science existed in the eternal present.

Is falsification immune to criticism then? The answer must be unfortunately not. The real problem is still the relation to the theory and the observation. All we can say deductively is that if there is O, then the falsity of T follows if the O is not given, but it tells us nothing about the standard of the evidence itself. What if the evidence is incorrect? Perhaps when person who said that the raven was white and no idea what white was. Perhaps the photograph of the white raven was created in Photoshop, and no such evidence exists.  Popper does not have a better story about the correctness of evidence than the positivist.

Moreover, when we actually look at science, it does not take the simple form of ‘All swans are White’…. Rather, sciences are made up of complex collection of universal statements which are interrelated to one another.  Now if a prediction tells us the theory is false it tells is that one of the premises might be wrong but not which one or even that our own experience might be the problem.  It might not the theory that is out, but the ‘test situation’ itself, because we cannot isolate the premise which allows us to falsify the theory (this is known as the Duhem/Quine thesis). So to use Ladyman’s example, if we were to try and predict the path of a comet, the law of gravity would not be sufficient, so if the predication were incorrect we would not know that it was the theory of gravity that was being falsified or something else (Ladyman 2002, pp.77–8).

Even if such an isolation were possible, falsification does not seem to capture actually what science and scientists do, for when we look at the history of science we do not find one great conjecture following another, but that scientists stick to their theories despite the fact that they can be falsified or they adopt a new hypothesis even though all the known evidence at the time should have killed them off at birth. This is what we find when we look at the detail of the eventual transition from the Aristotelian to the Copernican view of the world as Feyerabend and Kuhn describe it. It is certainly was not the simple falsification of the one by the other. Science works, to some extent, because scientists are dogmatic and not open to falsification. If that is the case, how is it possible to differentiate, or demarcate, science from any other dogma? Will we not have to use different criteria?

Works Cited

Chalmers, A.F., 1999. What is this Thing Called Science?, St. Lucia, Qld.: University of Queensland.

Ladyman, J., 2002. Understanding Philosophy of Science, London; New York: Routledge.

Popper, K.R., 2002. Unended quest, London; New York: Routledge.

 

[1] When we look at science as a method this is a problem. We might ask, however, if we think of science as an activity, whether it is such a problem.

[2] The source of this story can be found in Popper’s autobiography (Popper 2002, pp.30–8).

[3] Does this open Popper to a more pragmatic account of science than an epistemological one? For if testability is inter-subjective how are we to describe it? Popper appears to want to separate questions of method from question of practice, but later criticisms will in turn want to question this distinction by asking whether it is really the case, when we look at the history of science, that scientist really are committed to the principle of falsifiability.


Falsification

November 3, 2013

What we want is some criterion which will allow us to distinguish science from any other discourse. In other words what makes science, science, as opposed to religion? What is specific to the method of science? Our simplest response to this question is that science deals with facts that are objective (out there in some way) and that religion has to do with belief and is subjective. We might want to say, then, that science is true, and religion is not. When we looked at this simple definition, however, the less certain and clear it seemed. For the idea that science is made up of many observations of facts that are then converted into theories breaks down in the problem of induction, which in its most succinct form, is the impossibility of leaping from a singular judgement to a universal one. No amount of logical finessing will get you from a particular to a universal. This would seem to imply that science is no more objective than religion, and that a theory is as much a belief as any faith.[1] Moreover, it was also clear that the inductionist picture of science was not accurate at all, since facts are not just littered throughout the world such that we pick them up and notice common characteristics from which we then construct some universal law. On the contrary, we already come to facts with a pre-existing theory, which determines which fact we take as relevant or not (or even which fact we can see). As Ladyman explained, Newton did not find the law of gravity in Kepler’s data, he already had to have it in order to interpret the data (Ladyman 2002, pp.55–6).

This reversal of the relation between theory and facts, that theory is first and facts second, is the basis of the next philosophy of science that we shall look at, Popper’s theory of falsification, and indeed rose out of the insurmountable problems of ‘inductionism’. His argument is that we should give up induction as the basis of science, but such a rejection would not lead to irrationalism. Rather we substitute for induction, deduction. But did we not argue already in last lecture that deduction could not be the basis of science, since deduction is merely tautological? Deductive logic tells us nothing new about the world, but only analyses what we already know, whereas as would say that science actually tells us something about nature that we didn’t know before.

Deduction does not work as a basis of science only if we move from the singular to the universal, but if we go from the universal back to the singular then deduction does work. Indeed, this move from the universal back to the singular is exactly, Popper argues, how science operates. We do not start with facts and then make laws, rather we start with laws and then we attempt to test them with facts. The logical point is that we can’t go from observations to theories, even if the observations themselves are true, but it is possible the other way around. We can go from theories then back to observational statements to show that the theory is false. Thus to use Chalmers example, if someone was to see a white raven outside the lecture room today, then this would prove deductively that the statement ‘All Ravens are black’ is false. Such deductive arguments are known as modus tollens, which take the form if P, then Q. ⌐Q, therefore ⌐P (Chalmers 1999, p.61).

When we look at the history of science, this seems exactly what happens. Take for the example, Eddington’s proof of Einstein’s theory that gravity bends light. If the theory was correct then a star that was beyond the sun should be displaced from the direction of the observer so that we could see it. Normally the light from the sun would mean that these starts would not be visible to us, but would be if the light of the sun was blocked. Eddington managed to measure just such a displacement with the eclipse of the sun in 1919. For Popper, the point of this story is that he could have proved otherwise. In other words, Einstein’s theory could have been falsified, if there had not been any displacement.

The real difference between science and religion, or any other discourse that is not science, is not the theories or hypotheses that they put forward, but how they test them. Popper is adamant that science is creative as any other human discourse and that the origin of this creativity is outside any logical explanation. That someone comes up with such an idea at such a time cannot be rationally explained. Thus we don’t know how Galileo or Einstein came up with their ideas, and why not someone else, or at different time and place, but what we do know that what makes these creations scientific, as opposed to anything else is that they can be falsified (this is the difference between context of discovery and context of justification). In the opposite case, it does not seem possible to falsify a religion logically. I can always find a reason to believe something. Think for example of the classic problem of evil in theology. How do I justify the existence of God with evil in the world? It is perfectly possible to find such a reason, as Leibniz did that this is the ‘best of all possible worlds’, and it is just our lack of human understanding that prevents us from seeing it so.

Here we might need to know a little of the story of Popper’s life. When he was young he was a communist and of course Marxism was treated as a science. He says that one day in went on a march with his friends and they were attacked by the police and some of them were killed. He was so shaken by this incident that he had to speak about to his political leaders. They told him that these deaths were necessary for the political emancipation of the workers as was explained by scientific Marxism. But what then would falsify Marxism, for they did not seem to be any instance, including the death of his friends that could not be explained by it.[2]

This is precisely the difference between a science and a pseudo-science (religion is only a pseudo-science when it takes itself to be answering scientific questions, otherwise it is perfectly meaningful for Popper): a pseudo-science has the answer to everything and can never not be true, whereas a science does not have the answer to everything and can always be false. It is this that demarcates, to use Popper’s word, empirical science, from anything else and it is a question of method, rather than logical form, by which he means the positivist obsession with the correlation of statements with aspects of reality. Metaphysics and religion are only pseudo sciences when they pretend to be sciences. If they do not, then there is nothing intrinsically wrong with them. They are certainly not meaningless which is just derogatory word, rather than having any useful philosophical sense.

If what makes a scientific theory scientific is falsification, what exactly makes a falsification? Can any falsification be scientific? Such a broad generalisation does not seem to be correct because just to falsify something would not make it a scientific theory. I could falsify physics, by quoting Genesis but no one would think that I was being scientific. The answer here is intersubjective testability. One cannot conceive of how it would be possible to set up an experiment that would test my falsification of physics that claimed God had created the universe in the way that it is described in Genesis. One can imagine, however how it might be possible to test the falsification of Newtonian science through the prediction made by Einstein, which is entirely what the example from Eddington proves, and it is perfectly possible that other scientists could conceive of such an experiment, whether in principle or in practice.[3]

Could a theory always secure itself by simple adding an ad hoc modification every time a falsification was produced? Thus, to use Chalmers’s example, we could take the generalisation that all bread was nutritious to be falsified by the death all the members of French village who ate bread. We could then qualify our theory by saying that all bread is nutritious except when it is eaten by these members of the French village and we could do this every time any falsification was discovered. Such ad hoc modification would completely destroy any progress in scientific discovery. How then can we distinguish between an authentic and inauthentic ad hoc modification (Chalmers 1999, p.75). In this example, the modification cannot be falsified, so it does not tell us anything new about the world. It in fact tells us less than the original theory that all bread nourishes. So an authentic modification must be one that is also falsifiable. If we had said instead that all bread nourishes except one that is contaminated by certain fungus called Claviceps purpurea, then this would be an authentic ad hoc modification, since it could be tested and falsified, and thus does tell us something new about the world.

This distinction between authentic and inauthentic ad hoc modifications of scientific theories, however, tells us that we should not overestimate falsifications of theories. When we look at the history of science we can see that ad hoc modifications can confirm rather than deny a theory. Take the case of the discovery of Neptune. Irregularities in the orbit of Uranus predicted that there must be another planet that had not be observed. Rather than reject Newton’s theory, scientists argued that a planet must exist that would explain it. Thus, the fact that Neptune was found in 1846 confirmed Newton’s theory rather than falsified it. Rather than seeing science as just a series of falsifications which lead from theory to the next, Aristotelianism to Newtonism to Einstein, we should see it as the confirmation of bold conjectures and the falsification of cautious ones. For what difference does it make to science if one falsifies conjectures such as the universe is made of porridge or confirms a cautious one? But how then do we determine what make a bold conjecture? The only answer to this must be background theories themselves, for only in relation to them could we know what would be bold or timid. The background knowledge is therefore the cautious conjecture (what we take to be correct) and the bold conjecture flies in the face of what everyone thinks is the case. We can see, then, what the real fundamental difference between the falsificationist and inductionist is. The first takes the history of science seriously, and the second has no conception of the history of science at all. There is no background knowledge. Rather facts are accumulated as though there were no context at all and science existed in the eternal present.

Is falsification immune to criticism then? The answer must be unfortunately not. The real problem is still the relation to the theory and the observation. All we can say deductively is that if there is O, then the falsity of T follows if the O is not given, but it tells us nothing about the standard of the evidence itself. What if the evidence is incorrect? Perhaps when person who said that the raven was white and no idea what white was. Perhaps the photograph of the white raven was created in Photoshop, and no such evidence exists. Popper does not have a better story about the correctness of evidence than the positivist.

Moreover, when we actually look at science, it does not take the simple form of ‘All swans are White’…. Rather, sciences are made up of complex collection of universal statements which are interrelated to one another. Now if a prediction tells us the theory is false it tells is that one of the premises might be wrong but not which one or even that our own experience might be the problem. It might not the theory that is out, but the ‘test situation’ itself, because we cannot isolate the premise which allows us to falsify the theory (this is known as the Duhem/Quine thesis). So to use Ladyman’s example, if we were to try and predict the path of a comet, the law of gravity would not be sufficient, so if the prediction were incorrect we would not know that it was the theory of gravity that was being falsified or something else (Ladyman 2002, pp.77–8).

Even if such an isolation were possible, falsification does not seem to capture actually what science and scientists do, for when we look at the history of science we do not find one great conjecture following another, but that scientists stick to their theories despite the fact that they can be falsified or they adopt a new hypothesis even though all the known evidence at the time should have killed them off at birth. This is what we find when we look at the detail of the eventual transition from the Aristotelian to the Copernican view of the world as Feyerabend and Kuhn describe it. It is certainly was not the simple falsification of the one by the other. Science works, to some extent, because scientists are dogmatic and not open to falsification. If that is the case, how is it possible to differentiate, or demarcate, science from any other dogma? Will we not have to use different criteria?

Works Cited

Chalmers, A.F., 1999. What is this Thing Called Science?, St. Lucia, Qld.: University of Queensland.

Ladyman, J., 2002. Understanding Philosophy of Science, London; New York: Routledge.

Popper, K.R., 2002. Unended quest, London; New York: Routledge.

 


[1] When we look at science as a method this is a problem. We might ask, however, if think of science as an activity, whether it is such a problem.

[2] The source of this story can be found in Popper’s autobiography (Popper 2002, pp.30–8).

[3] Does this open Popper to a more pragmatic account of science than an epistemological one? For if testability is inter-subjective how are we to describe it? Popper appears to want to separate questions of method from question of practice, but later criticisms will in turn want to question this distinction by asking whether it is really the case, when we look at the history of science, that scientist really are committed to the principle of falsifiability.


Falsification – Lecture 4

October 20, 2012

What we want is some criterion which will allow us to distinguish science from any other discourse. In other words what makes science, science, as opposed to religion? What is specific to the method of science? Our simplest response to this question is that science deals with facts that are objective (out there in some way) and that religion has to do with belief and is subjective. We might want to say, then, that science is true, and religion is not. When we looked at this simple definition, however, the less certain and clear it seemed. For the idea that science is made up of many observation of facts which are then converted into theories breaks down in the problem of induction, which in its most succinct form, is the impossibility of leaping from a singular judgement to a universal one. No amount of logical finessing will get you from a particular to a universal. This would seem to imply that science is no more objective than religion, and that a theory is as much a belief as any faith. Moreover, it was also clear that the inductionist picture of science was not accurate at all, since facts are not just littered through out the world such that we pick them up and notice common characteristics from which we then construct some universal law. On the contrary, we already come to facts with a pre-existing theory, which determines which fact we take as relevant or not (or even which fact we can see). As Ladyman explained, Newton did not find the law of gravity in Kepler’s data, he already had to have it in order to interpret the data (Ladyman 2002, pp.55–6).

This reversal of the relation between theory and facts, that theory is first and facts second, is the basis of the next philosophy of science that we shall look at, Popper’s theory of falsification, and indeed rose out of the insurmountable problems of inductionism. His argument is that we should give up induction as the basis of science, but such a rejection would not lead to irrationalism. Rather we substitute for induction, deduction. But did we not argue already in last lecture that deduction could not be the basis of science, since deduction is merely tautological? Deductive logic tells us nothing new about the world, but only analyses what we already know.

Deduction does not work as a basis of science only if we move from the singular to the universal, but it we go from the universal back to the singular then deduction does work. Indeed, this move from the universal back to the singular is exactly, Popper argues, how science works. We do not start with facts and then make laws, rather we start with laws and then we attempt to test them with facts. The logical point is that we can’t go from observations to theories, even if the observations themselves are true, but it is possible the other way around. We can go from theories then back to observational statements to show that the theory is false. Thus to use Chalmers example, if some one was to see a white raven outside the lecture room today, then this would prove deductively that the statement ‘All Ravens are black’ is false. Such deductive arguments are know as modus tollens, which take the form if P, then Q. ⌐Q, therefore ⌐P (Chalmers 1999, p.61).

When we look at the history of science, this seems exactly what happens. Take for the example, Eddington’s proof of Einstein’s theory that gravity bends light. If the theory was correct then a star that was beyond the sun should be displaced from the direction of the observer so that we could see it. Normally the light from the sun would mean that these starts would not be visible to us, but would be if the light of the sun was blocked. Eddington managed to measure just such a displacement with the eclipse of the sun in 1919. For Popper, the point of this story is that he could have proved otherwise. In other words, Einstein’s theory could have been falsified, if there had not been any displacement.

The real difference between science and religion or any other discourse is not the theories or hypotheses that they put forward, but how they test them. Popper is adamant that science is creative as any other human discourse and that the origin of this creativity is outside any logical explanation. That some one comes up with such an idea at such a time cannot be rationally explained. Thus we don’t know how Galileo or Einstein came up with their ideas, and why not someone else, or at different time and place, but what we do know that what makes these creations scientific, as opposed to anything else is that they can be falsified (this is the difference between context of discovery and context of justification). In the opposite case, it does not seem possible to falsify a religion logically. I can always find a reason to believe something. Think for example of the classic problem of evil in theology. How do I justify the existence of God with evil in the world? It is perfectly possible to find such a reason, as Leibniz did that this is the ‘best of all possible worlds’, and it is just our lack of human understanding that prevents us from seeing it so.

Here we might need to know a little of the story of Popper’s life. When he was young he was a communist and of course Marxism was treated as a science. He says that one day in went on a march with his friends and they were attacked by the police and some of them were killed. He was so shaken by this incident that he had to speak about to his political leaders. They told him that these deaths were necessary for the political emancipation of the workers as was explained by scientific Marxism. But what then would falsify Marxism, for they did not seem to be any instance, including the death of his friends that could not be explained by it.

This is precisely the difference between a science and a pseudo-science (religion is only a pseudo-science when it takes itself to be answering scientific questions, otherwise it is perfectly meaningful for Popper): a pseudo-science has the answer to everything and can never not be true, whereas a science does not have the answer to everything and can always be false. It is this that demarcates, to use Popper’s word, empirical science, from anything else and it is a question of method, rather than logical form, by which he means the positivist obsession with the correlation of statements with aspects of reality. Metaphysics and religion are only pseudo sciences when they pretend to be sciences. If they do not, then there is nothing intrinsically wrong with them. They are certainly not meaningless which is just derogatory word, rather than having any useful philosophical sense.

If what makes a scientific theory scientific is falsification, what exactly makes a falsification? Can any falsification be scientific? Such a broad generalisation does not seem to be correct because just to falsify something would not make it a scientific theory. I could falsify physics, by quoting Genesis but no one would think that I was being scientific. The answer here is intersubjective testability. One cannot conceive of how it would be possible to set up an experiment that would test my falsification of physics that claimed God had created the universe in the way that it is described in Genesis. One can imagine, however how it might be possible to test the falsification of Newtonian science through the prediction made by Einstein, which is entirely what the example from Eddington proves, and it is perfectly possible that other scientists could conceive of such an experiment, whether in principal or in practice.[1]

Could a theory always secure itself by simple adding an ad hoc modification every time a falsification was produced? Thus, to use Chalmers’s example, we could take the generalisation that all bread was nutritious to be falsified by the death all the members of French village who ate bread. We could then qualify our theory by saying that all bread is nutritious except when it is eaten by these members of the French village and we could do this every time any falsification was discovered. Such ad hoc modification would completely destroy any progress in scientific discovery. How then can we distinguish between an authentic and inauthentic ad hoc modification (Chalmers 1999, p.75). In this example, the modification cannot be falsified, so it does not tell us anything new about the world. It in fact tells us less than the original theory that all bread nourishes. So an authentic modification must be one that is also falsifiable. If we had said instead that all bread nourishes except one that is contaminated by certain fungus called Claviceps purpurea, then this would be an authentic ad hoc modification, since it could be tested and falsified, and thus does tell us something new about the world.

This distinction between authentic and inauthentic ad hoc modifications of scientific theories, however, tells us that we should not over-estimate falsifications of theories. When we look at the history of science we can see that ad hoc modifications can confirm rather than deny a theory. Take the case of the discovery of Neptune. Irregularities in the orbit of Uranus predicted that there must be another planet that had not be observed. Rather than reject Newton’s theory, scientists argued that a planet must exist that would explain it. Thus, the fact that Neptune was found in 1846 confirmed Newton’s theory rather than falsified it. Rather than seeing science as just a series of falsifications which lead from theory to the next, Aristotelianism to Newtonism to Einstein, we should see it as the confirmation of bold conjecture and the falsification of cautious ones. For what difference does it make to science if one falsifies conjectures such as the universe is made of porridge or confirms a cautious one? But how then do we determine what make a bold conjecture? The only answer to this must be background theories themselves, for only in relation to them could we know what would be bold or timid. The background knowledge is therefore the cautious conjecture (what we take to be correct) and the bold conjecture flies in the face of what everyone thinks is the case. We can see, then, what the real fundamental difference between the falsificationist and inductionist is. The first takes the history of science seriously, and the second has no conception of the history of science at all. There is no background knowledge. Rather facts are accumulated as though there were no context at all and science existed in the eternal present.

Is falsification immune to criticism then? The answer must be unfortunately not. The real problem is still the relation to the theory and the observation. All we can say deductively is that if there is O, then the falsity of T follows if the O is not given, but it tells us nothing about the standard of the evidence itself. What is the evidence is incorrect. Perhaps when person who said that the raven was white and no idea what white was. Perhaps the photograph of the white raven was created in Photoshop, and not such evidence exists.  Popper does not have a better story about the correctness of evidence than the positivist. Moreover, when we actually look at science, it does not take the simple form of ‘All swans are White’…. Rather, sciences are made up of complex collection of universal statements which are interrelated to one another.  Now if a prediction tells us the theory is false it tells is that one of the premises might be wrong but not which one or even that our own experience might be the problem.  It might not the theory that is out, but the ‘test situation’ itself, because we cannot isolate the premise which allows us to falsify the theory (this is known as the Duhem/Quine thesis). So to use Ladyman’s example, if we were to try and predict the path of a comet, the law of gravity would not be sufficient, so if the predication were incorrect we would not know that it was the theory of gravity that was being falsified (Ladyman 2002, pp.77–8).

Even if such an isolation were possible, falsification does not seem to capture actually what science and scientists do, for when we look at the history of science we do not find one great conjecture following another, but that scientists stick to their theories despite the fact that they can be falsified or they adopt a new hypothesis even though all the known evidence at the time should have killed them off at birth. This is what we find when we look at the detail of the eventual transition from the Aristotelian to the Copernican view of the world. It is certainly was the simple falsification of the one by the other. Science works because scientists are dogmatic and not open to falsification. If that is the case, how is it possible to differentiate, or demarcate, science from any other dogma? Will we not have to use different criteria?

Works Cited

Chalmers, A.F., 1999. What is this Thing Called Science?, St. Lucia, Qld.: University of Queensland.

Ladyman, J., 2002. Understanding Philosophy of Science, London; New York: Routledge.


[1] Does this open Popper to a more pragmatic account of science than an epistemological one? For if testability is inter-subjective how are we to describe it? Popper appears to want to separate questions of method from question of practice, but later criticisms will in turn want to question this distinction by asking whether it is really the case, when we look at the history of science, that scientist really are committed to the principle of falsifiability.