 Okay, thank you very much. It's good to be here to be able to make a presentation in person. Thank you for coming and today I'll try to defend a rather provocative view. The option of my talk is simply this. I'll try to show that it's possible to reduce abductive reasoning or inference to the best explanation, which is a particular case of abduction, which is very much used to defend the scientific realism. Thus to reduce abduction to deductive reasoning, that is to deductive or a special kind of deductive reasoning in the sense that it has as one of its crucial premises an inductively empirically confirmed causal statement. Ecosality will be understood in the million sense, I mean the John Stuart Mills sense. Okay, so here's the menu. I'll shortly remind you, sorry for those of you who already know where I stand as a philosopher of science. So I will briefly explain what I believe philosophy of science is. I also then discussed some popular virtues of theories and the ones which have been considered to be true to topic by some scientific realists and especially explanatory virtues. And I'll show that explanatory virtues are not too conducive. Then I briefly present the bottom of the strategy in favor of scientific realism and then a conclusion. So what do I believe philosophy of science to be? First of all, I'll focus on the products of scientific activity and not on scientific activity itself. I know that it's very fashionable nowadays to adopt a pragmatic attitude in philosophy of science and to start from the practice of scientists, scientific practice, historically situated or present-day scientific practice and then try to draw us a conclusion from that. I do not adopt that perspective. I will focus on the products of scientific activity in my theories. It's a classical approach, but it's not very much practiced today. So I call that approach contemplative, like Plato would say for example. So I try to contemplate the theories and to see what are the virtues. In this case, what are the virtues of those theories which are truth conducive and truth topic. And I adopt an empiricist rather than a strict empiricist but not too strict approach in philosophy, which is not naturalistic. So I'm not a naturalist philosopher, I don't think philosophy of science is science. I'm not going to discuss that. I'm just stating, we can go back to that in the discussion if you wish, but just stating that I believe very bluntly that science is descriptive whereas philosophy of science and philosophy in general is normative. It has to do with the good and the bad, some values. So I'll try to find out which are good arguments in favor of scientific realism. So the philosophical question I'll address here is the following. What are the characteristics of the qualities of the scientific theory which give us better reasons to believe that some parts of the theory or some propositions or statements of the theory are true or are more likely to be true. So of course the brief answer is this as an empiricist is that perceptual experience is king. That's the main bottom line that we have to base our philosophical claims on experience. So I defend a version of scientific realism which is epistemological and which I formulate in this way. The arguments in favor of our belief in the existence or reality of some unobserved entities or things. And I take entities and things to be sets or bundles of instantiated properties. So a table for example is a set of instantiated properties. The table functions, the word table functions as an abbreviation of an open set of characteristics like being hard, being stable, having four legs and things like that. So what are the good arguments and also the bad arguments which entitle us to believe that some things which are unobserved exist and also the truth of some propositions about those entities. So are there arguments which give us better reasons? Not cogent reasons in the sense that we would be obliged on the basis of those arguments to believe in the reality. It's not of course irrational not to believe in the existence of unobserved entities and anti-realist philosophers are not irrational. So it's important for me and also in general to show that there are better arguments. There are more reasons, it's more rational to believe in the existence of some entities and some propositions about unobserved entities rather than not believing in them or suspect or believe about them. Suspect or believe is a kind of atheistic position which suspensionally means agnostic. So agnostic position and atheistic position would be well those unobserved entities do not exist. So I'm against a skeptical attitude with respect to scientific theories in the sense that I have tried to show that we have good reasons to believe that some parts of our best theories are true. So in practice specifically we have more reasons to believe in the existence of say protons and bacteria than not. Okay now some popular virtues of theories that show of course empirical adequacy. Empirical adequacy is this very briefly, a theory is empirically adequate if it saves phenomena. It saves phenomena but empirical adequacy in a restricted sense, not all possible phenomena that could be discovered in the future but empirically adequate now that is the predictions of a theory so far as we know are correct or more or less correct. This is of course a necessary but not sufficient condition, it's a necessary condition for be able to defend scientific realism about some specific theories but we have the standard argument against the position which would say well a theory is empirically adequate that's fine. And then therefore the parts of the theory at least that allow us to make the empirical predictions which are successful those parts are true and but the standard argument is the end of the determination is always possible that another empirical equivalent theory, empirical equivalent theory saves the same phenomena. It's easy to show that either in a semantic approach or the syntactic approach that it is always possible to construct alternative theories which are equally empirically adequate. So it's not sufficient, it's necessary but it's not sufficient. Empirical adequacy we need more than empirical adequacy to defend subversion of scientific realism. Another virtue is the empirical scope that is the wider scope, the empirical scope of a theory and the better theory is of course that it's a virtue but again the under determination thesis comes into the picture and doesn't allow us to choose or to decide that a given theory among those which are empirically adequate is at least partially and approximately the true one. Simplicity, well simplicity it's also a virtue the theory would prefer, would prefer theories that are simple to theories which are more complicated. In fact it is a pragmatic virtue and pragmatic virtue doesn't have any truth to purchase in my opinion. The main reason is I will come back to this several times. There is no reason to believe that nature is simple according to our criteria of simplicity. So even if we managed to agree on what is simplicity which is not that clear as you know, we still, there is no guarantee that this virtue of simplicity connects with external reality. There is no a priori reason to believe that our simplicity requirements are subjectively subjective requirements of simplicity which we like simplicity to be corresponds to something real. In other words nature is not necessarily complied with our subjective requirements. Now if a theory contains inductively confirmed causal laws then that would be the chief virtue besides empirical adequacy that I will try to show that it is truth conducive. Of course I will have to specify what it means for law to be causal but if we have causal laws, I mean if a theory contains causal laws then with the help of those laws and from actually observed effects we can deductively infer the existence of causes that are observable in principle. So from observed phenomena and causal laws which are inductively empirically confirmed then it's possible to defend the existence of some causes which are observable in principle but not yet perhaps observed. And then if this works we do not need to resort to influence to the best explanation or abduction. And this is the position I'll try to argue for. Now most scientific realists as you know they defend an explanatory strategy that is if a theory explains well or even better, provides a better explanation in any sense. Of course then you have to agree or so on what counts as a good explanation which is not that easy to do but it's supposed that it's possible to have some agreement or what's a better or a good explanation. Then between competing theories, empirical equivalent theories which are equally good at saving phenomena, you would be in a position according to the realist to be in a position to assert that the parts of a theory which gives it better explanation. There is a more explanatory power than another one but this one, this one, this theory which better explanatory power is more likely to be true than another competing theory which is a theory equivalent. So an empirically adequate theory that explains better the observed phenomena is more likely to be partially true than a competing theory which explains them in a less satisfactory way. And the selected parts of the theory, of course scientific realism is always a selective thesis, you cannot by means of general arguments defend that the whole statements that are contained in the theory are true or approximately true. So the selected parts of the theory which play an essential role which are indispensable in the explanation of some phenomena are more likely to be true. So that's probably the kind of argument which is very simplified kind of argument which is usually proposed by scientific realist is the explanatory power. So what are the problems with this kind of defense of scientific realism? But there is a first problem, what counts as a good explanation? I mentioned just mentioned that. The second problem, what are the reasons to believe that a good explanation in whatever sense of good explanation, what are the reasons to believe that a good explanation is truth topic or truth conducive? That is, well, let's suppose that we have an empirically adequate theory, it explains phenomena, it's empirically adequate, it explains better the phenomena than some other theory. Do we have reason to believe that the theory which are more explanatory power is true? What would be the grounds, the empirical grounds as an empiricist? What would be the empirical grounds to defend this view? And this is, of course, I've been pointed out by several empiricists, philosophers, and ultimately passers-by. Now we have to make, to go further, we have to make important distinctions. This is a distinction that Peter Lipton, Peter Lipton in his book, The Best Explanation, has made clear. An explanation is lovely if it complies with our requirements of what counts as a good explanation. Typically, a good explanation is the explanation which provides comprehension, understanding. And again, this notion of loveliness is fishy, but that doesn't really matter as we will show for the next, for what follows. You can choose any philosophical count of explanation, then the argument which follows works as well. Now an explanation is likely if it is likely to be true. So what are the reasons to believe that the loveliest explanation is also the likeliest? Now let me quote Lipton here. What reason is there to believe that the explanation that would be the loveliest if it were true? That's an important insert in this quotation by Lipton is also the explanation which is the more likely to be true. Why should we believe that we inhabit the loveliest of all possible worlds? I would take if it were true out of the quotation, because it's important to separate very clearly, very sharply, the internal property of loveliness. Why is it an internal property? Because the explanatory power of theory is an internal characteristic of a theory. It is the architecture of the organization, the intrinsic made up of the theory which makes it, according to some criteria, a better explanatory theory than another one. So explanatory power is an internal virtue of theories. On one hand, likeliness that is likely, is the theory likely to be true? That's an external characteristic, because truth, or at least in the correspondence view which I tend to favor, the truth is something which is external. The truth of the theory is not an internal characteristic of the theory, it depends on phenomena and things out there, so to speak. So the argument, the objective argument of the explanatory strategy of the inference to the best explanation is an argument which is supposed to go from an internal property of a theory, namely its explanatory power that is its loveliness, to an external quality of a theory, namely its being, at least partially, an approximate truth. That's the argument. And here I follow Lipton. We have Voltaire's objection. So you remember Carl D. Voltaire ironically criticizes Leibniz who defended that we live in the best of possible worlds because God is good and all powerful, and then he chose to create the best of all logically possible worlds. And then you have this famous sentence back indeed in the book by Voltaire. I took a poll of you, not a minute at the moment, and then he shows that, of course, there are many earthquakes, and this one earthquake made a very deep impression on Voltaire, and of course there are many, many problems in this world, many tragedies and things like that. So it seems a bit boastful or audacious to defend that everything, to put it where you are the mayor of the world. But the point, of course, of the argument is that there is, because Leibniz's argumentation is a priori, of course, it's a idealist argumentation. There is no empirical reason to believe that external reality complies with our human or scientist or collectively scientist requirements of white counts as a good explanation. There is no reason that the reality has to comply with what we like, what we love, to be coherent with the terminology we have adopted here. I think that's a very strong argument against any kind of explanatory strategy, any kind of explanatory strategy. Whenever philosophical account, I repeat, whatever philosophical account of explanation you have, it's an internal property of a theory, and then why does the external world would have to comply with those explanatory requirements? And in particular, this voter's objection applies to the favorite argument mobilized by a scientific realist, namely the No-Miracle argument, Putnam's No-Miracle argument, which is briefly, we do know that, which is briefly the following, that science is successful, everybody is successful, it manages to, scientific theory manages to successfully predict phenomena, future phenomena as well, then it would be a miracle if those theories were not at least partially true. That's a typical example of influence to the best explanation. Here, there is only one reasonable explanation, whether theory must be true or at least approximately true, otherwise they would have success. Well, okay, but empirical adequacy, success here is empirical adequacy, empirical adequacy is now a guarantee of truth because of the internal determination objection. And by the way, Putnam says that this argument, the No-Miracle argument is a naturalistic argument, it's a scientific argument in favor of scientific realism, but truth is certainly not a scientific concept. Truth is that it's not the same status as an electron or a proton or a bacteria, so I think there is no reason to believe that it has some empirical basis, if you accept that science is by and large successful. I know sometimes it doesn't work, but there is a falsified approach sometimes, but even if we accept that empirical basis as the broad success of science, the explanation, if we want a scientific explanation of this, well, we cannot, I think, resort to the concept of truth, which is not like positing the existence of unobserved entities and things like that to explain phenomena. So instead of this explanatory approach to defend scientific realism, I defend an inductive approach, which it's bottom-up, which starts from experience, whereas the objective strategies are top-down, top-down, because what you do in an objective or in reference to the best explanation argument, you consider a set of possible explanations of a given phenomenon. So you have several competing theories, and up there, which are at the top, so to speak, and then you pick up the one which you believe has the greatest explanatory power. This one is the true one, so this is a top-down explanation, whereas bottom-up explanation is a strategy we start with phenomena, causal relationships, as I'll try to show, causal, empirically proven relationships, and then allow you to infer the existence of unobserved causes of the phenomena. So you start from the effects, and you go up to the causes, whereas the top-down approach, you start from the theories, alternative theories, you pick up the one which has the best explanatory power, and then you see how this one is the true explanation of phenomena. So what's the bottom-up inductive strategy in favour of empirical scientific theories? Well, first of all, we have to make another important distinction between properties that are observable in principle OP properties, on the one hand, and purely theoretical properties. Pt properties, purely theoretical properties, pt properties. Now the properties that are observable in principle are of course the directly observable properties like hardness and things like that, and also detectable properties. The detectable properties in my terminology are observable properties which are indirectly observationally accessible by means of instruments which improve perceptual abilities, typically the telescope or the microscope, but other instruments are like that too. And I already said that here, but I repeat for those who don't know this rather controversial thesis, properties such as mass, charge, temperature, many properties which occur in the scientific theories level, properties which are referred to in scientific theories are observable in principle. Like mass, why? Well, the objection of course is the well-known objection, the terms, mass, charge, etc., are theory-laden. They are defined, their meanings are defined in the context of theories and their meanings have been determined after a long historical process, typically temperature, you have a beautiful discussion of temperature, for example, in a professor's book scientific representation. Well, but once we understood, we have understood those terms, we have understood those terms, typically when we had the training in science, or in physics, or in biology, we understand the meaning of the terms and then we are able to apply those terms in perceptual and political situations. For example, temperature, for temperature you can, if you have a kettle with boiling water in it, then in the kitchen, then you can touch the table in the kitchen and then you can touch, if you are daring, touch the kettle, you can say, oh, this kettle has a higher temperature. So, you do not observe the temperature as such as you can observe hardness, but you can claim on the basis of very simple experiences that the temperature of some objects is higher than the temperature of another object, when you have a grasp of the concept of temperature. So, a statement such as the temperature of the kettle is higher than the temperature of the table in the kitchen, that's a true statement, which can be buttressed, the truth, the belief in its truth can be buttressed by very simple, immediate experiences. Of course, our senses are crude, but there are the basis of all even scientific assertions, because when you use instruments to measure temperature, you check the precision of those instruments by comparison, by comparison with your immediate experiences in the first place, and then you verify also that if different measuring devices that you can read, you can read, of course, by your senses, you can give approximately the same measurements. So, I think the reliability of instruments also is this immediate experience at the end of the day. Of course, this is a very broad, I'm trying to be an empiricist as much as I can, but many empiricists will certainly bark at this kind of contention, so it's a very broad and extended conception of observable properties. Then the purely theoretical properties, but in a sense all properties are theoretical, because they are denoted by theoretical terms, the terms that belong to a scientific theory, and if a term occurs in a scientific theory, it's always theoretical in that sense. But there are terms in the scientific theory which denote entities which do not have observable properties even in the broad sense. An electron as a mass and a charge, which are observable in principle in my sense, but there are other properties which are not observable even in principle. I mean the spin, the internal spin of an electron that's not observable in principle. It's not like spinning on itself, the spin of the electron. But in the elementary particle physics, you have several examples of those purely theoretical properties. So PT properties are beyond our observational reach as purely theoretical. They are observationally transcendent. They transcend the domain of observability. So the claim I try to defend is that we have more reasons, more reasons to believe in the reality of observable properties that are directly or indirectly observed, that is detected. In order to be able, you know that in the discussions on scientific realism, people speak a lot about observable and unobservable properties, which is very deaf, of course. But I think it's important to realize that it's not enough for a theory to be observable to claim that it exists. You must also observe, actually, they must have been actually observed. It's not mere possibility. So to have a good case in favor of belief in the existence of reality of an observable property must be really observed, or indirectly observed, that is detected. Detected, that means of instruments, and those instruments typically work on the basis of causal laws. That might be ignored. The typical example that I often use is that when Galileo claimed that there are where mountains and lakes on the moon, everybody equally agrees. Even the hard-nosed Aristotelian, they agree on the observations. Even if they didn't know about the laws of the telescope at that time, they agree. That doesn't mean that the Aristotelians bought the new philosophy, defended by Galileo, because they introduced had hoc hypotheses like the moon is within a hard crystalline sphere, etc. So finally, those lakes and mountains, they were motionless, immobile, and this was a way of saving the immutability of the skies. But you can easily check the reliability of a telescope by means of simple earthly, mundane experiments. When you see a ship in the telescope, and then you see that the ship has three masts, for example, then the ship comes close to the harbor and then you see how bingo. The ship does have the three masts that you saw in the telescope. So even if you don't know the laws, you can assert the reliability of an instrument that there are some causal connections. Okay, if you cannot do that, if you cannot verify those causal connections, then you do have no reason to believe that these causal connections are there, and you can only do that if you deal with properties which are observable and invisible. Otherwise, it doesn't work. Okay. So indirectly observed properties are properties which are closely connected to immediately observed properties. And the existence of a causal connection must be attested by previously inductively verified causal generalizations or laws that mention observable properties only. Of course, you can verify a law only if those laws connect causes that are observable and observable and then observed in some instances, and then you could generalize this by induction. So you have to deal all the way through with the observable properties. Now, when it's a causal law, I wrote the papers some time ago with Moldau and unfortunately, who is now working in, and I quit private companies working with the university, the VUB in Brussels. Okay, so according to us, a causal law is a universal proposition that mentions a time variation of some properties, the effect, and the other terms in the law refer to the cause. And this is in line with the Neomilion empiricist's regularity view of causation. You can verify this kind of law by means of the million methods of agreement and difference. The important thing is that again, that in those laws, you must be able to detect empirically, to ascertain empirically the presence of properties which are denoted by the terms in the laws. And the effect, again, is denoted, well, intuitively, the effect is a variation, a change, a modification of some property, a change in the course of events, an event between a space-time instantiation of a property or some properties. So this is denoted in the context of mathematical sciences in the formula by derivative or passion derivative and things like that. And the other terms are referred to the purported causes of those effects and those terms denote real properties if those properties are again observable in principle and have been immediately observed or more often detected. So that you have been able to, so that you are able to empirically and inductively verify the cause of all. Let's now look at some examples of truth-dropping arguments. An inference to the presence of an unseen mouse, which is discussed in Van Fraassen's book in 1980, the scientific image and he criticizes, of course, inference to the best explanation, but not specifically this one because this one has to do with properties which are observable in principle. And the other argument is the inference to the existence of an acute, which are taken, those two examples are taken to be classical examples of inference to the best explanation or abduction. In the Stanford encyclopedia of philosophy, if you look at abduction, the inference to the existence of an acute, that's the example which is extensively discussed by Igor Duvin, who is a well-known specialist on abduction. Okay. So the challenge now for us is to reconstruct those arguments as detective arguments from premises. Well, that's all arguments from premises and with the conclusion. So we can reconstruct those arguments from inductively confirmed premises stating causal connections. Some premises state causal connections. Now, a brief terminological remark. Induction here is taken in a restrictive sense because some authors, they use the word induction to refer to all kinds of ampliative inference. That is an inference whose conclusion has a content which goes beyond the content of the premises. Okay. So here I take induction in a restrictive sense, not any kind. I'm not meaning any kind of ampliative inference. So I mean arguments of the kind. Well, a raven A is black, a raven B is black, a raven C is black, therefore all ravens are black. And of course, these arguments do not provide any certainty. They didn't provide any certainty. Okay. So let's look at the objective form of the argument in favor of the existence of a mouse at home. So we start from an observed fact. There is gray hair on the floor, cheese disappears, specific little noises are heard. Okay, that's the first premise which states and describes immediate empirical evidence and description of facts. Then the hypothesis age is the following, the presence of a mouse in the house, best explains the facts F. Okay, best explanation. Suppose this is the best explanation according to any criterion of explanation again. So, and that is the conclusion age is likely to be true. There is a mouse in the house. Okay, so I do not claim that this is a bad argument. Okay, I do not claim that this is a bad argument. I simply claim as I'll try to show simply claim that this argument, which is good. It's good because it relies on a hidden, not mentioned premise. And this premise is a premise that states a, is empirically, whether two premises, they are empirically attested by induction. So what we propose is a reconstruction of the argument, which is deductive. Well, of course, abduction is deductive, but which is a deductive defense bottom up argument which starts from empirically verified premises. Okay, let's go. So by definition, a mouse is an animal with four legs, a long tail, small ears, a pink snout, et cetera. The word mouse, like the word table, functions as an abbreviation of a set of properties, which are associated with go together. Instead of speaking of something which is hard at four legs that can be used to eat, use a word table. So these terms, of course, again, we can discuss this, but these many times in the language, table, mouse, the function, like abbreviations. But I am in good company with this. You can wrestle, Armstrong, and so on. I mean, there are many philosophers who have defended this. And we have previously, directly observed in this case that the animals that shed gray hair, eat cheese, make some typical noises, et cetera, also possess the other properties of the animals we call mice. To shed gray hair, eat cheese makes some typical noise, also possess the other properties, namely having four legs, a long tail, small ears, pink snout, et cetera. Such constant association between properties is empirically and observationally accepted by induction. We have observed a certain number of animals which have all those properties. So having some properties that gray hair, and also causal properties, in a sense, that is shedding gray hair, eating cheese, those are processes. Those are processes that disappear in cheese, is caused by the mouse eating the cheese. So those constant causal correlations between some events, disappearance of cheese and eating cheese by mice, have been observed, and can be tested by the methods of agreement and difference. So gray hair on the ground, disappearance of cheese, regularly occur after the loss of gray hair to the ingestion of cheese, and these are causal processes. Observation previously taught us, before we make an argument, is there a new mouse in my house? We have seen mice, but it's a new one. Observation previously taught us that mice defined as a bove and only mice have properties as gray hair loss, et cetera, causally correlated or connected with the instantiation of directly observed properties gray hair on the floor, et cetera. By induction, on the basis of previous observations of mice, we conclude that the latter properties, the properties that gray hair on the floor, et cetera, are clues to the presence of a mouse. So these clues give us better reasons to believe in the existence of a mouse. Of course, if we directly test the patient's observation, if we directly observe the mouse, then we reinforce the belief in the existence of the mouse in the house. We have thus detected the presence of something that has certain properties, gray hair, et cetera, and this is an indirect observation in the mouse. Indirect observation in the sense that it is based on those causal connections. I know that this is a rather stretched meaning. I stretched a little bit the usual meaning of the word observation here. So relying on an inductively previously verified statement, we can claim that observed facts, the clues are causally connected with some unobserved OP properties. They are observable. Of course, otherwise, we couldn't have verify the causal connections that we claim are to unobserved OP properties. And therefore, we have more reasons to believe the presence of a mouse that possesses the properties causally connected with the clues, as well as other properties, the properties which are the associated properties. The later indirect observation of the mouse will, of course, strengthen our belief in its existence. So we have this following reconstruction. Maybe I'm insisting on that, but I want to make those things clear. So we have first premise facts F, the gray hair on the floor, et cetera. We have second premise inductively directly verified causal relations C. The facts mentioned in premise one are causally connected with the presence of an entity that sheds gray hair, eats cheese and makes specific noises. And then we have the inductively directly verified association A, entities that shed gray hair, et cetera, also have the associated properties of being four-legged animals with a pink snout, et cetera. Conclusion, a mouse is present, which is not a certain, which is not a, we are never certain of the truth of that, but because these, the two premises, there are two premises which are, we are not even absolutely certain of the facts F, but we are certainly not certain of the inductively verified premises because there are inductions here. And so an induction is an imperative reasoning. Okay. So this conclusion is reached because the presence of the mouse provides the best or loviest explanation of the observations, but because we make an inductive influence from previously directly observed causal connections and an association of properties characteristic of specific entities also directly observed previously. We also have a premium bonus, so to speak, a lovely explanation of facts. Of course, we have a lovely explanation of facts. The presence of the mouse does explain nicely why we observe gray hair on the floor, et cetera. We have an explanation, but we have this explanation because premise C, which is as that is a causal process, is a causal explanation, provides the basis of a causal explanation here. It's a kind of explanation. There might be other kinds of explanation, but here we have an explanation. The causal explanation, we relate the effects of certain effects to a cause. The argument for the presence of a mouse is deductive and logically valid. If the premises FC and AR2, this argument is sound and also provides a correct explanation of the phenomena. So, we do have an explanation, but it's not the explanatory power, the loveliness of that explanation, which is behind, which grounds the conclusion that there is a mouse present in the house. Objection. Of course, there are objections. Always objections is philosophy. This deductive argument isn't convincing because we still must take that in this specific case, we do have an occurrence of the alleged cause of connection. We know that there are cause of connections. I suppose that in general there are cause of connections. But here, you have the gray hair, you have the specific noise, et cetera. And we know that mice have such and such and such and such. But in this case, what evidence do we have that there is this causal connection? The causal connection between this gray hair and this mouse in the house. That's the crux of the matter here. Well, there might be other causes of the phenomena. That's a typical objection of the interest of the best explanation as well. There might be other causes. But here, since I want to be a non-explanatory, then I have to reply to those objections of the alternative explanations, not on the basis of the explanatory power. All right. To be coherent. Hopefully. So there might be other causes like maybe a rat. A rat that has gray hair, you know, the noise, the cheese, anything, including cheese and so on. All right. Or a neighbor, you know, who doesn't like me for whatever reason, manages to have a copy of the key of my house. And when I away, he put those, he has mice at home and he put gray hair on the floor and he put cheese and the day after he made the cheese disappear and things like that. I mean, you can imagine other explanatory stories that may explain the evidence. Okay. But you eliminate, we can eliminate these alternative hypotheses. Not on the basis of inferior explanatory power, but on the basis again of experience. Because what you said, look, the rats, they don't have that kind of gray hair. We have seen that and they don't make those little noises that we have heard before produced by mice. So we can discard. We cannot, of course, discard any unconceived alternative. The word inconceived alternative expression, which often comes in this kind of discussion. But that doesn't matter anyway, because maybe there are some alternatives. But why are they? Why give me one? If you say simply it's possible that you have other alternatives, then maybe we can discard them. And of course we are not into the certainty. There is no certainty. There might be alternatives. But we are in a position, do we have good reasons now on the basis of the evidence that we have now to have more reasons to believe in the existence of mice rather than not? Okay. So we can also eliminate the hypothesis of the malevolent neighbor, because he is very nice and so on and so on. And I have a security key and things like that. Again, those beliefs which are grounded on inductive empirical reasoning. Another objection. Some say the IBE, in reference to the best explanation, is just the following. The best explanation is the one which is the likeliest to be true. Okay. So a genuine IBE is the one which is inferring likeliness from loveliness. Okay. But if you pick the likeliest explanation as the best one, then you get a circular argument. Again, this is discussed nicely by Lipton in his book. What you want as an explanation is what you want is an argument that goes from an internal property of a theory that is its explanatory power, its loveliness, to something which is not intrinsic but extrinsic, external, that is, its loveliness to be true and to fit reality, to fit the facts. Okay. If you say that the best explanation is likeliest, then you don't have a, perhaps you have a difference in the best explanation. We are not discussing terminology issues here. You don't have an interesting argument. What you want is to go from loveliness to lackliness. And there is a problematic connection between loveliness and truth. That's the point of what there's objection. And as I tried to show the elimination of alternative explanations for us from like lack empirical evidence or falsification, lack of empirical or that there is specification which is based on observations. Well, to conclude, let's briefly look at the Neptune's example. To recall the beginning 19th century, the French astronomer Bouvard detected by means of a telescope that the planet Uranus had anomalies in its trajectory, in the sense that its trajectory did not conform classical mechanics, even if we take into account the other known planets, typically the biggest one, Jupiter. And the anomalies couldn't be caused by a comet, because the comet is tiny and then doesn't have a motion. You could make an alternative explanatory hypothesis and come back to that, but this has a heuristic road. And then you check whether it works empirically, but not on the basis of loveliness. Come back to that again. Since the laws of classical mechanics have been widely confirmed, Adams and Laverrier concluded that a known planet caused the anomalies. Of course, you could say, well, that's a classification of Newton's laws, but since Newtonian mechanics had been confirmed in many instances, that was rather unclosable to put into question Newtonian physics. And in 1846, the visible properties of a planet called Neptune were observed by telescope by a German astronomer, you and Gala in its predicted location. Planets, so we have to start with definitions, with the definition of planets, because as we started from the definition of a mouse in the previous example. So planets are observationally defined as light points that seen from Earth move periodically along the constellations of the zodiac. According to classical mechanics, the planets also revolve around the Sun at mass speed and acceleration. Their motion is described by Newton's laws. Relying on Newton's inductively verified cause of laws, we can infer from the effect, the anomalies, its cause, the motion of a known massive body, a planet whose motion conforms classical mechanics. And then we have the following reconstruction. Fact F, anomalies in the trajectory of Uranus. We have inductively verified correlations. According to Newtonian theory, anomalies are causally related to some massive bodies. Inductive elimination of alternatives E, other hypothetical cause of connections with comets already known planets are not empirically verified. Association A, the properties of mass visibility and periodic motion around the Sun define what a planet is. Conclusion, a new planet Neptune probably exists and we have a further confirmation detection at the predicted positions. So again, we do not rely on the explanatory power of the hypothesis of a new planet to conclude that this new planet is likely present, but we rely on inductively confirmed causal laws, Newton's laws and associations observed associations in the case of planets. And this is a bottom up approach in favor of the existence of this new, as yet unobserved planet. Now, I do not deny of course that abductive reasoning are useful, because abductive reasoning does help in framing new hypothesis. Why not? You can always imagine new hypothesis or new theories provided they are empirically adequate. That would provide a nice according to a criterion, your criterion, an explanation of phenomena that's not prohibited, of course. It has a heuristic value only. It does not have a truth-productive ability. Simply, it helps you to frame new hypothesis, then those hypothesis you will see whether it fits the phenomena. If it's very empirically verified by the phenomena. You see, maybe it's a comet. Fine. This is not, of course, a stupid hypothesis, but then you see what is the mass of a comet. Typically, we have observed previously that comets have very small mass and this very small mass and they have a certain kind of motion. And then you put that in the framework of Newtonian mechanics, you calculate and then you see, well, it doesn't work. So, you put it aside. That's not an abductive reasoning. That's to check whether a proposition is true on the basis of observation. You do not rely on the good explanatory and purport it or let the explanatory power of the hypothesis of a comet. And also such abductive reasoning is guided by background knowledge. Background knowledge that is what you know, what physics or science knows. The knowledge that science provides you about what's going on in the world. You cannot frame anything. You cannot suppose that a god or mansions or the miracles occurred. That's something that it discarded beforehand on the basis of previous broad experience, which is encapsulated in the background knowledge that you have. Well, PT properties, of course, these are beyond the pale. Conclusion. Allegedly explanatory virtues are not conducive to truth. Explanatory loveliness is not true to topic. If a theory has the virtue of containing empirically and actively verified goes along, we can deductively infer the existence of causes that are observable in principle. And that may be detected later from observed effects. Then in this case, we have more reasons to believe in the reality of these causes than not. Okay. Thank you very much. Looking forward to your questions. Just before the question, we should take a few minutes to change the air. That's the reasonable thing to do, but very short time. Just open the windows a little. A few minutes. You open on your side. Put back your slides. Sorry. Slides. Charles, do you see the slides? You can ask questions here. Here, on the left. No, no, no. If someone is slow, you can leave them. We'll do what we can. Do you have any questions? Do you have any questions? No. Thank you very much. Just one question. I'm not sure. Do you think it's a possibility that people can ask questions? Just for the time I ask the question. Thank you very much. I'm not sure about one thing. So you showed us a function could be replaced by a deduction based on inductive premises. Am I right in thinking that you claim that this is only possible for observable properties? This would not work for spin instances like that. Is that the case? In that case, we need an explanation for spin instances. No, in that case, I think that we ought to be agnostic about the existence of spin. Even if our explanation does not work for this kind of property. No, because the definition of the best explanation is no. This is a volunteer objection. Of course, you can posit the existence of a spin of a particle to account for some experiments. They can go spin to these things. For the electron. Fine. You can do that. But then, since the volunteer objection, you have a nice explanation. And I don't dispute. That's a nice explanation. You have spin, etc. And it also has heuristic values and things like that. Fine. I don't think the philosopher is a censor. It's a policeman that is on the shoulder of a scientist. You are not to speak about spin anymore because you have no reason to believe that it really exists. But the volunteer objection applies to that. Because the only way you can argue in favor of the existence of these purely theoretical properties is explanatory power. And then explanatory power does not give you the leverage to go out to reality. Because again, a volunteer objection. It's unfortunate. I regret that. But regret is not a subjective feeling. It's psychological. It doesn't have any truth, conducive power either. And I think you ought to be anti-realist philosophically about the existence of spins. If these objections, volunteer objection against abduction holds. Because since spin is a PT property and which is transcendent, so to speak, there is no way you can ascertain that there is a causal connection between the property of having a spin on the one hand and other properties which will be the effect of this having a spin. You can of course observe purported supposed effects, but you cannot detect the presence of the spin by means of other laws which have been previously verified, because it's impossible. It's impossible because the spin is not an observable property. That's my point. I'm thinking about that. So if spin is the best explanation, the existence of spin is the best explanation of some muscle. But it's also the best explanation of another sort of a phenomenon and another one. And we have different reasons. This is the best explanation for several phenomena. And so we can perhaps back up one argument to the explanation by another one. In that case, perhaps there is some... It's just not one phenomenon. The only reason you have is that relation to that phenomenon. As an explanation of that phenomenon, you also have other reasons. In that case, is it stronger? I don't think so, because if you have a bad argument, and you repeat that bad argument in some other instances, you still have a global bad argument. If you think that the relation to the best explanation is not true, it's a topic for the reasons I tried to show. Then you can have... It certainly shows that the positing the existence of a spin is a very useful science. You can work pragmatically in this case. Pragmatically there is some virtue. Pragmatic virtue. I'm not disputing the pragmatic virtue of positing the existence of a spin. Certainly the fact that it works in several instances, then it gives good reasons to believe in the pragmatic utility of the spin. And what about if there is some empirically backed observation that the several phenomena which are explained by spin are correlated. So they appear together. So they seem to be linked, because we have some reason to think that they are linked. We have no explanation except... Yeah, that's the typical argument used by Rechenbach, the principle of the common cause. You have to say, okay, then it seems that you have a common cause. And yeah, that's true. But since you have no reason to believe that in each particular instance the cause is this one, it's not empirically verified, you also don't have any reason to believe that there is this common cause. Of course, the consequence of the argumentation, I didn't see that, but that's a good point. This kind of influence to the common cause principle which is used by Rechenbach, who is considered to be the physicist after all, which is criticized by Van Fraassen, as you probably know, then the principle of the common cause. It's not acceptable for someone who doesn't believe in abduction or influence to the designation. That's the consequence. You cannot believe in the principle of the common cause and be against abduction. That was the first statement. Yeah, several things, but thank you for this talk. It seems to me that while it's kind of almost obviously true that because something has some exploratory virtues, you cannot conclude from that that there's no, you cannot say that it's true or whatever. But there might be, it's not implausible to say that you can say something that gives us reasons to believe certain stuff about truth. Not per se that it's true, but that it's more likely or something. Because the things we know about the heuristics or the momentary features of explanatoryness, they have developed for a reason. We have learned that by these kinds of explanations, we get successful scientific practice, which is mostly empirical undertaking. You could have some kind of evolution argument or something that selects for the best norms for explanation. The best in the sense that it has the most chance of getting to truth. So that could be a way to say that, well, probably are the best tools we have now to come to explanations. Actually might have as good reason to believe that they have some more likelihood to get to truth than things that are not good explanations. So I don't know whether it's entirely internal, I mean it's mostly internal and there's nothing with absolute certainty to say that it gets you to truth. But it might be that it evolved in such a way that there's a slightly better chance that it's not unrelated to truth. It's not completely coherent this thing. Okay, absolute certainty, it's a will of the wish, as we all agree on that. But I think that if I understood correctly, you explained quite convincingly why people believe in some abductive explanations and their conclusion. So what you gave, it seems to me it's a naturalistic account of some beliefs. Fine, that's fine. And it is true that when we propose an explanation, this explanation is based on, in most cases, on some statements or propositions which are by large inductively empirically confirmed. So that's what I mentioned in the background knowledge. So it's not an explanation, the explanations that I propose are not out of the blue. They are plausible and their plausibility is based on what we previously observed. So that's why I said the argument in favor of the existence of a mouse is not considered as a bad argument by Valfrassen. Because the explanation that is provided is at the end of the day based on empirically verified and inductively correlations and associations. But if you want to by kind of argument infer the existence of spin and things like that, you don't have a sufficiently solid background knowledge, solid in the sense, again empirically buttressed to sustain the existence of these purely theoretical properties. By analogy, I mean a scientific release they say, okay, well, scientists use that argument, inductive argument, which is true. And deposit entities, the existence of entities like Neptune, that works fine. So why not extend this kind of argument to purely theoretical properties like spin, charm, strangeness and things like that? But that's I think it's a sophism. Because it's true that the form of the argument seems to be the same, but the nature of the premises or the hidden premises that you are relying upon is not the same. It's not empirically grounded. Enough empirically grounded. I have a question, but I will ask one. So your argument rely a lot on the fact that causality, causal law or dynamical laws or physical dynamical laws. And so when you have this conception of causality, it's easy to exclude common cause arguments and all that. But of course, most scientists today defend some kind of manipulation account of causality. Like Pearl, where the principle of common cause is at the basic of the construction. So they cannot get around, they cannot say it's not there. It's an axiom of all the research of causal relations. So would you not say that at least in certain contexts when it's related to observable problems, since the ground research, the basic research of causal relations is based on this axiom, or is it also an inference to this explanation that I'm doing without? I'm not familiar with what you mentioned, but I tend to believe that the manipulability, I'm not sure I will be able to answer to your question in a very complete and accurate way, but just a few remarks. First of all, I tend to believe that manipulability is based on causal laws. If you are able to manipulate, then there are some causal laws operating that you might ignore. But I know that there is a good inductive support that if I take this glass and I put it around, it will go the way I want it to. I don't know what the causal laws are behind that. Well, not all of them, but not all of them, of course. First of all, a second remark. The principle of the common cause is perfectly valid if the common cause is observable in principle. That's what I would reply. But if it's not, then I'm rather skeptical about that. But it would be interesting to have some reference in the literature about that. But just to add, you're right that it's quite possible that many of the manipulation account is grounded in more processional causal law. This is the position of Peter, for example. But it's not a proof. We don't know that. There's a nice argument that they wish to say it could be true. David Lewis explained how even the fact that the manipulation account seems directional is not proving exactly that, but it's proving that for counterfactual. So you could have directional causality that is grounded in reversible laws. It's quite possible, so it shows up formally. But why not? We don't know. It's an hypothesis. We cannot check for our cases. And you're right that it's true that most of the research I have in mind, like Pearl, like all these guys that are building news models to discover causal relation and disease and complex factors in economy, it's always related to observable variables. They don't talk about what would be non-observable. So in that case, you would say, okay, this principle could be used because it has some empirical support. Fair enough. I have other questions, but I will leave them here. Thanks a lot for the very nice presentation. So I just have a few questions. Sorry a bit. So yeah, we have to select one question. So it's just a small point about the mouse example. So perhaps if you could please go back to the slide where you present the example. And especially I want to draw your attention about the claim made to the effect that we have observational evidence that only mice can cause those effects on observable variables, which is something that I intuitively resisted a bit. And I think it's related to the program of unconvincing alternatives. Okay, okay. Well, I took the example of temperature because I've already presented the example of the mask here in several seminaries. Well, first of all, you have to distinguish between inertial and gravitational mass. So let's take the mouse example, not the mask. Oh, sorry, sorry. I understood. I understood mass. Sorry, sorry. Is that the slide you wanted? No, I think it's slightly more text. Yes, that's the one. So the first paragraph of observation was given to us at mice, defined as above, and only mice have properties as great errors. And I'm curious about how you, if I push you, say how do you know by observation that only mice will say well? You should know by observation of mice. Okay, okay, okay. Okay, well, that's a good point. Yes, okay. I should drop the word only that on the basis of our experience so far, we think that there could be other animals which resemble very much mice, but which differ from mice with respect to some other property. And then, okay, I agree. But I think that changes, you know, the point because if you say mice, we observe that mice are that. And then someone would say, well, of course, there might be other animals. But that is the unconceived alternative issue. And then what you have to do is to contradict me if I claim that there is a mouse in the house. That there is another. There are other animals that have been observed and they have the exact same properties as the ones, as the clues that we have observed. And but they differ from mice with respect to other properties. For example, they have a black snout or something like that. But okay, I agree. I think that's a good point. That was a very short point. So I anticipate that we fall back on the problem of unconceived alternatives. And your solution to it is to bring, like, trying to think correctly to bring some probabilistic considerations so that the available hypothesis is the one about mice is the most probable rather than not. No, it's simply more reasons to believe. We have more reasons to believe. That is, I quite agree that there is the inter-determination thesis and there are possible alternatives that are unconceived alternatives, but unconceivable, right? I quite agree that there is this possibility. But they cannot be taken into consideration because they are simply put on the table, so to speak. And then we cannot check whether they are true or not. I stick to what you already know. So until you provide me an alternative, I don't need to put it into account. I don't even need to put the probability, if I understand correctly. Because probability would be trying to guess about things that I don't know about. I thought that somebody introduced the idea that you take the most, like, clear or... Okay, okay, okay. Perhaps we'll be more careful about your likelihoods. We lack absolute certainty because the possibility of other alternatives jeopardize the certainty that you can have about your conclusion. But I wouldn't speak in terms of probability or likeness. But I spoke of likeness because the premises are not absolutely certain. Perhaps the example is the idea that I suppose in many cases we have... So we know that there are various things that could cause, like, a lot of evidence, like, the mice and stuff. And even though it's okay, there are some, like, truly imaginative hypotheses with a true possible explanation that doesn't correspond to anything that hasn't been observed before. But if you do, like, like, criminal forensics, you know, some criminals... So planting evidence, something that has already happened, something that happens, we have observed in the past. So it seems like what's the difference between the hypothesis that evidence was planted or the hypothesis that actually the criminal that might have succeeded in this thing actually did it is something that has to do with... Since you mentioned likeness, I would say that you can introduce some sort of uncovered probability sequencing. It's perfectly not... So it's a terrific question. Perhaps it's not what you're not in mind, but if that's the kind of thing that you lean towards then I'm not sure that the argument that you provide instead of the hypothesis explanation is actually purely deductive argument, seems more like a probabilistic argument. But it's a... I think it's a deductive argument, but I use the word likely in the sense of Lipton. You know, it's not sure, but it's more certain or less uncertain that it's true. But it's true that you could introduce the probability at the level of some premises because they are inductively addressed. I don't know why, but that's a huge topic. I have a follow-up just to give you a look at this. So let's just do it, the premise, the inductively obtained premise of the elimination of the alternatives, right? It's only... Sorry, I just write something. Okay, sorry. This word only mice here is not... It's important because you have this premise of the elimination of alternatives and this only mice is supposed to do that, right? It's not your neighbor who's playing tricks on you. It's only mice that... Yes, yes, yes. But I wonder whether this is really part of the... I mean, and then you get to the elimination of the alternatives by empirical verification, but this is not part of the usual adduction process that's afterwards. You look for some explanations, probably the best ones, and then you're going to... You'll check the nature and check which ones are good, but that's already the second phase of that. I mean, that's not part of the same arguments, I would say. No. The creative process, and then you have to go back to nature to go into... I agree, I agree. So the first part is heuristic. I see. The first part is heuristic. You propose a hypothesis and a possible explanation, but then you cannot infer the truth or the... Well, you cannot infer that this explanation is more likely to be true than another one. You have to make the second step in order to have a good reason or more reasons to believe in the truth of that explanation. Because you gave the example with the comments and the comics. And I didn't think that was very convincing because the reason why you dismiss a comment as the explanation for... Uranus or Jupiter? But you have to explain on that. You have to explain the movement of another planet, which was the Jupiter. Uranus. Uranus. Uranus. Uranus. Uranus in the trajectory of Uranus. You frame some hypotheses like another unknown planet or Jupiter. You have to take into account maybe Saturn or perhaps there are comets. Yeah, but the comet thing is specific because there the reason why you dismiss it is not so much empirical, but there's already theoretical reason that comets don't get this size and so on. It's empirical because you eliminate the comets on the basis of empirically verified of the properties of comets. For example, typically they have very small mass. And second thing is that based on Newton's laws, you insert the mass of the comets and you see whether you can, on the basis of Newton's law, you can correlate the mass and the trajectory of the comet with the deviation of the trajectory of Uranus. And it does work. Okay, but there's quite a difference between this sort of more theoretical work and why you dismiss comets and seeing that, I mean, going afterwards and check empirically that certain potential explanations do not work. You can do this in advance without any new knowledge. Say, well, it's unlikely that it's a comet because comets we've seen such and such. While with the mice, you might have to check whether your neighbor can have been inside of your house or no. It seems very important whether you verify after you've done the abduction process empirically or you do this as a premise of the abduction reasoning. And in the comet case, it seems to be a premise that allows you to eliminate while in the case of your neighbor putting fake evidence of mice in your house, it might be more an afterwards thing, another premise of your house. The point with the comet, I agree that there is a difference, but with the neighbor. But the point is that to discard the comet and to discard the malevolent neighbor, you based your elimination on some empirically verified causal relationships. But of course, a causal relationship you're relying upon to eliminate a comet is more theoretical than the case of the neighbor, of course, because it's new to the mechanics. But the point is that these are both cases, those causal connections, these hypothetical causal connections are not addressed by observation. All right, it's four o'clock. But we can lay by default. So if we should agree, you can continue ten minutes off. Yeah, it's your choice. Okay, I'm not sure I understand the observable and principal scope of this notion because let me give you an example. So if you understand and you also agree or you understand, but I'm not sure I try to understand the extent because if I understand correctly, using your distinction, you would make a difference between the inference for the existence of net two and the inference for the existence of neutrinos, which were also based on observable anomalies that were... Sorry, I interrupt you because I'm a realist about some properties. So the neutrino, it's again a name, what are the properties of neutrinos? What is the property which is typical of a neutrino and only neutrino, which is observable in principle and which is connected to observation? Typically, I would say energy. Energy is an observable in the wide sense. It's an observable property. And then if you are a particle which has this amount of energy, such amount of energy, and it's this amount of energy is connected to by means of laws, verify laws to some observed properties, then you have a good case for the existence of neutrinos. So it's a realism about instantiated properties. Okay. So if the neutrinos and net two are very similar case, according to what you just said, because they are based on causal laws, dynamical laws, but your notion of causal laws, that you presented here, I think. Yeah, yeah, it cares a lot. So why not include spin? So spin is detectable. Energy is detectable through quite almost direct means now. So if you accept, when I try to understand it's why you accept certain instrument-driven properties and not others. Yeah, okay, it's a good question. Because just to finish, I would be, I think the detection of neutrinos, not at the beginning, but now is as calibrated as telescopes. It took time, but now it's very calibrated. It's almost direct, almost in a certain sense. So why, where do you come? Because the idea is this. Because wherever, you know, even if you have a very big spin, an intimate spin, that will remain totally inaccessible to us. Of course, a very tiny amount of energy, like the amount of energy of the neutrinos is so intricately inaccessible to us, right, directly. But different amounts of energy are empirically accessible to us. So we can give an empirical meaning to the term energy, and we cannot do that with words. So that's the idea. So in that case, you should look at the, there's a Newton's principle in the Principia, that exactly argue for that. What you can observe, property that you can observe by cross-property, should exist at all scales. A property that you cannot observe by cross-property, maybe they don't exist. But I wouldn't buy the fact that if property exists at some scale, it will exist at all scales. But it's just... Where does he write that? It's the analogy of nature. It's often translated as the analogy of nature in Latin, I forgot. It's in the principle of the good science and the good books. It's analogies of nature, yes, yes. There's a long paper, the historian of physics, about just what is the meaning exactly of that. And it's something about properties that I can attribute energy to very small thing, because I can attribute energy to Microsoft. Right, that's right. That's where you agree with. The paper, I can find the historian, there's a good paper to explain, because it's two lines in Newton, but there's a good historian to explain the context and why it should be some kind of... If you recall that... Transduction, it's kind of an addiction from the microphone. Yeah, exactly, it's sympathetic with that, yeah. It's okay. Okay, thanks.