Thursday, January 31, 2013

This will probably only interest two to three people :)

This is a discussion from a little while back on facebook with Daniil Gorbatenko about praxeology, preferences, choice, etc. I thought it was interesting at the time and meant to post it but forgot until Daniil reminded me to recently.

It kind of picks up in the middle of the conversation (although not too far in) - we were getting into the weeds on a post on Steve Horwtiz's wall and decided to take it to messaging. Daniil strikes me as making very strong claims here. I'm not sure all Austrians agree with this take on praxeology, but I think there are good arguments against it. There's also simply a disagreement over useful modeling assumptions. There are a lot of wrong abstractions we make that have a big payoff and relatively little cost. The retort "but that's not true" has very little bite in this case, unless you can show that the cost that comes with it not being true is bigger than I think it is.

Daniil summed up his position in discussion in this post. I will do no summing up, but you can get a sense of the concerns I have about his position and about the strengths of the mainstream viewpoint.

*****

DK: People have preferences before an exchange happens. Those preferences are consistent with a whole set of possible exchanges. Obviously only one exchange happens. But it's not wrong to say at all that a person with free will and preferences has a set of possible exchanges and it's not wrong to talk about bounds to that set. Now, to talk about functions we have to make some assumptions about those preferences. But the assumptions aren't all that strenuous and nothing is really lost in making them for analytic purposes.

DG: re: "People have preferences before an exchange happens. Those preferences are consistent with a whole set of possible exchanges."

No, we don't have the grounds to say that. It's an unjustified psychological assumption that mainstream economists use to be able to use calculus.

DK: No, we do have grounds to say that. I prefer coffee to tea. That will inform exchanges I make. The assumptions we make are around the rationality of those preferences - transitivity, completeness. The first assumption there is a reasonable one. The second one probably isn't true but its being wrong doesn't carry much cost and it provides a lot of analytic benefit. THAT is for the calculus. The idea that humans have preferences is very different, and we do have grounds for that.

DG: From the praxeological standpoint you only have preference when it is reflected in the particular exchange. Otherwise you may have preferences in the psychological sense about many things, though I doubt that you may have in your mind at one moment a systematics set of preferences about all the goods. Introspectively, I'm sure this is not the case. But anyway, from the praxeological standpoint it's irrelevant. For a more detailed exposition of this point I recommend chapter 9 of this text http://mises.org/journals/scholar/long.pdf

DK: Well if you want to redefine the word that's one thing. But then if you redefine the word you can't very well use that definition to dispute the definition used by mainstream economists! I can do that. You say exchanges exist. I redefine exchanges to mean "a round square". I insist to you that exchanges don't actually exist. In fact from my perspective they are a logical impossibility. That doesn't get us anywhere, now does it?

DG: Definitions matter here because free will implies that there is no causal connection between psychology and choices. That means that psychological preferences may not explain exchanges the way you want them to.

DK: re: "though I doubt that you may have in your mind at one moment a systematics set of preferences about all the goods"

In my conscious mind? No. In my brain? I carry around a lot of preferences that I am not consciously holding at any given moment. For example, I had to think a second or two to come up with my coffee vs. tea example. Now - as to the point about preferences over "all" goods, this is an assumption that is made for mathematical reasons. But I can't conceive of how making this assumption causes any real problems.

re: "Definitions matter here because free will implies that there is no causal connection between psychology and choices."

No, it doesn't.

DG: re: "In my conscious mind? No. In my brain? I carry around a lot of preferences that I am not consciously holding at any given moment. For example, I had to think a second or two to come up with my coffee vs. tea example."

It doesn't at all imply that you'll necessarily choose in accordance with these preferences which returns us to the free will point. Free will implies that there is no sufficient cause of your particular choice.

DK: re: "It does. Free will implies that there is no sufficient cause of your particular choice."

That sounds more like something like "stochastic will" than "free will". But I'll work with that definition. Again, mainstream economists make rational choice assumptions so the math works. If you want to argue that people will always act exactly according to their preferences is a rational choice simplification that seems right to me. But it still seems like a very reasonable simplification to make for the math. Certainly it doesn't challenge the point that preferences exist prior to exchange and that they inform the exchange.

DG: re: "That sounds more like "stochastic will" than "free will". But I'll work with that definition."

Well, this aspect free choices share with stochastic processes.

re: "If you want to argue that people will always act exactly according to their preferences is a rational choice simplification that seems right to me."

Just for clarification, Austrians (at least of the persuasion that I adhere to) do not believe in our being able and are not trying to, explain human choices in terms of their prior psychological preferences. They (we) say nothing about that. And we believe the attempts of mainstream economists to psychologize economics are mistaken precisely because of the free will property. Again, free will implies that there is no remotely precise mapping from psychological preferences at some point in time (even if they exist without your thinking of them which is not at all clear) to choices at another point in time.

DK: So you don't just think that psychological preferences sufficiently explain choice. You actually think they are a bad predictor of choice. That is interesting. On what grounds do you make that claim? That seems like a very strong claim indeed. Not sufficiently explain I would agree to and I would say that we argue they sufficiently explain it for the math. But to say that it's "not remotely precise" seems absurd.

DG: re: "That is interesting. On what grounds do you make that claim? That seems like a very strong claim indeed."

Well, because free choices are not outcomes of a stochastic process. Thus, the talk about predictors is misplaced.

DK: So you believe its true because you assert its true. I see. Hmmm...

Let me clarify. I believe that psychological preferences are very close causal predictors of choices because I prefer coffee to tea and I buy a lot more coffee than tea. I prefer jeans to khakis and I buy a lot more jeans. I prefer bacon to sausage and I buy a lot more bacon. It seems to me that these psychological preferences I have right now, sitting down, not confronted with any choices are fairly stable and have a lot to do with the choices I do make. That's my reason. Can you tell me your reason other than just saying "because its true"?

DG: My reason is twofold. First. I know from introspection that I made many choices differently from what I had thought I preferred. Second, I don't think you have psychological preferences unless you conceptualize them. And due to time limits and distractions to other things you're only able to conceptualize very little of your psychological preferences, anyway. Conceptualize prior to choice situations.

DK: re: "First. I know from introspection that I made many choices differently from what I had thought I preferred."

This seems very atypical. I mean, everyone has impulse purchases. We can file that phenomenon under "maybe choices aren't perfectly rational". But do you think this is typical?

re: "Second, I don't think you have psychological preferences unless you conceptualize them"

Then why is my "coffee preferred to tea" preference so persistent between the times that I do and don't make the effort of conceptualizing it? The consistency seems to suggest that these preferences exist in our brains even if we're not conceptualizing them at any particular moment.

DG: re: "This seems very atypical. I mean, everyone has impulse purchases. We can file that phenomenon under "maybe choices aren't perfectly rational". But do you think this is typical?"

Yes, I believe that this is more typical than the reverse. Most of our choices are spontaneous, in my view.

re: "Then why is my "coffee preferred to tea" preference so persistent between the times that I do and don't make the effort of conceptualizing it? The consistency seems to suggest that these preferences exist in our brains even if we're not conceptualizing them at any particular moment."

I'm afraid you're here conflating the praxeological preference that you infer from your choices of coffee over tea with psychological preference. And that's why you are probably so sure that you are always acting in accordance with your psychological preferences. You infer them largely from your actual choices which is a mistake.

DK: On psychological/praxeological preferences - so you are claiming that my preference for coffee over tea comes up repeatedly simply as a matter of chance - that there is nothing more fundamental in my brain driving that (even though that preference is not always conscious for me). What you call praxeological preference I call "revealed preference", and it seems to me there if very good reason to believe that those revealed preferences are grounded in pre-existing preferences.

DG: re: "On psychological/praxeological preferences - so you are claiming that my preference for coffee over tea comes up repeatedly simply as a matter of chance"

No. It comes as a result of evaluative process that is neither stochastic nor determined by previous mental or brain states.

re: "What you call praxeological preference I call "revealed preference", and it seems to me there if very good reason to believe that those revealed preferences are grounded in pre-existing preferences."

'Grounded' is a loose word that can mean many things. I'm not saying that choices are not at all related to psychological preferences. What I'm claiming is they are neither strongly, nor loosely determined by them (there is no probability distribution of choices derivable from psychological preferences).

16 comments:

  1. Daniil Gorbatenko isn't really the view of the rest of us.

    Daniil seems to think of things this way so he can find a space for free-will. That's not the reasons others have done it, the traditional reason was to avoid the tarpit of psychology. Mises (for example) wasn't averse to the idea that human behaviour is totally deterministic at it's final reductionist root in physics and chemistry. His point was that investigating it was far too complex, he judged that at his time of writing that it was better to mostly ignore the problem and take preferences as givens.

    I generally agree with you that there's nothing much wrong with assuming some preferences and assuming they exist over a reasonably long period of time. As a modelling assumption it's reasonable, and the kind of thing many Austrian Economists have done. We have to always remember that they're assumptions though, and may be violated in some cases.

    What kind of calculus applications of this idea are you thinking off? I think a lot of the uses of preferences that people talk about don't require as much psychology as they think.

    ReplyDelete
    Replies
    1. Current, in contrast to you I didn't pretend to speak on behalf of all Austrians. I think your pretense to speak on behalf of all Austrians, except me, is as unjustified.

      Delete
    2. Fair Enough. My first line was presumptuous.

      Can you expand on what you wrote though?

      To a degree I agree with the view you describe about economics. I think that assumptions that are made about psychology are really assumptions about demand in disguise. I don't agree with you about free will, I'm a compatabilist about that. Nor do I think that psychology is completely intractable, though I think it's very difficult.

      Like Mises I want to separate my views from psychology as much as I can for the sake of simplicity. But, like Mises I don't want to abandon reasonable assumptions about the future. All through his work Mises assumed the kind of thing Daniel was discussing above. He assumed that demand for coffee at time X is at least related to demand for it at time X+1. You can view this as an assumption about psychology or construct it differently, though it can be difficult to construct it differently.

      Do you think that Praxeology explains time-preference?

      Delete
    3. =Can you expand on what you wrote though?=

      Could you list the questions you'd like me to answer?

      =All through his work Mises assumed the kind of thing Daniel was discussing above. He assumed that demand for coffee at time X is at least related to demand for it at time X+1.=

      But by demand, I think, Mises meant preferences deducible from choices, not the psychological preferences that Daniel is fond of.

      =Do you think that Praxeology explains time-preference?=

      I never deeply thought about this question before but I think time preference is, contra Mises, in a sense inherent in the notion of choice. When you choose an alternative you aim at a certain element of the future state of the world that you imagine (the goal). Time that has to elapse until that element is there is not an object of your choice, unless you need the element to be there at a certain moment in the future. It is something that stands between you and your goal. Thus, choosing a shorter time period from the choice to the goal is a means to attain the goal in a better way which must be clear to any choosing agent. And the Austrian definition of reationality implies that the agent will choose the means that, to his judgment, best suit the attainment of the goal. Thus, the agent will exhibit time preference.

      Delete
  2. What in the difference between a choice and a preference? I am going to say that a choice is observed behavior. One alternative is realized, others are not. If they are the same thing, then we have nothing to talk about. ;) I am for the moment going to say that a preference is not choice, but is related. We might say that preference is vicarious choice. If you ask me whether I prefer chocolate or vanilla ice cream, I may imagine choosing between them and report the vicarious choice. But there is something else going on. If my vicarious choice is always the same as my actual choice, then, unless we are interested in vicarious behavior, we do not need to talk about preferences at all. But in fact, even if I prefer vanilla to chocolate, I may at times choose chocolate over vanilla. So preference is **not** vicarious choice, but is an abstraction of choice. It may have to do with frequency of choice, prototypical choice, default choice, any or all of the above, or something else.

    Preference is not a given. There is no reason for preference to mean the same thing to economists that it means to cognitive scientists. Or even to mean the same thing in differing theories. It all depends on how things hang together, on the significance of preference and what role it plays vis-a-vis the rest of the concepts in the theory.

    ReplyDelete
  3. I think free will is a post decision rationalization of intuitive choices based on subconscious preferences. Preferences are circumstance and time sensitive, multidimensional and often contradictory, the resolution to a specific choice is largely instinctual, and the only evidence of any conscious decision is when we are so indifferent to alternatives that we have difficulty deciding.

    ReplyDelete
  4. As you know, I do a lot of research on stereotypes (cognitive), prejudice (attitudinal), and discrimination (I know we have different definitions, but typically, behavioral). Decades of research on these topics consistently reveals a weak link between prejudice and discrimination -- attitudes do not strongly predict behavior. Why? Well, because there is a lot that happens between attitudes and behavior, including other attitudes (a desire not to appear prejudiced) and external constraints (situations and the actions of other people). This discussion is fascinating to me because it highlight one of the (apparent -- this is a limited N conclusion so might be biased) differences between psychology and economics. Researchers in psychology live in a messy, messy world, and almost all of our findings concern weak relationships and moderators of those weak relationships -- and we're increasingly trying to incorporate both internal (attitudes) and external (situations) variables/constructs to better understand why these messy links exist and how they work. It's just interesting to see how people with other backgrounds deal so differently with similar topics.

    ReplyDelete
    Replies
    1. Dr. J, the whole point of my position is that I don't think that economists should deal with the topics you mentioned, should not study the connections between attitudes and choices.

      And your point about the messiness of connection between prejudice and discrimination is actually a beautiful demonstration of what I was trying to explain to Daniel - that there is no one-to-one mapping between psychological preferences (you may call them 'attitudes' if you wish) and actual choices.

      In other words, I believe that choice is not a psychological phenomenon, that it is, if you will, a higher-level phenomenon that, in the language of the modern philosophy, supervenes on the psychological level and differentiates us, humans, from other animals.

      There are of course connections between the two levels but they are neither logical, nor mathematical.

      Delete
    2. citizen-global -

      Just so we're clear - the one-to-one mapping is a modeling convention, everyone knows its a modeling convention, and for the class of choices we're talking about it's quite good.

      You have to remember that this is just the motivating model too. Nobody actually observes preferences. Any empirical work is going to have stochastic element to it.

      This all brings me back to my initial point: the explanatory power gained by the modeling convention dwarfs any associated costs. So what argument is there really for abandoning it?

      If we didn't work with models that abstract from reality to understand reality we wouldn't have a very good understanding of reality.

      Delete
    3. Dr. J -
      re: "Why? Well, because there is a lot that happens between attitudes and behavior, including other attitudes (a desire not to appear prejudiced) and external constraints (situations and the actions of other people)."

      Obviously we're talking about a somewhat different choices here, but I think economists actually do rather well thinking about the issues that you've raised. So we would say that people have two preferences in this case - a preference for a certain amount of discrimination and a preference not to appear discriminating. They're going to try to optimize the various trade-offs between those two preferences (because they're going to work against each other - so you have to trade-off between them), subject to constraints.

      And we actually call them "constraints" too. Usually we talk about budget constraints, but we get into other things as well. The framework for thinking about this is called "constrained optimization".

      So if it's just a matter of adding other competing preferences into the decision making process, I don't think that presents a big obstacle.

      It gets trickier if people are consistently non-rational (and they can be - granted there's a lot of economists working on that too), or if there's a degree of randomness so that the noise swamps the signal (which I gather is more where citizen-global is coming from).

      Delete
    4. Re: "I think economists actually do rather well thinking about the issues that you've raised"

      I did not intend to imply that economists don't do well in thinking about these issues-- my only goal in posting was to share an observation I've had about the differences in how different sciences look at similar issues. If anything, the fact that we look at similar things from slightly different angles (and using slightly different language and techniques) can be a major strength. It's the elephant analogy -- maybe psychologists got hold of the tail and economists got hold of the leg, and so forth.

      Speaking of interdisciplinary work, when are we ever going to get together and do our econ-psych crossover brainstorming?

      Delete
    5. Yep - I didn't think you were implying it. Just relating how I would think through the points you made.

      We do need to do our brainstorming... what are your good days? Are they the same as the fall?

      Delete
    6. Daniel, you wrote:

      =If we didn't work with models that abstract from reality to understand reality we wouldn't have a very good understanding of reality.=

      We are going into deep philosophical waters here but it seems inevitable.

      There are two types of abstraction, non-precisive (usually called 'modeling') and precisive (usually called 'abstraction'). The former distorts the reality it describes, the latter just abstracts from the individual features of the described objects without distorting anything.

      In the example of human beings the Austrian theory relies on precisive abstraction. It states the obvious about all human beings abstracting from their psychological differences and adding some more obvious facts about the world in which the human beings live, deduces certain necessarily true implications using praxeological categories. Later on, it is possible to look at a certain historical episode and, depending on which praxeological categories are present there, say what happened in such episode. For example, if an episode was characterized by credit expansion (excess of investment expenditure over savings through bank credit), then it is possible to say about such an episode that more investment projects failed in it than would have been the case without credit expansion. In such explanation the reality is preserved intact.

      What mainstream economists do is non-precisive abstraction. Implying that human choices are just revelations of underlying structures of preferences with a stochastic element involved they try to approximately reconstruct the behavior of certain variables in the real economy abstracting in the process from the fact that what they think of as variables aren't such, that say every price is a unique fact, not a value of certain variable. Such reconstructions can have a connection with reality only by chance.

      The upshot is that the mainstream models are mathematical descriptions of non-existing worlds, not descriptions of economic reality which does not contain mathematical concepts of the sort used in mainstream models while it obviously contains the concepts with which the Austrian theory operates.

      Delete
  5. So much to write about and so little time....

    Praxeology is one of those things that everyone adapts to their own liking. I don't think that citizen-global's praxelogy is the same as Mises' praxeology. I don't think mine is either but I think it's closer.

    Citizen-global, you seem to think of thought as a combination of "psychological" factors coupled with a mysterious element of "free-will". You then criticize mainstream economists for ignoring the free-will and concentrating on the psychology. I don't agree with this breakdown for two reasons.

    Firstly, I don't think free-will is separable from psychology or from the physical world. I'm a materialist, I see humans as being not that mysterious, just complicated. This viewpoint doesn't deny free-will at all, rather free-will is the action of a material being responding to its environment. Materialist explanations that deny free-will do so by confusing the issue. They define processes that take place within the brain (for example) as something separate from the material creature being examined, which is nonsense, they are part of it and those processes (be they deterministic, random or pseudo-random) are part of that creature's free-will. But, that's all by the by, and tangential to the discussion.

    Secondly, and more importantly, I think your theory *is a psychological theory* and as such is exactly the kind of thing I want to avoid. I think we can talk about economics without getting into abtruse philosophical discussion about compatibilism and other views on free-will, and without getting into equally abtruse discussions about psychology.

    I think part of the confusion here is that lots of the modern work on psychology that gets reported in the popular press is about irrational behaviour, rationalizing behavioir, about the unconcious and about instincts. But that doesn't mean psychology is all about that. Traditionally it was just as much about rational behaviour as the rest. You seem to be criticizing undue emphasis on this in economics, I don't think that's really fair. Recently it's made a big splash, but over a longer period of time it hasn't been that important to any school of economics.

    I don't agree that there is a difference between examining many praxeological choices and doing psychology. I agree that when a choice is actually made there's more at work than simply the pre-existing preferences of the chooser. There's new knowledge for a start. For example, if it was shown that coffee is very bad for health, but tea isn't then Daniel may decide to by coffee rather than tea. What are we doing when we taking a whole series of choices as an indicator of future choice? You could say that that isn't psychology, I wouldn't agree about that, I'd say it's a very simple form of psychology. It's certainly a borderline case.

    I don't think that ABCT really evade psychology. Rather, it depends on certain simple forms of psychology, like those I mention in the paragraph above. But, I agree that there are some theories that are precisive and some that are less so. Though I don't think there's any economic idea that is totally independent of empirical evidence. One of the closest things we have though is the theory of supply and demand. Unfortunately, I've often seen this described as though it depends heavily on psychology, which is doesn't.

    The real problem we have isn't that some schools have theories that depend less on psychology and other schools have theories that depend on it more. Rather, there are many ideas across all schools that are mis-described. Some theories are described as though they depend heavily on psychology though they really don't, others are described as though they don't depend heavily on psychology they they really do.

    Time preference is a tricky one, I'll talk about it another time.

    ReplyDelete
    Replies
    1. =Firstly, I don't think free-will is separable from psychology or from the physical world. I'm a materialist, I see humans as being not that mysterious, just complicated.=

      Well, that's where we fundamentally differ. I do believe that the mental level is not reducible to lower-level material phenomena. And I don't think there is anything mysterious here, it's just obvious from the clear incompatibility of free will with determinism and many other things.

      The complications actually come from the materialists/positivists' willingness to deny the obvious to cram reality into a certain scheme which is not actually dictated by anything.

      Delete
    2. "... it's just obvious from the clear incompatibility of free will with determinism and many other things."

      Why do you think it's obviously incompatible?

      Anyway, do you think that our differences of opinion over the philosophy of the mind have a bearing on Praxeology? I don't think they really do.

      Delete

All anonymous comments will be deleted. Consistent pseudonyms are fine.