Why I'm Not a Scientific Realist: A Philosophical Analysis

[00:00:00]hello youtube uh so i've mentioned occasionally in in some of some recent videos that i'm not a scientific realist uh i used to be but i changed my mind and i've been asked by a few people if i could do a video explaining why i changed my mind so uh this video is going to be a bit more personal i'm just going to explain some of the reasons that led me to uh change my mind on on realism if you're not familiar with the realism debate you might want to check out my series on scientific realism but very briefly scientific realism is the view that our best scientific theories are approximately true and the entities and processes postulated by those theories really exist there really are electrons neutrinos black holes genes the citric acid cycle and so on and these things really do behave in the way that our best scientific theories describe and i think this is basically the common sense view of science this is pretty much what everybody thinks about science you know before they before they do any philosophy of science science describes the world accurately so if i'm not a realist what am i well i would say i'm something like a constructive empiricist we should treat scientific theories as tools for the prediction and control of observable phenomena where the observable phenomena are essentially those things that are detectable with the unaided sensors so we should believe what our best theories tell us about the observable world but remain agnostic concerning anything they say beyond that the technical term here is empirically adequate a theory is empirically adequate just in case everything it says about the observable world is true i believe that theories are empirically adequate but that's all or maybe i should say more precisely i i believe our theories are approximately empirically adequate obviously scientists still make mistakes and no theory perfectly describes the observable but um you know it does get it nearly right um so why did i abandon realism well i suppose the first question there is why was i a realist in the first place uh part of it i i guess was as i noted realism is the common sense view it's it's kind of the default position um but what was my justification for being a realist well i basically accepted what's known as the no miracles argument hillary putnam puts this argument very succinctly he says the positive argument for scientific realism is that it is the only philosophy that doesn't make the success of science a miracle science is remarkably successful in many ways it makes incredibly accurate predictions it's applied in the development of very sophisticated technology and the thought is that this success would just be incomprehensible unless our theories were at least approximately true it would be a miracle for a bunch of false theories to exhibit such successes the best explanation for the success of our theories is that they are true that's what motivated my realism and i think that's what motivates most realists i mean just ask yourself why do you believe in electrons and not in for instance the greek gods i mean surely it's the difference there is that the theory that postulates electrons is so successful so we're inferring from the success of a theory to its truth so with this said there are a few reasons why i changed my mind about realism the first reason is a pretty common one i guess in the literature so um i won't talk about it in too much detail but um for the no miracle argument to work we need to assume that there's a connection between success and truth false theories will be unsuccessful whereas true theories will be successful obviously you have to assume that if you're going to say that success is a good indicator of truth and the problem is that both of these claims seem to be mistaken for one thing we know that there are false theories that are nevertheless successful newtonian mechanics is remarkably successful but it's completely wrong in terms of its description of the the underlying structure of space and time and mass i mean so it makes very very accurate predictions about what we observe and it can be applied in all sorts of contexts whether you're just dealing with like a pendulum here on the earth or whether you're talking about the orbits of the planets very remarkably accurate predictions in a wide range of contexts but it seems that the underlying metaphysics is now utterly rejected and perhaps a more important example of this actually is that it's not just that past theories are false falsehoods are also littered through contemporary science as well because scientists will constantly employ idealizations and abstractions to simplify their models for example in astronomy when scientists model the interiors of stars they make all kinds of simplifying assumptions they may assume that there is no density gradient throughout the star that the star is composed entirely of hydrogen they may ignore factors like radiation pressure in many models the the equation of state is that of an ideal gas but we know that no real gas is an ideal gas i mean so all of these things are false um idealizations such as these simplify the models and they make computation easier but it's important to emphasize that i mean the these these kinds of claims are false and many of them are not even close to being true so an ideal gas is uh composed of dimensionless particles that interact only through perfectly elastic collisions uh on the molecular level at least no real gas is anything like this i mean it's not like real gas molecules somehow are approximately like dimensionless particles no instead it's that in certain contexts real gases uh are observed to behave in ways that we would observe ideal gases to behave if there were ideal gases but on the you know on the underlying on the molecular level in terms of how it describes the underlying uh metaphysics as it were there's nothing like an ideal gas or in biology models that identify selective forces on populations may make the simplifying assumptions that the population is infinite or that there is no inter-generational overlap or that traits are passed unchanged from parents to offspring again the idea that say a population is infinite that's that's not even approximately true so for more detail on these points you might want to check out my video on models and fictions in the scientific realism series the important point is we certainly can't assume that success entails truth because there are all kinds of models that are very successful but that include claims that are just flat out false so it seems that falsehood is compatible with success on the other hand many many many claims may be true but unsuccessful so yeah realists often speak as though once you have truth you have success as long as you've got a true theory you'll be able to make good predictions you'll be able to apply that theory to create sophisticated technology and and so on and i mean so to some degree i can see the justification for this um if your theory provides a totally true description of the unobservable world then presumably a true description of the observable world will follow from that right you know if you have a completely true description of the unobservable that's going to entail a true description of the observable and of course making successful predictions is really just a matter of accurately describing certain observable phenomena so okay truth does entail success but the problem here is that nobody actually believes that our best theories are true this is completely implausible even not even the most extreme realist would say that our theories are true period what the realist says is that our best theories are approximately true and i mean you have to say this because nobody thinks that science is complete there are still changes to come in the future um we haven't got it completely 100 right at the moment so nobody's saying that our theories are totally true the claim is that they are approximately true but it's not at all obvious that an approximately true theory would be successful changing theories in relatively minor ways will often destroy their predictive accuracy just consider what would happen to the predictions made by science if you were to make relatively small changes to what physical theory says about the properties of fundamental particles or forces like if you take current physics and just alter it slightly so that the strong nuclear force is two percent stronger now if we think that current physics gives an approximately true description of the fundamental physical processes and forces then this modified physics where the strong nuclear force is two percent stronger would be approximately true in this respect also um yeah because again right the only change we're making here is increasing the strength of the strong nuclear force by two percent so it's like if if if physics is currently approximately true in in this respect in terms of how it describes the fundamental processes then that modification would would only be slightly less true right so it's still going to be approximately true but this modified physics would give a completely inaccurate picture of the observable world paul davies has suggested that if the force were different in this way then almost all the hydrogen in the universe would have fused into dye protons which is a particular isotope of helium within just a few minutes of the big bang that prediction is massively wrong um so you know if you if you tweak what theories say about the unobservable very slightly you will often get a completely useless theory perhaps a less extreme example but if you look at the copernican revolution i'm sure we would all agree that the copernican system was much closer to the to the truth about the uh real structure of the solar system than was the uh than were the ptolemaic or tyconic systems which both held that the earth was the fixed center of the universe i mean there's really no question that copernicus was closer to the truth because we've since become able to actually observe the real structure of the solar system so you know we know the copernican system was truer but initially the copernican system didn't offer any improvement in predictive accuracy it took a lot of time and effort to develop the copernican theory before it made better predictions than its rivals so the point is that having an approximately true description of certain parts of the unobservable world isn't enough to get you success um indeed probably almost all theories that would be approximately true of certain parts of the unobservable are in fact unsuccessful um and taking a false theory and making it truer won't necessarily make it more successful so there can be successful false theories and there can be unsuccessful true theories and this means that this the realist attempt to draw a connection between success and truth is i think looking pretty implausible okay so that's one concern about realism as i said this kind of argument is i think fairly familiar in the realism debate but that is you know that's one reason why i sort of drifted away from realism there are i guess two other reasons for my anti-realism which i think are a bit more original so the second reason is what i guess i'm calling the selection argument uh and uh i think this argument is basically original with me as far as i know i don't know anybody else who's offered this so the the realist says that success is explained by truth let's put aside the problems that we've just discussed and assume that this works right let's let's assume that if a theory is approximately true it will be successful so so it's at least plausible to explain success by appealing to approximate truth now obviously insofar as this is an argument for realism um there's a kind of there's a there's a challenge here to the anti-realist if truth doesn't explain success what does right what's your explanation for success um most anti-realists i think would tend to want to resist this demand for explanation explanations have to come to an end to somewhere and the anti-realist will say well it's it's perfectly legitimate to just say that we don't know why a particular theory is successful we don't have to give an explanation nevertheless anti-realists have considered alternative explanations one of which comes from baz van frassen van frassen says and i quote i claim that the success of current scientific theories is no miracle it is not even surprising to the scientific darwinist mind for any theory is born into a life of fierce competition a jungle read in tooth and claw only the successful theories survive the ones which in fact latched on to the actual regularities in nature of course van frasen is an anti-realist and this statement should be understood as meaning the observable regularities um anyway so scientific theories are successful because science has a darwinian selection mechanism for weeding out weaker theories well that's an interesting idea unfortunately it doesn't work as a response to the realist the reason is we have to distinguish two questions so the first question is why do we have successful theories and the second is why is a particular theory successful so it's like you know what first question is just okay how come we've managed to get all of these very successful theories the second question will like look at a particular theory and say well why is that successful why why is it that theory that makes such great predictions such you know such with such remarkable accuracy now van frassen's explanation is only an answer to the first question van frassen um so what what this explanation does is well the fact that science has a selection mechanism that explains why uh successful theories accumulate while the unsuccessful ones are lost but the realist explanation is an answer to the second question the realist explains the success of a particular theory say general relativity i mean to say general relativity is successful because science success for successful theories would be utterly idiotic um the question is right why was general relativity selected why is it why is it general relativity that is that makes such great predictions so selection and truth are in fact compatible theories may be selected because they are true and this is just what the realist says the realist will say the fact that a theory is true explains why it's successful and the fact that science is a selection process explains why this theory has been retained over its less successful rivals so van grassen's um alternative explanation is really no alternative at all right that's the background to my argument what i would say is i think there's actually a very important insight in van fresn's point that science is a selection process as i've noted anti-realists want to reject the realists demand for an explanation of the success of science and here's what i would say the fact that science is a selection process shows that we should resist the demand for an explanation of success in the terms that the realist requests right so given that science is a selection process the type of explanation that the realist provides is inadequate so let's consider how explanations work in the context of biology where we have a very well-developed theory of selection consider crows crows are remarkably good at flying and let's say we want to explain why it is that crows are such good flyers we want to explain that is why are crows so successful where success is the ability to fly and here's a possible way of proceeding well first of all we might specify a possible structure of the crow's wings and we show that this structure exemplifies certain aerodynamic properties then second we show how this structure might have developed in crows through the process of natural selection and then we conclude that well crows must have this type of wing structure that explains their success at flying this is exactly analogous to the realists selective explanation of the success of science right first the uh realist specifies a particular property of scientific theories namely truth that she thinks will lead to success where success is confirmed novel predictions or whatever um second the realist shows how our theories might have acquired you know how we might have acquired theories with this property via a selection process and she concludes that our best theories are in fact true now here is the uh the crucial point the explanation of the crow's ability to fly is acceptable only if we independently establish what the structure of the crow's wings actually is it's not good enough simply to note that crows are successful flyers and that a certain wing type x would make crows successful flyers and then to conclude that crows do in fact have wings of type x we also need to have independent evidence that the crow does in fact have wings of type x if we wish to explain an organism's success at achieving some goal by achieving by appealing to particular properties of the organism we must first independently establish that the organism has those properties that is a minimum requirement for an acceptable selective explanation in biology so consider the various hypotheses that are proposed in evolutionary psychology according to evolutionary psychology the human mind comprises hundreds or thousands of specialized mental modules that evolved in our police to seeing past to deal with specific adaptive problems for example our capacities for sophisticated social interaction are to be explained by mental modules that deal with specific social problems a serious social problem that faces any social group that exhibits altruistic behavior is the possibility of free riders right people taking from the group without contributing their fair share and so evolutionary psychologists have proposed that in order to solve this problem humans evolved what they call a cheetah detection module which allows us to determine when people are not playing by the rules the cheater detection module helps us to ensure that we're not going that we're not going to get swindled anybody who did not have a cheater detection module would have been more likely to lose their resources to free riders so less likely to reproduce and to rear offspring you know that's a plausible explanation of some of our social abilities but before we can accept this explanation we need independent evidence of these cognitive modules and we need we need independent evidence of the cheetah detection module and so evolutionary psychologists have in fact done empirical research intended to establish that humans do in fact have a cheater detection module you might want to check out my my video on evolutionary psychology for more details on this i mean that's a very brief statement of some of the ideas in evolutionary psychology but hopefully you can see the point i'm trying to get at um we have a kind of minimum requirement for a good explanation here i mean i should note that the explanations given by evolutionary psychology are far from conclusive there are many biologists and philosophers who reject the evolutionary psychology program even though they've got independent evidence for these modules um so the point is that evolutionary psychologists do attempt to meet the minimum requirement right they don't merely tell a plausible story along the lines of oh well humans have skill why trait x would explain skill y therefore we must have trait x they they they'll tell that story but that's just a speculative hypothesis they then try to give independent evidence that we do in fact have trait x now it should be clear that the realists appeal to truth to explain success does not me even this minimum requirement instead the fact that truth would explain success is taken to be the evidence for truth that's all the evidence the realist offers my argument is that given a selective view of science the realist can appeal to truth as an explanation for for success only if it is independently established that successful theories are in fact true and of course if it's been independently established that our best theories are true then um yeah appealing to the fact that truth is the best explanation success would be uh redundant as an argument for for realism um so the key point here is that success is no argument for truth even if the realist explanation did not face the many problems outlined in the first part of this video it still wouldn't be acceptable as i mentioned you know anti-realists will want to say that explanation has to come to an end somewhere and i mean one thought is that if science is our ideal of rational inquiry if science is you know the the best example the exemplar of rational inquiry then it at least makes some sense to think that we should apply the kind of standards of scientific explanations to explanations that we give in other contexts and if we apply the standards of selective explanations in science uh to the explanations we give in philosophy of science then i think the realist explanation is very greatly lacking incidentally the analogy to crows allows us to see exactly why the realist explanation is unacceptable and the reason is that selection processes are in a certain sense not discriminating there are many ways of achieving flight many types of structures that a crow wing might have if there is adaptive pressure on a population towards the ability to fly the the members of that population will develop whatever kinds of wings happen to work uh obviously this will be different for different species um some you know for some species many different solutions may be adequate and any one of these could could be chosen anything that gets the job done is acceptable uh that's why we can't make the inference well crows are successful flyers x would make crows successful flyers therefore crows have x because there are many ways of being a successful flyer the same is true within science science is a selection process but it does not select for truth directly it can't possibly do that because science is a human activity and theories theories don't come with a sign saying true there are many ways a theory can achieve success without being true since selection is indiscriminate there's no reason to expect that the true theories are going to be favored indeed many of the selective forces in science are tangential or even opposed to the truth so consider the following first a theory will be selected only if it is understandable to humans part of this possibly depends on our basic biological constitution there may just be some cognitive limits due to how brains encode information it might simply be beyond our biological capacities to understand or conceive of certain things in the same kind of way as it's perhaps beyond the well is it certainly beyond the biological capacity of a dog to understand quantum mechanics at weaker limits may be imposed by the general conceptual background over time in the past many of the concepts of general relativity and quantum mechanics would have seemed utterly incoherent so consider the the rationalist arguments like those of can't for the view that the truths uh of the structure of space could be established a priority we can establish a priori that space has a has a euclidean structure it took a lot of conceptual development before general relativity which proposes a the structure of space is non-euclidean it took a lot of conceptual development before that could have been accepted um so our world view our general worldview provides a lens that colors our evaluation of theories and may constrain what types of theories we can conceive second a theory needs to be coherent with other parts of science so that's going to be an issue if those other parts of science are false before the discovery of thermonuclear fusion the source of the sun's energy was explained as gravitational contraction causing uh compression and therefore rise in temperature of the gases in its interior in the 1860s lord kelvin argued on the basis of this theory that the sun would have a lifetime of tens of millions of years similarly kelvin used the temperature gradient of the earth's crust to argue that the earth must have been molten as few as 20 million years ago these claims caused some pretty serious difficulties for uh geological and biological theories like darwinism that required uh hundreds of millions or billions of years for the age of the earth a similar controversy occurred in the mid-1900s estimates of the age of the earth based on radiocarbon dating gave about uh 3 billion years while estimates of the age of the universe based on the rate of expansion as described by hubble's law it gave only about 1.8 billion years that led to the development of the steady state theory which proposed the continuous creation of new matter and so was uh consistent with expansion for a potentially infinite time of course that was later found to be unnecessary when the estimates based on hubble's law were raised um third different theoretical virtues may pull us in different directions david magnus in his article down the primrose path shows that in early debates concerning the mechanism of speciation biologists in the naturalist tradition held that conciliance and breadth which emphasized gathering evidence from a wide variety of sources was of the utmost importance importance other biologists such as defenders of hugo de vries's mutation theory placed more weight on experimental replication now these virtues are not easily compatible because the pursuit of consilience led to theories of wide explanatory scope that could be applied to a wide range of phenomena whereas the pursuit of experimental control led to theories of much more restricted scope that dealt with very specific processes and according to magnus the seemingly empirical debate concerning whether speciation required isolation really boiled down to these different epistemological virtues so we have these different with different theoretical virtues that pull us in different directions but of course only one theory can be the true one fourth theories will be more likely to develop to develop if the scientists working on them can easily acquire research funding theories are under pressure to demonstrate practical and technological applications kyle stanford in the article unconceived alternatives in conservativism in science has argued that the funding structure of science is likely to lead to increasing theoretical conservativism funding will tend to go to projects that are sure bets projects that are supported by or that try to extend already very well accepted theories with proven track records there's much less incentive to fund radical ideas because i mean radical ideas are just more likely to fail right they don't have that proven track record they're much riskier so they're going to receive less funding and that will mean they have less chance to develop so there is there is selection for increasing conservativism um and obviously that's there's no reason to think that that's going to push us towards truth so the overall point here is the scientific process can never select for truth period it can it can never select for theories that are true of the unobservable world at best it can only select for empirical adequacy for truth about the observable world furthermore many other factors that are selected for are at best tangential to truth um and so this is why if we want to say that the success of a theory is explained by its truth we need independent evidence that it is actually true um in the same way if we say that the crow's ability to fly is explained by some wing structure we need independent evidence that it has that wing structure so um that is the second reason why i'm not a realist all right a third reason why i'm not a realist although this isn't really so much an argument against realism but it's one thing that motivates me that i think makes anti-realism appealing um so again let me begin with a a quote from baz van frassen from preston says if i believe the theory to be true and not just empirically adequate my risk of being shown wrong is exactly the risk that the weaker entailed belief will conflict with actual experience it is but empty strutting and posturing this display of courage not under fire and a vowel of additional resources that cannot feel the pinch of misfortune any earlier so the point is the realist believes everything that the constructive empiricist believes um but the realist also believes additional things that can't face any additional empirical tests this should be obvious because the only way we can confirm or disconfirm a theory is through observation right so a person who takes a theory to be wholly true and the person who takes the theory to be merely empirically adequate will accept or reject the theory under exactly the same circumstances so we could we would say that the the realist version of a theory and the constructive empiricist version of a theory are equally vulnerable where two theories are equally vulnerable when they are disconfirmed by precisely the same observations now van frassen appears to adopt what uh what andre cooper calls a vulnerability criterion of belief van frassen seems to hold that if two theories are equally vulnerable we should believe the theory that is logically weaker the theory that says less about the world and so that favors constructive empiricism because realism and constructive empiricism are are both equally vulnerable um you know it does but but obviously the realist believes much more what the realist believes though the realest believes more but it it can't face any additional empirical tests now i mean it's not immediately obvious why we should adopt this kind of view why we should accept this so-called vulnerability criterion of belief um van fresen himself doesn't really offer that much in the way of a justification i mean we might ask why exactly shouldn't you hold beliefs that go beyond what is observable beyond what can be empirically tested um or on the other hand why even go as far as holding beliefs about what is observable i mean why not restrict your beliefs even further why not uh believe not in the observable but only in what has been observed then you're taking even fewer risks and so you're less likely to be wrong i mean van frastin accuses the realist of displaying courage not under fire but in the context of epistemology why would you display courage at all i could just form beliefs about what i've observed and not bother trying to infer anything about the observable but as yet unobserved but i would but notice that in epistemology we have to perform a kind of balancing act if our goal was simply to minimize the risk of holding false beliefs then we would become radical skeptics and believe absolutely nothing at all or perhaps we would believe just a few propositions that we think are absolutely beyond doubt um but certainly we yeah we shouldn't believe anything substantial about the world now in my view nobody is actually a radical skeptic we all do have beliefs about the world radical skepticism is not a live option psychologically uh it seems to me it's just not it just isn't psychologically possible for anyone to actually be a radical skeptic and even if you even if we could be radical skeptics it would be an absurd position to take it would render us unable to act um so we have other goals than simply minimizing false beliefs and there are broadly speaking two types of of epistemic error right so type one errors consist in believing what is false type two errors consist in failing to believe what is true failing to believe what is true is bad we want to have true beliefs i mean perhaps more importantly than just wanting to have true beliefs we want to act in the world we want we have desires and we want to fulfill those desires in order to act effectively to get what we want we need not just to avoid false beliefs we also need to have true beliefs so both types of error need to be minimized now obviously attempting to avoid one type of error will pull in a different direction than attempting to avoid the other type so applying some sort of rule or method that tends to decrease the false beliefs will also tend to decrease the true ones so this is the balancing act how best to minimize false beliefs and maximize true ones trying to balance these two goals is perhaps the central question of uh epistemology and philosophy of science and the suggestion then is well the vulnerability of criterion of belief strikes a good balance we avoid type one errors through adopting a skeptical attitude towards the unobservable don't hold beliefs about entities and processes that are beyond what can be observed hence beyond what can be tested empirically we avoid type 2 errors by holding the strongest possible beliefs given the vulnerability criterion we need to hold reasonably strong beliefs so that we're able to act so consider the view again that we should believe only what has in fact been observed not just what is observable that would entail a total agnosticism about the future which would be a disastrous position it would leave us utterly crippled we wouldn't know what to do so form beliefs not just about what you observe but about what is observable this allows you to know how to act in order to satisfy our interests we need to know what will happen to us given different courses of action we need to know how the external world will respond to our actions but crucially the only way that things outside our minds can have any effect on us at least any effect that we can be aware of is via our senses the observable world impinges on us the unobservable world can only impinge on us via the observable world so as far as the satisfaction of our desires is concerned the observable world is usually all that matters scientific theories allow us to predict and control the observable world with great accuracy science does this better than any other type of theory than any other type of institution so we should take our best scientific theories as a guide we believe in what those theories tell us about the observable world so that we can figure out what will happen given different types of actions that we might make and that allows us to to satisfy our interests and desires i mean perhaps one way to put all of this is to say no matter what your goals are no matter what your interests are you have to form beliefs about the observable world but that's not the case for uh the unobservable it doesn't actually matter in practice in terms of achieving your goals what your beliefs about the unobservable world are unless of course your goal is simply true beliefs about the unobservable you know if your goal is just to have true beliefs about the unobservable world if that's one of your goals then sure you're gonna have to form beliefs about the unobservable but of course nobody with a more empiricist attitude is likely to share that goal so those of us with a more empiricist attitude can treat scientific theories as tools for predicting and controlling the observable while remaining agnostic about the unobservable we can be we can be skeptical about the unobservable because it actually just doesn't matter what our beliefs in this respect are we can do everything the realist can do in practice i mean i should note as i've said this isn't really intended as an argument against realism in fact i take this to be something of a conciliatory point we're trying to figure out how best to minimize two types of errors where reducing one error will always increase the other there isn't going to be a single objectively right way of doing this it depends on what your goals are and it depends on how much risk you're willing to take so i i'm explaining why i'm not a realist but i'm not trying to refute realism um i'm explaining why i think we're not compelled to be realists anti-realism is a position that can be rational and justifiable there are at least some good reasons to consider it but i i don't know if i would say that anti-realism is rationally compelled either so um those are some of the things that have led me away from from realism um those aren't the only reasons actually but i think those are the main ones and that's all for now thanks for watching