One of the counter-arguments I have encountered several times (commonly from Austrians) is based on an utterly circular premise. (Or as Mark Nelson says, an argument with no loose ends). First you define preferences as being only revealed through actions. In other words, preferences are what you act as though you are trying to achieve. Then you define a rational being as one who acts in the best way to achieve his preferences. (It helps to use these two definitions far apart so that people don't realize that you've created a tautology). Its like if we give you a choice between 3 flavors of ice cream. You choose chocolate. We say "aha, his preferred flavor is chocolate!" Now we check to see if you are rational. Did you choose your preferred flavor? Yes, you chose chocolate.
This results in any being that can choose fitting the definition of rational. Rational becomes just another name for "Makes observable choices". That is a useful concept, but it is very weak. It tells us absolutely nothing about how the being will act as part of a society. It doesn't differentiate people from snails, or even from random number generators. We can always construct an absurd set of preferences to fit any set of observed actions. That does not make them rational.
- Mood:only-partly-exhausted
- Music:DJ Dado-Pure Moods-X-Files Theme
Profile
![[info]](http://l-stat.livejournal.com/img/userinfo.gif?v=92.1)
patrissimo- Patri Friedman
- Patri's World
Page Summary
- michaelduff : (no subject) [+3]
- zuleikhajami : (no subject) [+2]
- creases : (no subject) [+6]
- lovelyslr : Rationality defined. [+9]
- resipisco : (no subject) [+7]
- coldtortuga : (no subject) [+2]
- simonfunk : (no subject) [+3]
- vladkornea : (no subject) [+6]
- triple_entendre : If I'm rational, I'll eat my hat. [+0]
- (Anonymous) : Your definition is wrong [+0]
Comments
Seems kinds odd...not sure why I should be prejudged as a dogmatic lib :).
There is lots of emotion involved in consumer decisions. Its just that rationality is a decent approximation that makes analysis a lot easier. I think one important way in which economics will develop in the future is figuring out how to incorporate models of irrationality. But I don't think its bad to use the approximation for now.
When that world is all that we can imagine or understand, and its somewhat close to this one, fine. But the more we prove the differences, the more it behooves us to find even closer worlds, right?
Its much easier to mathematically analyze polynomials. But lots of functions are not polynomials. At some point you need to start analyzing those functions.
I don't think this is as tautological as you make it out to be. You seem to be ignoring the "best way" in the rational definition. I think it can make sense to say that preferences are revealed through actions. But that doesn't mean that everyone reveals their preferences by acting in the best way to attain the preferences. To twist your example a bit, imagine that the three ice cream flavors are behind glass cases which can be raised from the bottom. The person sticks their face at the glass over the chocolate and licks it, eventually grows frustrated, smashes it and takes the chocolate. Preference revealed? Yep. But the best way to achieve the preference? I think most people would agree no.
It's a silly example and I haven't seen the counterargument in action (or the argument *grin*) to know if its really applied in a useful way, but, at least in theory, I see the "best way" as providing space that keeps the definition from being circular (although then you get into the sticky ground of how do you define "best way" in most situations?)
Suppose we ignore the best way when creating preferences. We just say that whatever it looks like people are trying to achieve (getting at the chocolate ice cream in your example) is their preferences. Then we judge their rationality by whether they choose effective ways to get their wants.
That's totally fine by me, and you are right that it is non-tautological. But then the person in your example is irrational. And the whole point I'm trying to prove is that people are irrational. Hence they don't follow this line of thinking.
Instead, rather than observing your example and deriving "Person wants chocolate, person is irrational", they derive some weird set of preferences/beliefs to make the behavior rational, because they think that's what "preferences revealed by actions" means. And they have a definition which says that humans always act to follow the subjective option which maximizes their subjective happiness. So they must be doing so now.
So they'd say "person likes chocolate, has not encountered glass before, does not know it is transparent, does not know it is sharp when broken". Or "Person likes licking and smashing glass cases more than eating ice cream". If they check rationality, they do it against these preferences - and the person tautologically comes out rational.
Its less obvious when they do it because they tend to do in with examples that they've created.
The difficulty with "best way" is that, as you say, it can be ambiguous. Are their actions an irrational way of getting to a simple goal, or a rational way of getting to a more complicated goal? For my part, I think its ok to say "Well, we can't tell. Lets take more data, and use the model that best fits the facts".
I DEFINITELY agree that we should say "we can't tell, let's get more data" (or make an argument for best fit of available data if getting more data is not an option while recognizing that "best fit" does not equal "only possible explanation") when it's ambiguous.
What you think gets you what you want, doesn't necessarily get you what you want. That's where the economist (or any other scientist) comes in: he's in a position to say, Look, this course of action won't result in what you say you want. For example, raising minimum wage won't improve the standard of living for anyone, it'll only increase unemployment and increase workloads. If that's what you really want, it's not a matter of irrationality, but of sadism. If it's not what you want, then you were confused to pursue that course of action, and in that sense we can say it's "irrational".
Economists tend to assume certain specific interests; people don't necessarily have those interests. If the mutton-heads are in error, it's not in assuming that people are rational, but in assuming they always pursue the interests economists assume for their models -- "economic" interests. Mises was a big critic of this; he insisted that economic values can't be distinguished from social, cultural, spiritual, moral, familial, or aesthetic values. Any economic model that presupposed they could be distinguished, he argued, is flawed from the get-go. All of the above have economic, social, and political effects, and a comprehensive science of human interaction has to suppose that they will play roles in the economy.
Now, when it comes to their ultimate values -- Mises believed there was no basis on which we could judge some to be 'rational' and others to be 'irrational'. That includes whims, cravings, and lapses of long-term self-care for the sake of immediate gratification. Those are issues that individuals need to work out for themselves, and that they should bear responsibility for. People always act as they conceive their interests to lie -- but the way they assess their interests can vary according to a lot of influences, including everything from ideology to immediate temptation.
Yes, this definition is fine by itself, and I agree with it.
But what if you define "what you want" as "Whatever your observed actions are trying to effectively achieve?" Now we have tautology.
If we don't define "what you want" in this way, we need some alternate definition of wants. If we derive wants in some other way than just deriving them from observed actions, then we can apply this definition of rational.
f the mutton-heads are in error, it's not in assuming that people are rational, but in assuming they always pursue the interests economists assume for their models
Its just not true that assuming people are rational is not an error. If you read the literature on rationality, you will see that there is NO consistent set of interests which any person follows effective paths to proceed. People make consistent errors in finding and following effective paths, for *any* reasonable model of their preferences (one which does not include a huge number of strange preferences included to somehow make observed actions fit). People are so far from rational that I don't understand why anyone doesn't agree with it! Try reading some of the literature in the field. Our irrationalities and inefficiencies are pervasive and consistent.
That's not the definition of "what you want". It's the criterion of ends or desires insofar as they are economically relevant.
The tautology is that we always act on our most intense desire. That's not a circular definition; it's an axiom. It's supposed to be self-evident, in that, if someone doesn't act on a desire and acts on another one instead, the one not acted on must be of lower priority.
It's not supposed to permit empirical inquiry into what people actually desire. That's not relevant for economics; the interesting thing is to see how consensus and double coincidence of wants can occur regardless of people's wants. The wants themselves are (by Mises, anyway) held to be relative to the constitution of the acting subject -- ie., subjective, not capable of being empirically uncovered except in retrospect.
If you read the literature on rationality, you will see that there is NO consistent set of interests which any person follows effective paths to proceed. People make consistent errors in finding and following effective paths, for *any* reasonable model of their preferences (one which does not include a huge number of strange preferences included to somehow make observed actions fit).
Oh, I'd agree about that, definitely. (I'm not so familiar on literature on rationality as it is used in mainstream economics; but I know Mises on this point fairly well.) The Austrian principle is that people always pursue the course of action they anticipate to realize their desires, ex ante facto. Ex post, they may evaluate it differently, in terms of what their priorities should have been and in terms of whether the course they pursued really connected rationally with their more general priorities. That's all that's meant by saying that people always act rationally -- they always try. They don't always succeed, when they reflect on what they did later.
The real bugbear that Mises wanted to slay, is when this plain fact is used as justification for another, supposedly more supremely rational philosophos basileus to make people's decisions on their behalf. Ex ante, only one person is in a position to make a final decision about what means contribute to the most intense end, because only one person is privy to the actual intensities.
This seems like a reasonable axiom if changed to "most intense instantaneous desire". But I am then very skeptical of equating these instantaneous desires with true wants or preferences. Because there is lots of evidence which shows that people's desires are different in the instant, change from instant to instant, and change consistently from instant to medium-term to longer-term.
So people may be acting on their most intense immediate desire, but that demonstrates nothing about achieving happiness or long-term goals.
It's not supposed to permit empirical inquiry into what people actually desire. That's not relevant for economics;
It seems very relevant to me! How do you measure whether people are successful at getting what they desire (under different economic systems or policies) unless you have some idea about what it is?
That's all that's meant by saying that people always act rationally -- they always try. They don't always succeed, when they reflect on what they did later.
If they consistently fail in a certain way, how are they "trying to be rational". Isn't it more accurate to say they are trying, and succeeding, at being irrational?
When I see a bowl of chips in front of me, and I eat it, knowing I shouldn't, knowing it doesn't match my long-term goals, wishing I would stop, there is just no way that is rational. Either I'm irrational to be doing it now, or I'm irrational all the times I choose not to do it because its not in my face (ie not putting chips on the grocery list). Its either a net benefit or its not. How this common human behavior be called rational?
Mises was a moral relativist, so the only respect in which this problem would make sense to him, is when it's framed as a question of real long-term interests versus real short-term whims. Mises would say that long-term interests are only of economic relevance insofar as they actually manifest in current behavior; otherwise they're either idle fantasy, or pious pretense. So the conflict here is between immediate gratification and the expected payoff of renunciation.
When it's framed like this, it's clearly an issue of time preference. The agent chooses between a) eating the bowl of chips and staying chub (or getting worse), and b) foregoing them and staying (or getting) trim. The interest rate you pay is measured in pounds, and I don't mean sterling.
Whatever the outcome of that choice, it reflects the balance of desires that existed when the time came to make it. If the long-term is sacrificed for the short-term, it's because the short-term gratification was more intensely valued just then. To call that "irrational" would require referring it to a standard that transcended the person's actual values, and that moves us out of the realm of praxeology and into the realm of what Mises considered vain casuistry and solipsistic projection.
In any event, that kind of consideration has nothing to do with the Austrian conception of "rationality" or the application of that notion in their theories (actually, 'reason' and 'rational action' play a negligible role overall, and I think their theory is stronger for it). It is of personal significance for the individual, but has no bearing on exchange, the phenomenon of price, or political economy generally. So, the definition isn't circular.
Perhaps to you this makes the argument more convincing. I am very skeptical of arguments energetically advanced in order to contradict some looming threat. It seems to me there is a great tendency to be overzealous, to assume ones conclusions, to overgeneralize, all justified by the need to stop some great evil.
I wholeheartedly agree that an extremely minimal government is the way to maximize human happiness. Yet the idea that one person can never make a better decision for another is total bullshit. The truth is that most of the time, very few people could make a better decision. Nor is it easy to tell when these times are, or who these people are. Systems that try to have people control others work very poorly. Since most people tend to make pretty good decisions, the best system is just to let them.
I get to the conclusion through "mosts" and "probablys". So I might actually be right. The Misesian technique of turning "usually" into "always" is, in my opinion, intellectually dishonest. That sort of reasoning tends to get one lost in a fantasy world, rather than consistently reaching accurate conclusions. Shouldn't the latter be the goal of economics?
I play poker for one of my vocations. Time and time again, you see someone faced with the emotional pressure of a big decision. Glands dumping adrenaline into his bloodstream, it is incredibly difficult for him to think rationally. Sometimes it is the case that, WITHOUT EVEN SEEING HIS HAND, all the good players at the table know what the correct play is. In my experience, players are right when they are part of the crowd much more often than when they are in the hot seat. Yet they have the same information. How can you reconcile that with "rationality"?
As for how you can determine whether someone's achieved their goals and interests -- you have to wait until they tell you.
So people should stop using definitions like preferences as being revealed through actions, and actually use their minds to think of preferences on outcomes, which leads to preferences over strategies/choices/actions, find a model (to show that people given 3 choices will choose chocolate), see if the model works... and then test against observations.
Does this make sense?
Btw, I like the new icon.
(retranslating back to my language)
Yes.
Are you suggesting we predict choices by looking at strategy and use the success of the strategy to determine rationality?
Look at strategies when given specific choices. See if the strategy that you define as "rational" allows you to better predict the person's actions (choosing chocolate).
I'm not sure whether that's slightly different from what you're saying.
I think we're saying the same thing??
I'll elaborate more on this with a better example at some point (<-- almost a useless sentence, except that it came from me). There are better examples than choosing ice cream; there are examples that show that rationality helps predict unobservable choices as well as observable ones.
People make different choices when they are asked than when they are actually choosing. Which one is really their preference? I'm not sure. But if the goal of discovering these preferences is to predict the persons actions...then isn't it better to stick with the actions?
The point is that if stated preferences and actual choices differ significantly, then stated preferences are not very useful for building a model whose goal is to predict actual choices. Agreed? Unless you can find some correction, some way to map what people say to what they will do, then it doesn't matter in the slightest *why* people don't do what they say. Their words are still useless for predicting their actions.
But if the goal of discovering these preferences is to predict the persons actions...then isn't it better to stick with the actions?
If you want to throw away some data to simplify your model, sure.
It is true that "stated preference" doesn't map trivially onto "observed actions". But that doesn't mean that there isn't (at least in theory) useful information to be extracted there.
For one thing, it can help us work around the tautology, by providing an angle on the utility function other than "whatever function the person appears to be attempting to maximize".
What do you think? Could people be rational? Could people choose the most efficient action with respect to their goals and the information they have given enough time?
Despite my intuition and all the evidence against total rationality, I still wonder if some bounded form of rationality exists.
My definition of rational is practically explicit here. What's yours? Do you think potential for rationality is a function of intelligence? If so, would you agree to a definition of rational bounded by the person's intelligence?
Do you think there are cases in which the most rational decision for a person cannot be determined by somebody other than the person? Do you think there are cases in which the most rational decision for a person can only be determined by somebody other than the person? (These are probably obvious questions to you, but I don't know your definition of rational, so I thought I'd ask straight up as an exploration.)
I think the degree of rationality involved, called upon in a situation, depends on the weight of your decision. Deciding to go to bed doesn't really call upon making a rational choice. If someone changes their mind given no new information then their decision must not have required much thought.
So rationality is more useful when looking at decisions like strategies in games, wars, business, etc. where people actually have to consider the consequences of their actions.
So I guess I'm saying that rationality is preferences-over-consequences dependent.
I can see that being the case for a J, but what about us P's? The big life-altering decisions are fence-straddlers.
(I had a more thought out response before my system overloaded and crashed due to mass trojan infection earlier today.)
one more reason I just don't get p's...
terrible.
so big life-altering decisions aren't quite what i'm getting at here... there's hardly a rationality for something like marriage, or going to college in one place or another, because those depend on emotional context... even ordering food at a restaurant is feely-based...
but I'm sure that when you played Texas hold'em you tried to either
a. win,
b. learn more.
My definition for rationality is to use available information so as to maximize a utility function (a composite of different preferences, including discount factor, risk aversion, etc.).
Intelligence helps rationality, but it is not enough. We are not just irrational because our brains are too slow, we are irrational because the fundamental structure of our brains is irrational. Its not a general purpose computer, its a mish-mash of modules with bits of ape and lizard brains all over the place.
I think (contrary to libertarian philosophy) that in some situations, other people can better determine what is rational. For example, ones genes may exert a strong influence that does not make you happy (let's not use a condom, or let's eat that whole bowl of chips). Another person, not feeling that influence, could decide better.
Why are you not including the desires of one's genes in the utility function? Are they not a part of oneself? Is your utility function a composite of only conscious preferences?
Suppose that while holding a lottery ticket, I masturbate into a cup and die on climax. The lottery ticket turns out to be a winner, and the sperm are used to artificially inseminate thousands of women, all of whom share in the lottery proceeds to help raise the children. That would be a stunning, epic victory for my genes.
Yet I'd get much more pleasure out of directly inseminating a few women and directly raising a few children on a non-lottery budget. Wouldn't you?
He doesn't go into specific implementations, but they are easy to imagine: you could agree pay a fine if you transgress, etc. hey even physical punishment might be a good idea.
One problem is that you might actually change your mind for good reasons. But I, for one, am confident that my habits for procrastination and chocolate are (and will always be) bad for me. Although I don't think my procrastination would respond well to threats of punishment. Hm... perhaps if I thought of it as winning a prize...
Then we (the software guys) write down the problem formulation and note that it's NP-hard or NP-complete, and takes 47 hours to solve to optimality on the most fantastically zippy computer sitting around in the office. Also, the answer is "obviously stupid".
My suspicion is that, even if people knew what they wanted, and their multitude of desires did not conflict in any way, and they had perfect information about the state of the world and the sequences of consequences of their potential actions, it would still be impossible for any mechanical computational process to discover the "rational" approach to getting what they wanted. Hence either people act "irrationally" (more to the point, short-sightedly and often incorrectly), or people are somehow waaaaaaay smart.
There are situations in which people consistently do a worse job than very simple mechanical computation processes.
One example is medical diagnosis. They did a study where they took lots of medical diagnoses from experts, and derived a simple linear model with coefficients on each observation. This model was more accurate at diagnosis than the experts. While the experts did fairly well on average, they had a lot of arbitrary links with things like "A patient who wore this color, or had this one symptom, had this disease, so I'll tend to expect future patients wearing that color or with that one symptom to have it". Apparently hese links were rare enough that the average correlation was a good guide, but common enough that each doctor was off.
Several other studies have shown that embarassingly simple models perform better than experts in situations like these.
The implications are exceedingly interesting. For example your recent problems with getting allergy medication prescriptions basically comes down to experts not being at all expert.
http://hypatia.ss.uci.edu/lps/psa2k/eth
http://hypatia.ss.uci.edu/lps/psa2k/fif
http://www.blackwell-synergy.com/li
And it leads me to wonder: How does this affect our evaluation of political models? E.g., even skipping over the difficult question of choosing the goal(s) for such models, does the recognition of people's irrationality lend some credence to paternalistic governing? Conversely, how much of human irrationality is a consequence of interactions between modern contexts and antiquated evolutionary strategies? I.e., perhaps certain societal/legal structures "bring out the irrationality in people" (say by removing some natural feedback cycles).
Has anyone made a good attempt at designing a political philosophy begining with rational goals but based on a realistic model of human psychology? I've seen (mostly) rational goals pursued assuming rational participants (O-ism and kin), and I've seen irrational goals pursued...
The answer to your first question is that if the irrationality is not regular then you can still work on the assumtion of rationality and still come up with more or less realistic models. For instance, if 50% of stock market brokers act rationally, 25% overvalue stocks and 25% undervalue stocks, then a model of rational participants may still be able to explain the real world.
To your second question, about the viability of paternalistic governing -- the answer is still no, paternalism is nonsensical. A person does not need to be rational to make rational choices. As long as he _knows_ that he's irrational in a certain matter, he can always ask for advice from someone who can be judged to be rational.
As you say, this knowledge of his own irrationality can come from feedback cycles from the real world.
And I do think much of human irrationality is the modern/evolved environment problem. But some of it is also that we have a lot of hardwired shortcut modules, and even in the right environment shortcuts are sometimes wrong.
Now personally, I am wholly convinced of human irrationality, and still think paternalistic governing is a pretty bad idea. But I have the advantage that much of my reasoning is empirical, and so human irrationality is just totally irrelevant. Its much more of a problem for the axiomatic folk who start with human rationality and then conclusively prove that government is a bad idea. Yet another reason not to take that approach :).
I am at least willing to admit that people are irrational, and that this weakens libertarian arguments. I think its intellectually dishonest to avoid a true facet of life because it weakens ones philosophy. But I don't think its anywhere near a crushing blow. And I think that just as with any other system, we are more likely to design a good libertarian system if we are honest about the agents that will interact in it.
I'd appreciate it if you could demonstrate this. I do not mean point out a libertarian who stated that people are rational, but articulate the actual argument for libertarianism which is based on the premise that all people are rational all of the time.
Almost all of economics is based on the assumption that people are rational. As is the libertarian political system.
Nor do I disagree with this basis. Its a good first approximation. I just get annoyed when people mistake it for the truth. Or when they axiomitize their assumptions, then get surprised when people disagree. "How can you disagree - its an axiom!".
I read Mises' Socialism and much of Human Action, and I'm not sure what you're referring to. The first sentence of Human Action is "Human action is purposeful behavior." Mises continues:
Rand defined man as the rational animal. This definition serves to best differentiate man from all the other animals, in the context of our knowledge. It does not mean that all men are rational all of the time (and Rand obviously didn't believe that).
This may be just my ignorance, but I honestly have no clue what you mean by "Almost all of economics is based on the assumption that people are rational. As is the libertarian political system." My political and economic views largely coincide with Mises' and Rand's, and I have never seen either of them state anything that can reasonably be construed as "all men are rational all the time."
I was wrong about it being the first sentence of the book. it is the first sentence of Chapter 1, section 4 of Human Action.
As for man being "the" rational animal:
Rand defined man as the rational animal. This definition serves to best differentiate man from all the other animals, in the context of our knowledge. It does not mean that all men are rational all of the time (and Rand obviously didn't believe that).
If men are not rational all the time, then how are they qualitatively different from animals, as opposed to just quantitatively different? Surely a chimpanzee, in some situations, displays the same level of clever planning to maximally achieve its goals as a person [example]. Just not quite as often. Therefore its a spectrum.
To me, it seems meaningless to define man as "the rational animal". If we have good observations to demonstrate this conjecture, great. If you want to argue for why you think its true, fine. But as soon as you define it, making it an axiom, now you have totally closed the book on further evidence and debate. Its an effective technique of Rand's, but I find it intellectually pusillanimous. I have a math degree, and I find axiomatic logic beautiful - but I think it is dangerous and wrong to try to apply it to the real world. I have no axioms - I'm willing to question everything I believe.
I always find it a bit strange to hear athiests argue that homo sapiens is uniquely endowed with some special metaphysical quality which makes them qualitatively different than animals. Sounds just like the argument for a soul to me - and just as egocentric.
This may be just my ignorance, but I honestly have no clue what you mean by "Almost all of economics is based on the assumption that people are rational. As is the libertarian political system." My political and economic views largely coincide with Mises' and Rand's, and I have never seen either of them state anything that can reasonably be construed as "all men are rational all the time."
Perhaps your exposure has been more to non-traditional forms of economics such as the Austrian school. In neoclassical economics (which is the economic mainstream), the most common model used for the basis of analysis is one which people are rational all the time. Or at least their irrationalities are unpredictable and random, so tend to cancel out. There is certainly lots of work diverging from this, but it is the status quo in the field. I usually see it explicitly stated in economics books. For example, the Wikipedia lists it as one of the "Key assumptions of neo-classical economics which are widely criticised as unrealistic". Here is another example, from the online Essential Principles of Economics. Note that rationality is described as "the basis of neoclassical economics".
This is split into multiple parts because I hit the 4300-character limit.
Further reading shows that Mises uses the term differently than you're using it now.
Mises' definition of "rational" is the presumably then-common "relating to reason; not physical; mental". I read all this quite a while ago, and I see now why I didn't remember Mises claiming that all men are rational all the time in the sense you meant it--I understood his thoughts in context.
You're apparently not familiar with Rand's thoughts on the nature of concepts and definitions. She actually uses her definition of man as an example (Introduction to Objectivist Epistemology - 5. Definitions; first paragraph below mangled by me):
So much for Rand's alleged claim that all men are rational all the time.
I'm not sure if it comes across, but I'm agreeing with your post. ;)
Lots of good discussion in these threads; thanks for that. ("Make the ganglia twitch!")
Greetings, Patri:
I think you try to define "rational" as "the best use of the mind" or "of perceived choices". Ayn Rand surely tries to use it that way. But Austrians do not. Human action is not mindless reaction or gut impulse, it is purposeful.
But you need to separate the two concepts that you may attach to "rationality".
The Randian (as I undestand it) one is "a rational person makes good choices"
The Austrian one is "People act purposefully".
As you can see, the Randian definition is more akin to what you were discussing.
Best regards,
Juan Fernando Carpio
juanfer76@yahoo.com