The problem psychology faces is that humans are evolved to be so damn good at it. In fact, we probably have a more extensive and predictive psychological theory than we do an economic one.
We ascribe people a variety of theoretical constructs such as beliefs, attitudes, temporary emotional states like anger, sadness etc and dispositions. These theoretical constructs are incredibly powerful at predicting behavior. We can use the idea of beliefs to predict where an item is hidden based on what they've seen. We are able to pretty informatively guess how recent information (finding out about the death of a loved one or winning the lotto) will affect deciscions. We can even apply the theory to solve inverse problems and infer what kind of news someone must have received based on how they are acting. That's hard for most theories.
The problem is that any result that would improve our ability to predict the outcome of important choices substantially -- much less a general theory -- would offer substantial evolutionary advantage to the holder.
In short, psychology really can't hope to find any theory that's simple, significant and non-obvious. It might find theories that don't help much with individual choices (theories about crowd dynamics or group size or whatever) or only offer kinda unimportant insights (people pick the option on the right when they are identical) but it's fighting evolution to do more.
You could apply this same argument to economics, or any human activity for which advanced theorizing provides an advantage. Perhaps you could even measure the success of those theories by the increase in procreative success they confer to their theory-bearers, ie do economists fare better (eg by getting rich) or do psychologists fare better (eg by psychosexual seduction). But then it seems ridiculously simplified and ignorant of confounding factors to reduce a theory's success to its evolutionary advantage.
Edit: A day later and I'm still thinking about how confidently wrong this is - it fails to withstand even a moment's scrutiny. Humans also have an intuitive understanding of physics, with all the same predictive and inverse power you describe here. It took several hundred years to formalize and justify some of these intuitions, but generally physics has progressed well beyond quotidian human observation.
I'm really amazed that someone could be so enamored with their own bullshit as to think all of this out, type it up, and never second guess the basic premise. Sometimes I really loathe the intellectual arrogance of STEM.
And for an example of a non-obvious psychological theory, see object relations.
Edit 2: Imagine an early human saying this about agricultural theory. I don't mean to attack you personally, but this argument is unfathomably stupid.
And if you don't mean to attack me personally then don't. Maybe you've misunderstood my point which is that we are holding psychology to a different standard because we label the fundamental insights into how the brain behaves at a low level or in unusual conditions (where physics is pretty useful) like what happens when children aren't taught language as something other than normal psychology (neuroscience, some kind of abnormal psychology or development).
Yes, I don't mean claim that it's impossible to improve on anything humans have a good folk theory about. Obviously physics is a case where we have. But my primary point is we actually do have a pretty informative theory of psychology just not one that is better than the folk theory which is a different standard.
And it's not clear to me what is different about physics and theorizing about people is really that we are worse at psychology than that we use a different standard. The equivalent of things like QM and materials science is really neuroscientific claims which we are able to predict quite a bit about. Yes we can look at highly idealized models in physics but if what you care about is something like catching a ball they aren't really much of an improvement. Indeed, it's only relatively recently that we even understood the way people catch baseballs (in the sense that there is a very simple algorithm they use to stay lined up with it that's likely going to do better than someone who is using some Newtonian zero air resistance prediction and standing where that model predicts if they could even get the numbers in the first place).
So it's not clear to me the problem with psychology is so much we are doing worse with it than that we are asking something different. Just like with physics where most of our improvements are about predicting unusual situations beyond our normal range of intuitions we do pretty well in understanding that drugs that block certain dopamine receptors tend to help with skizophrenia.
But psychology does face an extra problem in that the thing being predicted is under evolutionary pressure to be unpredictable. The things we'd really like psychology to help us with are usually cases where if there was a simple predictive theory it would often be to the detriment of the individual being predicted. In other words we should expect that most aspects of normal psychology -- not unusual stuff like skizophrenia -- either are pretty easy to predict using our folk models or are going to require relatively complicated models (same way some problems in physics like turbulence are theoretically predicted but lack any simple approximation we can use).
Evolutionary psychology is a theory of psychology that has a very good and progressively developing track record. I agree with you - much of what came before was disjointed, and didn’t replicate. Evolutionary psychology and behavioral genetics however are grounded in hard biological science, although we aren’t yet seeing much practical application in the clinical setting yet.
The dismal science it is not. I only minored in Econ and did a poli sci major and later law school. I find without the Econ work I did, my thought process and way of looking at problems would be far too meandering and discursive
This is a pretty sloppy error for anyone familiar. They will use it to go "oh, he doesn't even understand what he's criticizing if he confuses these things."
The problem is that while not debuked it's not defined in any way that's simultaneously non-obvious and meaningful.
I mean sure, the claim that it's easier to rember/consider options that have been made salient is true but obvious. Yes, if you have someone read an article about death they are going to increase the relative rate they fill in g_ _ _ e with grave than grace. But that's just what our folk psychology would predict.
OTOH when you try to go further and make the claim non-obvious you don't get much that's both precise, testable and non-obvious.
I think you're mostly talking about social psychology here rather than the many other fields of psychology that I think have much better track records. In a clinical setting, I think CBT has broadly been very successful, and l think lots of cognitive psychological results seem both robust and theory driven.
Can you share the model used to predict results? Your statement is one I have never heard before. Whenever I see psychologist propose a theory it is more like a discussion than a real theory. I would love a link to more about this. All the psychology I have seen looks like literary criticism, not social science. I would love to see the theory and its predictive value you mentioned. I was not even aware that this existed.
This reads like a fish being unable to sense the water in which it swims. Microeconomics is useful because traders have persuaded rulers to set up institutions that allow it to be useful.
How useful is standard microeconomics in explaining potlatches, or hunter-gatherer societies or the Qin empire or the Armenian massacre?
Much of what you mention has little to no connection to economics. Furthermore, one needs data. I worked in finance for the first 14 years of my career, and economics has real value where there is data and competition. It is less useful where there is no data or bad data. It is useless where there are no institutions and some cruel monsters randomly makes decisions.
The key to both understanding economics and applying it often comes down to knowing where the tools apply. Economic theory is pretty useful in explaining the inflation in late Ming China as the Spanish mines in Mexico and Peru flooded the world with silver. We can find price data as well as local accounts of rampant inflation. Once the Qing took over it gets difficult because the new rulers of China were significantly less worldly and accustomed to ruling a large, diverse empire. There do exist aspects of Qing rule where economic theory can be useful (the maintenance of the water infrastructure, for example), but it really comes down to where one can get good data.
But for hunter-gather societies and other examples devoid of data, economics would be the wrong tool. Physics and chemistry would also be the wrong tools. So would psychology, not because it is irrelevant, but because there are no real tools. There should be, but those who practice seem to have a bizarre hatred for logic/math.
This is a misreading of the history of economics and a misunderstanding of science IMO. The main problem with economics is the fantasy that it is just like physics, where there are discoverable, pithy equations that form the foundation of everything. In psychology or economics, any undergrad math or stats major can make up a handful of equations that theoretically shows anything they might want. The left thing we need is a generation of psychologists making up mathy (as in truthy) BS pretending they're doing valuable modeling of human psychology.
Many tools in economics are great at providing real, measurable predictions. There are many industries that depend on economic tools, especially commodity markets, finance, and other fields. Keep in mind that physics had real application problems before computers made calculation and measurement easier. Economics is a new field. It is not a hard science like physics, but it is a real social science. Psychology, it probably could be, but it needs more rigor. The benefit of equations is that they are hard logic. There is a massive benefit to model building. Even thinking in a logical manner has dramatic benefits. Psychology could accomplish so much more if they stopped engaging in literary criticism-style discussion and replaced it with strict tools that obey logic. Probability math allowed for physics to greatly advance in the applied sphere, and a good deal of financial economics piggy-backs on the work of physicists. There is not reason psychology could not make that jump into social science, with real tests and real theory. That is the way fields both grow and deliver real value to humanity.
Of course pattern matching is a real thing. I've got a PhD in statistics and have worked professionally in stats/ML and now AI for 20 years. Lots of economists do solid applied statistics. But all attempts in economics to describe the full economic system (i.e. macro) with "models" is literally a joke. It's all made up. The Fed, to take an example of economics influencing the real world, does short term statistical models and uses heuristics and predictive modeling insights. "Models" in the physics sense of having a set of equations that describes reality have nothing to do with it. They can't or else they'd do a disastrously bad job.
I agree with every word you say. Macro models are pretty much useless. They have a long way to go. But you must understand that the Fed only deal with a tiny slice of economics (monetary policy). I could say to you that physics is useless because in some area our knowledge is weak. That does not make physics useless. It simply means that this specific area requires more work and development.
For the economics that professionals use every day in operations research, financial modeling,..., it is all solid and useful. Monetary policy is extremely complex and we know very little right now, but we will get better. That is how civilization works.
To bring it back to you, I could say your career was a waste because there are no good home butler robots that can take care of babies and clean windows, but that is not all research into AI. Hopefully we will get there, but the absence of a home butler robots does not mean AI is worthless.
This is beside the point I'm making. Nothing that's useful about economics comes out of the let's pretend we're physicists vein of 20th century economics. It's all either finding broad patterns in history (Keynes) or a particular niche of applied statistics and applied math (ops research, auction theory, EA style micro program evaluation, etc.). The physics-pretense version of economics is a total disaster and has had very negative societal effects because it's been such effective propaganda. What I was reacting against was the idea that psychology should imitate this mistake.
Thank you for writing this interesting & thoughtful piece.
I’m sympathetic to this argument, and yet I feel this post may be over-stating its case. For example, economics is increasingly empirical and a-theoretical, and even in theory, things are pretty eclectic. What’s exactly the baseline economic prediction — do you allow for hyperbolic discounting, or diagnostic expectations, or some non-standard type of equilibrium?
That said, in a more modest version, I’m a strong believer that having precise, quantitative mathematical models is extremely useful, for various reasons.
> Our work on how to optimize an auction requires no experimental proof whatsoever – it stands on its own. So too does much of microeconomic theory.
The only models that stand on their own are the models of pure mathematics. Physics doesn’t even stand on its own, physicists and engineers look for empirical evidence of their theories and designs.
And economics is far away from a science, lacking predictive or explanatory power.
I can read it as I have a mathematical degree but unless it’s verified by actual empirical results it’s useless. It may well be mathematically useless anyway, I’ve seen that in economics papers before. However even if the bestest and brightest minds of economics have worked out optimal auctions I’m not sure that exonerates economics in general, since macro economics is clearly a clown show.
Wasn’t there a period post 2008 when, having not predicted another recession, some humility entered the profession?
Predicting recessions would be like physicists connecting the Bohr-Einstein models of the atom. That is some BIG stuff. Where modern economics shines in the small stuff, and the actual practitioners use a whole lot of data. That is what they fit their models to. Economics has real value. It is used all over the place from commodities to insurance to marketing (marketing is only now starting to really use tools).
What psychology would benefit from is a more rigorous, strictly defined method(s) of analysis. Right now too much of it is like discussions in the humanities, and real breakthroughs will only come when the field is able to build stronger theory and tools. The author wants to se psychology thrive instead of endless discussion and little advancement, or worse, actual harm to patients.
I must agree with you here, and it is basically this point that has turned me away from fixating all my energies upon Austrian economics. If all you have is theory, even mathematically fortified theory, it still may say diddly squat about what happens in the real world because some of the concepts may be insufficiently well-defined or the axioms not compliant with reality. Even so, there's the *very hard* problem of proving that you've encompassed all possible mechanisms/structures in your analysis, which is referred to as a classification problem in math.
I come from a math and physics background, so I'll give an example of a classification problem. In linear algebra, you can prove that finite dimensional vector spaces over a given field are completely determined up to isomorphism by their dimension. This seems easy enough, but you don't realize how easy you have it until you consider the matter of modules over a given ring, even if the ring is an integral domain! For context, integral domains would be rings if they only had the extra axiom guaranteeing multiplicative inverses. At least in the case of the ring being a Principal Ideal Domain (PID), you can classify the finitely generated modules over them up to decomposition into cyclic modules (or primary decomposition), but this is a far cry away from having a comprehensive invariant given by dimension alone. Or to give a physics example, that the orbitals for a Hydrogen atom take specifically those shapes and that no others are possible is a result arrived at after a painstaking process of modeling the atom and its attendant forces, plugging in the appropriate expressions to the equations at issue (primarily Schrodinger), and then trying a whole bunch of methods/techniques to find a closed form solution. Even then, if that didn't match up with empirical reality, that would tell us that perhaps the Schrodinger equation was wrong, or our model of the atom, or both. Luckily these seem to hold up, at least within the kinds of energy/scale regimes that we've measured it at.
But to take an example of something which is backed by robust and rigorous theory but fails to match reality, you've got string theory. Even dark matter seems to be mostly discredited at this point, and it was itself hypothesized as a reason why some astronomical phenomenon didn't seem to match up with predictions made by general relativity. Basically, even when you have everything mathematically well-defined, theory is never sufficient on its own unless you want to stay within the math world. Empirics provide support that the theory says something meaningful about the reality we find ourselves in. To OP, just remember that the assumptions upon which neo-classical economics rests do not perfectly conform to human nature and reality, so you are at best working with decent approximations.
If there aren't any theories of economics with genuine predictive power, then the advantage of economics over psychology is merely that it's more up-front about how useless it is. The fact that no one theory of economics has beat out all of the others, or been implemented to create significant economic advantages for parties in the know, suggests to me that economics is as useless as any application of "scientific rigor" to the study of human behavior.
Psychology is largely pre-paradigmatic, and it is because it is much harder to understand the complex causal pathways. It is not for lack of trying, just there are no theories that stand up. I think that the epistemic difficulties you enumerate for psychology are something that may be more significant in degree than the empirical aspects of economics, but ultimately affect all scientific inquiry. There is no such thing as unbiased published research reflective of reality. Facts are theory laden, systems and models have degrees of arbitrariness and reduction, theories are interpretations.
I made a similar argument below. It's not yet clear how to develop theories and models for something as context sensitive as the human mind. Often times the best we can do is to say what "tends" to happen. The average result.
Odd to come across this tonight. Just a few hours ago I told someone the identical thing. Psychology research is on a level with chiropractic. Economics has a long way to go, but at least economist try, except for the likes of Robert Reich.
The problem with psychology is more specific: it has no theory of consciousness. This qualification is crucial. Psychology purports to be normative about consciousness but does not have a theory of consciousness, it therefore always ‘begs the question’. While psychology can demonstrate how certain inputs result in certain outputs, it can theorise about mechanistic or statistical patterns, but it misleads when it asserts what patterns we ought to have, or what outputs we ought to desire. The situation is roughly the same with economics: it has a theory of mechanistic and statistical patterns but cannot not tell us what patterns we ought to pursue, or what kind of value we ought to maximise.
It's worth distinguishing parts of psychology. We have a theory of habit formation and because of that we can make predictions about when people will and will not develop habits.
But we do not *really* have a theory of Risk Aversion (for example). We have some ideas around when it and will not replicate/generalize, but those contextual sensitivities are not encoded into an actual theory or model. Instead we just say, "people tend to be Risk Averse" as if that descriptive statistic (an average) were a theory. It's not, and this causes problems for replication, generalization, and prediction.
It is also worth noting that economics is ok with "as if" models in a way that most psychologists are not. The rational axioms of economics have been falsified, but it largely doesn't matter. The models are accurate enough.
Meanwhile psychologists want to find fundamental truths about how the brain works. This is difficult because the brain is more context sensitive than mathematical models allow. The number of exceptions and context sensitivities which modulate an effect is far too many for us to model accurately. (Hence the reliance on tendencies)
The problem psychology faces is that humans are evolved to be so damn good at it. In fact, we probably have a more extensive and predictive psychological theory than we do an economic one.
We ascribe people a variety of theoretical constructs such as beliefs, attitudes, temporary emotional states like anger, sadness etc and dispositions. These theoretical constructs are incredibly powerful at predicting behavior. We can use the idea of beliefs to predict where an item is hidden based on what they've seen. We are able to pretty informatively guess how recent information (finding out about the death of a loved one or winning the lotto) will affect deciscions. We can even apply the theory to solve inverse problems and infer what kind of news someone must have received based on how they are acting. That's hard for most theories.
The problem is that any result that would improve our ability to predict the outcome of important choices substantially -- much less a general theory -- would offer substantial evolutionary advantage to the holder.
In short, psychology really can't hope to find any theory that's simple, significant and non-obvious. It might find theories that don't help much with individual choices (theories about crowd dynamics or group size or whatever) or only offer kinda unimportant insights (people pick the option on the right when they are identical) but it's fighting evolution to do more.
Interesting argument
You could apply this same argument to economics, or any human activity for which advanced theorizing provides an advantage. Perhaps you could even measure the success of those theories by the increase in procreative success they confer to their theory-bearers, ie do economists fare better (eg by getting rich) or do psychologists fare better (eg by psychosexual seduction). But then it seems ridiculously simplified and ignorant of confounding factors to reduce a theory's success to its evolutionary advantage.
Edit: A day later and I'm still thinking about how confidently wrong this is - it fails to withstand even a moment's scrutiny. Humans also have an intuitive understanding of physics, with all the same predictive and inverse power you describe here. It took several hundred years to formalize and justify some of these intuitions, but generally physics has progressed well beyond quotidian human observation.
I'm really amazed that someone could be so enamored with their own bullshit as to think all of this out, type it up, and never second guess the basic premise. Sometimes I really loathe the intellectual arrogance of STEM.
And for an example of a non-obvious psychological theory, see object relations.
Edit 2: Imagine an early human saying this about agricultural theory. I don't mean to attack you personally, but this argument is unfathomably stupid.
And if you don't mean to attack me personally then don't. Maybe you've misunderstood my point which is that we are holding psychology to a different standard because we label the fundamental insights into how the brain behaves at a low level or in unusual conditions (where physics is pretty useful) like what happens when children aren't taught language as something other than normal psychology (neuroscience, some kind of abnormal psychology or development).
And if you don't mean to attack me personally then don't. You aren't required to call it unfathomably stupid you can just say it's wrong.
Yes, I don't mean claim that it's impossible to improve on anything humans have a good folk theory about. Obviously physics is a case where we have. But my primary point is we actually do have a pretty informative theory of psychology just not one that is better than the folk theory which is a different standard.
And it's not clear to me what is different about physics and theorizing about people is really that we are worse at psychology than that we use a different standard. The equivalent of things like QM and materials science is really neuroscientific claims which we are able to predict quite a bit about. Yes we can look at highly idealized models in physics but if what you care about is something like catching a ball they aren't really much of an improvement. Indeed, it's only relatively recently that we even understood the way people catch baseballs (in the sense that there is a very simple algorithm they use to stay lined up with it that's likely going to do better than someone who is using some Newtonian zero air resistance prediction and standing where that model predicts if they could even get the numbers in the first place).
So it's not clear to me the problem with psychology is so much we are doing worse with it than that we are asking something different. Just like with physics where most of our improvements are about predicting unusual situations beyond our normal range of intuitions we do pretty well in understanding that drugs that block certain dopamine receptors tend to help with skizophrenia.
But psychology does face an extra problem in that the thing being predicted is under evolutionary pressure to be unpredictable. The things we'd really like psychology to help us with are usually cases where if there was a simple predictive theory it would often be to the detriment of the individual being predicted. In other words we should expect that most aspects of normal psychology -- not unusual stuff like skizophrenia -- either are pretty easy to predict using our folk models or are going to require relatively complicated models (same way some problems in physics like turbulence are theoretically predicted but lack any simple approximation we can use).
Evolutionary psychology is a theory of psychology that has a very good and progressively developing track record. I agree with you - much of what came before was disjointed, and didn’t replicate. Evolutionary psychology and behavioral genetics however are grounded in hard biological science, although we aren’t yet seeing much practical application in the clinical setting yet.
The dismal science it is not. I only minored in Econ and did a poli sci major and later law school. I find without the Econ work I did, my thought process and way of looking at problems would be far too meandering and discursive
> Priming, since debunked,
"Priming" is not debunked.
"Behavioral priming" is debunked.
This is a pretty sloppy error for anyone familiar. They will use it to go "oh, he doesn't even understand what he's criticizing if he confuses these things."
The problem is that while not debuked it's not defined in any way that's simultaneously non-obvious and meaningful.
I mean sure, the claim that it's easier to rember/consider options that have been made salient is true but obvious. Yes, if you have someone read an article about death they are going to increase the relative rate they fill in g_ _ _ e with grave than grace. But that's just what our folk psychology would predict.
OTOH when you try to go further and make the claim non-obvious you don't get much that's both precise, testable and non-obvious.
I think you're mostly talking about social psychology here rather than the many other fields of psychology that I think have much better track records. In a clinical setting, I think CBT has broadly been very successful, and l think lots of cognitive psychological results seem both robust and theory driven.
Can you share the model used to predict results? Your statement is one I have never heard before. Whenever I see psychologist propose a theory it is more like a discussion than a real theory. I would love a link to more about this. All the psychology I have seen looks like literary criticism, not social science. I would love to see the theory and its predictive value you mentioned. I was not even aware that this existed.
This reads like a fish being unable to sense the water in which it swims. Microeconomics is useful because traders have persuaded rulers to set up institutions that allow it to be useful.
How useful is standard microeconomics in explaining potlatches, or hunter-gatherer societies or the Qin empire or the Armenian massacre?
Much of what you mention has little to no connection to economics. Furthermore, one needs data. I worked in finance for the first 14 years of my career, and economics has real value where there is data and competition. It is less useful where there is no data or bad data. It is useless where there are no institutions and some cruel monsters randomly makes decisions.
The key to both understanding economics and applying it often comes down to knowing where the tools apply. Economic theory is pretty useful in explaining the inflation in late Ming China as the Spanish mines in Mexico and Peru flooded the world with silver. We can find price data as well as local accounts of rampant inflation. Once the Qing took over it gets difficult because the new rulers of China were significantly less worldly and accustomed to ruling a large, diverse empire. There do exist aspects of Qing rule where economic theory can be useful (the maintenance of the water infrastructure, for example), but it really comes down to where one can get good data.
But for hunter-gather societies and other examples devoid of data, economics would be the wrong tool. Physics and chemistry would also be the wrong tools. So would psychology, not because it is irrelevant, but because there are no real tools. There should be, but those who practice seem to have a bizarre hatred for logic/math.
This is a misreading of the history of economics and a misunderstanding of science IMO. The main problem with economics is the fantasy that it is just like physics, where there are discoverable, pithy equations that form the foundation of everything. In psychology or economics, any undergrad math or stats major can make up a handful of equations that theoretically shows anything they might want. The left thing we need is a generation of psychologists making up mathy (as in truthy) BS pretending they're doing valuable modeling of human psychology.
Many tools in economics are great at providing real, measurable predictions. There are many industries that depend on economic tools, especially commodity markets, finance, and other fields. Keep in mind that physics had real application problems before computers made calculation and measurement easier. Economics is a new field. It is not a hard science like physics, but it is a real social science. Psychology, it probably could be, but it needs more rigor. The benefit of equations is that they are hard logic. There is a massive benefit to model building. Even thinking in a logical manner has dramatic benefits. Psychology could accomplish so much more if they stopped engaging in literary criticism-style discussion and replaced it with strict tools that obey logic. Probability math allowed for physics to greatly advance in the applied sphere, and a good deal of financial economics piggy-backs on the work of physicists. There is not reason psychology could not make that jump into social science, with real tests and real theory. That is the way fields both grow and deliver real value to humanity.
Of course pattern matching is a real thing. I've got a PhD in statistics and have worked professionally in stats/ML and now AI for 20 years. Lots of economists do solid applied statistics. But all attempts in economics to describe the full economic system (i.e. macro) with "models" is literally a joke. It's all made up. The Fed, to take an example of economics influencing the real world, does short term statistical models and uses heuristics and predictive modeling insights. "Models" in the physics sense of having a set of equations that describes reality have nothing to do with it. They can't or else they'd do a disastrously bad job.
I agree with every word you say. Macro models are pretty much useless. They have a long way to go. But you must understand that the Fed only deal with a tiny slice of economics (monetary policy). I could say to you that physics is useless because in some area our knowledge is weak. That does not make physics useless. It simply means that this specific area requires more work and development.
For the economics that professionals use every day in operations research, financial modeling,..., it is all solid and useful. Monetary policy is extremely complex and we know very little right now, but we will get better. That is how civilization works.
To bring it back to you, I could say your career was a waste because there are no good home butler robots that can take care of babies and clean windows, but that is not all research into AI. Hopefully we will get there, but the absence of a home butler robots does not mean AI is worthless.
This is beside the point I'm making. Nothing that's useful about economics comes out of the let's pretend we're physicists vein of 20th century economics. It's all either finding broad patterns in history (Keynes) or a particular niche of applied statistics and applied math (ops research, auction theory, EA style micro program evaluation, etc.). The physics-pretense version of economics is a total disaster and has had very negative societal effects because it's been such effective propaganda. What I was reacting against was the idea that psychology should imitate this mistake.
*last thing we need
Thank you for writing this interesting & thoughtful piece.
I’m sympathetic to this argument, and yet I feel this post may be over-stating its case. For example, economics is increasingly empirical and a-theoretical, and even in theory, things are pretty eclectic. What’s exactly the baseline economic prediction — do you allow for hyperbolic discounting, or diagnostic expectations, or some non-standard type of equilibrium?
That said, in a more modest version, I’m a strong believer that having precise, quantitative mathematical models is extremely useful, for various reasons.
> Our work on how to optimize an auction requires no experimental proof whatsoever – it stands on its own. So too does much of microeconomic theory.
The only models that stand on their own are the models of pure mathematics. Physics doesn’t even stand on its own, physicists and engineers look for empirical evidence of their theories and designs.
And economics is far away from a science, lacking predictive or explanatory power.
https://www.eecs.harvard.edu/cs286r/courses/spring07/papers/myerson.pdf Yeah no, read this and get back to me. If you know the parameters, this is how you do it. There is no better way.
I can read it as I have a mathematical degree but unless it’s verified by actual empirical results it’s useless. It may well be mathematically useless anyway, I’ve seen that in economics papers before. However even if the bestest and brightest minds of economics have worked out optimal auctions I’m not sure that exonerates economics in general, since macro economics is clearly a clown show.
Wasn’t there a period post 2008 when, having not predicted another recession, some humility entered the profession?
Predicting recessions would be like physicists connecting the Bohr-Einstein models of the atom. That is some BIG stuff. Where modern economics shines in the small stuff, and the actual practitioners use a whole lot of data. That is what they fit their models to. Economics has real value. It is used all over the place from commodities to insurance to marketing (marketing is only now starting to really use tools).
What psychology would benefit from is a more rigorous, strictly defined method(s) of analysis. Right now too much of it is like discussions in the humanities, and real breakthroughs will only come when the field is able to build stronger theory and tools. The author wants to se psychology thrive instead of endless discussion and little advancement, or worse, actual harm to patients.
I must agree with you here, and it is basically this point that has turned me away from fixating all my energies upon Austrian economics. If all you have is theory, even mathematically fortified theory, it still may say diddly squat about what happens in the real world because some of the concepts may be insufficiently well-defined or the axioms not compliant with reality. Even so, there's the *very hard* problem of proving that you've encompassed all possible mechanisms/structures in your analysis, which is referred to as a classification problem in math.
I come from a math and physics background, so I'll give an example of a classification problem. In linear algebra, you can prove that finite dimensional vector spaces over a given field are completely determined up to isomorphism by their dimension. This seems easy enough, but you don't realize how easy you have it until you consider the matter of modules over a given ring, even if the ring is an integral domain! For context, integral domains would be rings if they only had the extra axiom guaranteeing multiplicative inverses. At least in the case of the ring being a Principal Ideal Domain (PID), you can classify the finitely generated modules over them up to decomposition into cyclic modules (or primary decomposition), but this is a far cry away from having a comprehensive invariant given by dimension alone. Or to give a physics example, that the orbitals for a Hydrogen atom take specifically those shapes and that no others are possible is a result arrived at after a painstaking process of modeling the atom and its attendant forces, plugging in the appropriate expressions to the equations at issue (primarily Schrodinger), and then trying a whole bunch of methods/techniques to find a closed form solution. Even then, if that didn't match up with empirical reality, that would tell us that perhaps the Schrodinger equation was wrong, or our model of the atom, or both. Luckily these seem to hold up, at least within the kinds of energy/scale regimes that we've measured it at.
But to take an example of something which is backed by robust and rigorous theory but fails to match reality, you've got string theory. Even dark matter seems to be mostly discredited at this point, and it was itself hypothesized as a reason why some astronomical phenomenon didn't seem to match up with predictions made by general relativity. Basically, even when you have everything mathematically well-defined, theory is never sufficient on its own unless you want to stay within the math world. Empirics provide support that the theory says something meaningful about the reality we find ourselves in. To OP, just remember that the assumptions upon which neo-classical economics rests do not perfectly conform to human nature and reality, so you are at best working with decent approximations.
Paul Lutus made this exact point back in 2018, in what amounts to a PhD thesis:
https://arachnoid.com/no_theory_no_science/index.html
Evolutionary psychology.
If there aren't any theories of economics with genuine predictive power, then the advantage of economics over psychology is merely that it's more up-front about how useless it is. The fact that no one theory of economics has beat out all of the others, or been implemented to create significant economic advantages for parties in the know, suggests to me that economics is as useless as any application of "scientific rigor" to the study of human behavior.
Oh, dear. He doesn’t know, does he.
Psychology is largely pre-paradigmatic, and it is because it is much harder to understand the complex causal pathways. It is not for lack of trying, just there are no theories that stand up. I think that the epistemic difficulties you enumerate for psychology are something that may be more significant in degree than the empirical aspects of economics, but ultimately affect all scientific inquiry. There is no such thing as unbiased published research reflective of reality. Facts are theory laden, systems and models have degrees of arbitrariness and reduction, theories are interpretations.
I made a similar argument below. It's not yet clear how to develop theories and models for something as context sensitive as the human mind. Often times the best we can do is to say what "tends" to happen. The average result.
Odd to come across this tonight. Just a few hours ago I told someone the identical thing. Psychology research is on a level with chiropractic. Economics has a long way to go, but at least economist try, except for the likes of Robert Reich.
The problem with psychology is more specific: it has no theory of consciousness. This qualification is crucial. Psychology purports to be normative about consciousness but does not have a theory of consciousness, it therefore always ‘begs the question’. While psychology can demonstrate how certain inputs result in certain outputs, it can theorise about mechanistic or statistical patterns, but it misleads when it asserts what patterns we ought to have, or what outputs we ought to desire. The situation is roughly the same with economics: it has a theory of mechanistic and statistical patterns but cannot not tell us what patterns we ought to pursue, or what kind of value we ought to maximise.
It's worth distinguishing parts of psychology. We have a theory of habit formation and because of that we can make predictions about when people will and will not develop habits.
But we do not *really* have a theory of Risk Aversion (for example). We have some ideas around when it and will not replicate/generalize, but those contextual sensitivities are not encoded into an actual theory or model. Instead we just say, "people tend to be Risk Averse" as if that descriptive statistic (an average) were a theory. It's not, and this causes problems for replication, generalization, and prediction.
It is also worth noting that economics is ok with "as if" models in a way that most psychologists are not. The rational axioms of economics have been falsified, but it largely doesn't matter. The models are accurate enough.
Meanwhile psychologists want to find fundamental truths about how the brain works. This is difficult because the brain is more context sensitive than mathematical models allow. The number of exceptions and context sensitivities which modulate an effect is far too many for us to model accurately. (Hence the reliance on tendencies)