Articles

Alternative Categories

In Uncategorized on February 15, 2010 by Joshua Smart

Collin and I have been working on a paper to respond to Witt with the following structure: 1. What are Witt’s categories? 2. What is wrong with Witt’s categories? 3. Alternative categories. 4. Examples of alternative categories.

Below are brief rundowns of our proposed categories and Collin will post our analysis of Witt’s categories. We greatly appreciate any feedback you can give. Are they sensical? Do we miss any major ways in which evolution is applied to economics? (Especially appreciated from economist friends).

—–

  • Universal Darwinism

Universal Darwinism is a term first coined by Richard Dawkins in 1983, but the locus classicus for the concept is Daniel Dennett’s 1995 book Darwin’s Dangerous Idea. This approach sees Darwinian evolution as an algorithm involving variation, selection, and retention that is substrate neutral. Darwin described how this algorithm applies to biological systems. Economists who employ UD attempt to describe how it applies to economic systems. This is not meant to imply anything concerning the interaction or relationship of biological and economic systems, but merely uses Darwinian principles as a heuristic to understand economic systems.

  • Evolutionary Metaphor/Analogy

This approach is not committed to Darwinian evolution actually occurring in economic systems (as UD is). Instead, it attempts to use metaphor or analogy to biological evolution to aid in the description of economic systems. Those who employ this strategy attempt to abstract away from the particulars of biological evolution to generic concepts that are present in economic systems as well. They may attempt to draw specific analogies between the systems (e.g. find economic analogues for genotypes and phenotypes), or they may attempt to abstract away to the most generic principles of evolution (e.g. Witt’s “novelty emergence and dissemination”).

  • Evolutionary Modeling

This approach applies mathematical models that were originally developed to describe the evolution of biological systems, and employs them to describe the evolution of economic systems. Using these models does not commit one to any claims about the connection between the underlying entities and principles governing economic and biological systems. That is to say, the motivation for employing evolutionary models is pragmatic, not metaphysical.

  • Evolutionary History Acknowledgement (working on a pithier name for this one)

This approach to economic research incorporates information about the evolutionary origins of the preferences and behavioral tendencies of economic agents. This approach does not borrow biological evolutionary explanations or analytic tools to describe economic systems. Rather, it incorporates information provided by evolutionary biology into attempts to understand economic agents.

  • Summary

Our distinction between these potential meanings of “evolutionary economics” arises from what is being borrowed from biology. In EHA, it is information concerning the nature of economic agents. In evolutionary modeling it is mathematical machinery. In evolutionary metaphors and analogies it is general features. And in UD, it is nothing (rather, shared principles are employed).

Advertisements

14 Responses to “Alternative Categories”

  1. I’ve actually been meaning to email Colin and Yasha about this, but haven’t had a chance to yet (being sick will do that). Some observations that might be helpful with your second goal:

    1) I think that Witt gets himself in trouble by misunderstanding what a scientific analogy is. My understanding is that in science, analogy is a somewhat technical term with two potential meanings – either two or more systems that share a direct isomorphism or they share a qualitative similarity (see Maynard Smith’s “The Concept of Information in Biology”). Talking about different kinds of analogies as well as metaphors makes me suspect that Witt either doesn’t understand this difference or isn’t being exact enough.

    2) I don’t think Witt is totally clear on what UD is. If UD is, as you say, substrate neutral, then the monism/dualism distinction doesn’t matter and the neo-Schumpeterian approach is potentially also an instance of UD.

    I have a couple other minor thoughts, but maybe we could discuss them in person? It sounds like I’m thinking about the same things, at least.

  2. I’m a fan of Josh and Collin’s four alternative categories here as they seem to make much more sense and are clearer cut than Witt’s.

    I’m wondering though how different the categories of Universal Darwinism (UD) and Evolutionary Modeling (EM) are though. UD appeals to some “algorithm involving variation, selection, and retention that is substrate neutral” whereas EM “applies mathematical models that were originally developed to describe the evolution of biological systems, and employs them to describe the evolution of economic systems.” Is EM merely the mathematical formulation of the (possibly informal) algorithm that is appealed to in UD? Do each of these imply the other? That is, if we are committed to UD, are we also committed to EM and vice versa?

    In both UD and EM, you make sure to emphasize that neither entail an explicit connection between biology and economics (a connection which some could argue exists under the Evolutionary History Acknowledgment view). However, UD and EM both implicitly posit some sort of connection between biology and economics right? That is that phenomena in biology and economics are subject to the same forces or laws or whatever. Though perhaps this does not count as a connection since this does not entail that changes in biology will result in changes in economics or vice versa (unless the change is due to the independent force or law or whatever). Am I just babbling off base here?

  3. In both UD and EM there is a framework that is substrate independent (and that was discovered in biology). In UD it consists of an algorithm (though not a mathematical one). In EM it consists of mathematical equations. In both cases the disciplines of economics and biology are not connected directly to each other, rather there is a tool that is useful for understanding both.

  4. As an example of EM (in the opposite direction), I think of how evolutionary biology has borrowed the use of optimality theory (i.e. optimality modeling) from neoclassical economics. These models were originally developed by engineers, but they also get used within anthropology. How do they all relate? The borrowing of these models results, not because each system is governed by the same evolutionary principles (the stronger claim made by UD & EMA), rather each system shares a basic mathematical relationship that optimality models are able to capture. The structure is that of maximization in the context of various constraints and trade-offs. That is what the equations of optimality theory are designed to capture. However, the employment of these models in different domains involves very different underlying assumptions, concepts (e.g. utility vs. fitness), governing principles, constraints, etc. Therefore, the models are borrowed simply because the same forms of mathematical relationships obtain. This, however, falls short of claiming that the evolution of each system is governed by Darwinian evolutionary principles. The differences show up when we look at what the variables represent and what assumptions accompany the mathematical model in different domains.

    However, it is important to keep in mind that each of these categories postulates a different type of connection between biology and economics. Some combinations are perfectly consistent (e.g. UD and EHA). Indeed, I suspect that many economists who support one of these strategies incorporates at least one of the others as well. As you mention, it would seem strange to be a UD, but reject EM. If the systems are evolving according to the exact same Darwinian principles, then the same modeling techniques had better have some application in economics as well! Also, it seems to make little sense to endorse UD (that the systems are subject to the same principles of evolution), but deny EHA (that information about agents’ biological history is essential to economic inquiry). The important point is that they are two logically independent claims. Neither implies the other.

    The vital goal, however, is to keep each of these potential connections separate. We suspect that much of the confusion that arises from Witt’s categories is the failure to adequately distinguish the types of connections between the two domains. Rather, he focuses on the distinctions in POSITIONS that have been taken. This causes a good deal of trouble (or at least cofusion) when the positions the authors have taken make claims about several different ways in which that the two fields are connected. This conflation of issues can also be clearly seen in the Hodgson paper. Working out the possibile (and the impossible) combinations of strategies is certainly important, but I think keeping the above categories separate will help to clarify, “What is specific about evolutionary economics?”, what the viable positions are, and which arguments are relevant to which strategies.

  5. Oh, by the way. I misspoke when I said “each system is governed by the same evolutionary principles (the stronger claim made by UD & EMA)”. Although each of these positions makes a claim that is independent (and in some sense stronger) than that made by EM, the two views are very different. Only UD claims that the systems are subject to the same evolutionary principles. EMA merely claims that there are more abstract similarities between the two systems that can be utilized in the formation of metaphors, analogies, or perhaps a more generic concept of evolution. This is an important diffence. Though both are different (though perhaps not consistent with the denial of) from the connection claimed by EM. One could accept EM without accepting either UD or EMA. Just wanted to clear up any confusion that might cause later on.

  6. The summary is very good in distinguishing the differences of all those theories. I guess more strictly speaking, the differences concern what is borrowed from the evolutionary theory rather than biology. That is a minor point, though.

    Leo is unclear about the distinction between UD and EM (evolutionary modeling), while I am unclear about the distinction between UD and evolutionary metaphor/analogy (EA). Josh said that “This approach [EA] is not committed to Darwinian evolution actually occurring in economic systems (as UD is)”. In my opinion, UD is commited to Darwinian evolution occuring in every evolutionary phenomenon, including economical evolution. As far as I understand Witt, theorists of UD are those who use evolutionary metaphors, And Hodgson, a proponent of UD, actually spent some time explaining why using the metaphors are not bad.

    But maybe sometimes it is difficult for one to subscribe to one but not the other theory, as Collin said. If so, then it will be helpful to point out who subscribe to which theory and see the connection among all the theorists from this angle.

    Jake seems to invoke a difference between metaphors and analogies. Jake, can you clarify that distinction? Saying that metahpors have two potential meanings just seems to say that the metaphors are ambiguous. But I think you don’t mean that.

  7. Wenwen,UD is a theory that Darwin described an algorithm that can be applied to disparate domains. One domain (which was his context of discovery) is biology. Economists that employ UD think applies in the economic domain.

    When economists draw evolutionary analogies they are noting important similarities (at some level of abstraction) between biological and economic systems, and attempting to draw useful inferences from those similarities. But they are not committed to any underlying reason for those similarities. (Whereas UD is committed to the operation of the Darwinian algorithm as an explanation for similarities.)

    One might subscribe to UD and think in terms of analogy because it’s helpful in the context of discovery, but there would be no need to use it in the context of justification.

  8. I just wrote up my thoughts On Witt. It’s pretty rough, but it’s longer than I was expecting (just over 1800 words). The file is in my folder on the public drive. I’m just now reading Josh and Collin’s blog posts on the subject, so the only exposure I’ve had to their ideas was during the seminar. This might be a good case study in parallel memetic evolution :-). I spilled a bit of ink trying to nail down what a scientific analogy is, and Jake sums it up in one word: isomorphism. I think I said things other than that about analogies, but that’s a pretty good summary.

    Josh said:

    “When economists draw evolutionary analogies they are noting important similarities (at some level of abstraction) between biological and economic systems, and attempting to draw useful inferences from those similarities. But they are not committed to any underlying reason for those similarities. (Whereas UD is committed to the operation of the Darwinian algorithm as an explanation for similarities.)”

    Damn, I just argued pretty much the opposite. I didn’t think there’d be any empirical/scientifically relevant difference between using an evolutionary model in an analogical way and using an evolutionary model with the assumption that the process you’re modeling really *is* an evolutionary process. Now that you mention it, I suppose a modeler could just be using an evolutionary model as a “good enough” approach with the understanding that the real process is non-darwinian. (In fact, I think Witt mentioned something along the lines of game theory modeling based on adaptive genetic algorithms in order to model non-darwinian adaptive learning in humans.) The question of analogy/reality could be settled by seeing if a non-darwinian model more accurately represents reality. Which means neo-Schumpeterism might not be an instance of U.D. Oops.

  9. It looks like Colin and Josh kept Witt’s 2×2 matrix. I expanded to a 4×4. What you’re calling EHA is one type of what I’m calling “economic naturalism.” You’re welcome to use it if you want a catcher name. The other type of economic naturalism investigates two-way interaction between the economic and biological spheres.

  10. Josh, having read your comments, now I am not clear what evolutionary analogy/metaphor means to show. According to you and Collin, it only shows that economics and biology are analogous. They can be analogous in terms of the Darwinian principles, or they can be analogous in terms of novelty emergence and dissemination.

    My take is that novelty emergence and dissemination are not particular of Darwinism, so they are not the distinctive features of Darwinian evolution. As for UD, if it is an algorithm that applies to different areas, doesn’t it follow that the algorithm is where evolutions in different areas are analogous?

    As for the context of discovery/justification distinction, if we just take the metaphor to describe the similarities, rather than to justify anything, it seems that the similarities must follow from UD.

  11. Dan: I really like “evolutionary naturalism”. That is exactly what we mean by EHA, and your distinction about the different types of interaction is important to note as well. One thing of clarification. What Josh and I are proposing is NOT an alternative matrix to Witt’s 2×2. Rather, we want to distinguish four different types of questions (that is, four different types of connections) concerning the relationship between explaining economic systems and evolutionary biology. Some answers to these four types of questions will be consistent, others will not. Which combinations are viable still needs to be worked out a bit. That is, we don’t want to claim that every position will need to take a stand on each of these issues in order for us to evaluate that approach. They are each separate issues that need to be addressed separately. What the resulting matrix of possibilities will look like depends on what combinations are allowed. In that sense at least one might say that we are proposing a 4×4 matrix where some of the cells are likley to be blocked out as inconsistent approaches.

    Wenwen: I think you are exactly right to push the relationship between analogies between Darwinism in biological and Witt’s generic concept of evolution: novelty emergence and dissemination. As you say, novelty emergence and dissemination are not particular of Darwinism. The point I want to stress is that neither are the properties that enter into the analogies and metaphors. It is precisely their being present in both systems that makes the metaphor/analogy work. In short, I see little difference between claiming that two systems are analogous because they have a set of properties in common and claiming that there is a more generic concept of evolution that each system instantiates. In either case we abstract away from the particular details to some more generic similarities between the two systems. Whether we call it a metaphor or an abstraction to a more generic concept of evolution seems irrelevant to what is really going on: abstraction from the particular details of Darwinian evolution. In contrast, the UD claims that no abstraction is required since the particular details are exactly the same.

    I must admit I’m unclear on how to put that idea, or whether it is on the right track. I can see a connection between metaphorical use and abstracting away to a more generic concept, but I can see some subtle differences. So far though I don’t see those difference as relevant to the overall strategy. I hope Jake’s paper ideas can help us sort this out a bit more.

    Cheers,
    Collin

  12. Wenwen said:

    “having read your comments, now I am not clear what evolutionary analogy/metaphor means to show.”

    Well, on my account, this corresponds with Question 2: “What is the relationship between the processes that drive biological evolution and the processes that drive economic change?”

    Possible answers (modified from my initial draft):
    A. The mechanisms that drive economic change do not resemble darwinism in any important respect.
    B. The mechanisms that drive (certain kinds of) economic change resemble darwinism metaphorically. (This approach was inspired by darwinism, but is not “really” darwinian. <- I'll clean this up later.)
    C. The mechanisms that drive (certain kinds of) economic change can be approximated using darwinian models, but these mechanisms are not darwinian in reality.
    D. The mechanisms that drive (certain kinds of) economic change are isomorphic with the mechanisms that drive biological evolution.
    N. (n/a) This approach does not address the topic of economic change.

    Jake said:

    "I don’t think Witt is totally clear on what UD is. If UD is, as you say, substrate neutral, then the monism/dualism distinction doesn’t matter and the neo-Schumpeterian approach is potentially also an instance of UD."

    I would agree. Per my typology, I'm defining UD as *D (Answer "D" to Question 2, agnostic on Question 1). After rethinking my typology a little, I would argue that neo-Schumpeterism is 2D (D to Question 2, treats interaction with biological sphere as noise). If so, then neo-Schumpeterism is an instance of UD.

  13. Wow! This is exceptionally useful science. (I am leaning on D. McCloskey’s (1985) epigram: “good science is good conversation”). I am reading and re-reading back and forth through the thread to better understand the rhetoric. But my early thought is that the analyses of these typologies is leading somewhere of value.

    One quick riposte to: “When economists draw evolutionary analogies they are noting important similarities (at some level of abstraction) between biological and economic systems, and attempting to draw useful inferences from those similarities. But they are not committed to any underlying reason for those similarities.”

    I believe the first statement to be true. I believe the second statement is true only if the analogy is not formalized and not used for the purposes of modeling causal relations. If otherwise, then there is a need to investigate the reasons. My belief on this is tied, in part, to Collin’s ideas about optimality in analogous systems. I think of the use of genetic algorithms as an engine for simulating change in economic systems, for example.

    But, as I said, I need more reflection (as my French colleagues say) on this thread. Reflection is lubricated by good wine (insert allusion to French here) and I am working on that this evening.

    Thanks for this stimulating series of posts.

  14. I’ve been thinking a lot lately about the relationship between using mathematical models and claiming that an analogy exists between two systems (this grew out of the useful discussion in which I you all pressed me on this point of Josh and I’s categories). What is the relationship between using the same modeling techniques to study a system and drawing an analogy between the two. I think there are several key differences, although there is some overlap. First, it is worth noting that when two systems are analogous, that would seem to be strong motivation for using similar models for those systems. That is, assuming tha the analogy has something to do with the structural properties of the two systems (e.g. the functional roles involved). However, this is not the only reason to use modeling techniques from other fields, though it is probably the most common. Another reason might simply be predictive accuracy or the computational tractability offered by the technique. For instance, when biologists use optimality models formulated by engineers, the only thing that is assumed to be the same in the two systems is the maximization of some design variable subjec to various contraints and trade-offs. This similarity is purely mathematical. This is the only similarity I can see being drawn between say designing a bridge and designing phenotypic traits. It claims nothing more about the connections between what the variables are intended to represent or the other concepts involved in applying the modeling technique. The only similarity is the mathemtatical structure, just as the applicability of calculus across domains implies very little about the systems that can be usefully described by it. This mathematical similarity is a rather minimal connection between the systems being used (unless for some reason the modeling technique builds in something more). This is all the modeler is committed to concerning the similarities of the systems in question.

    In contrast, when one uses an analogy between two systems, one seems to be claiming something stronger. These analogies will come in varying strength. Some will draw only minimal connections between the two systems, others will claim far more. The important point is that these similarities are not restricted to those of underlying mathematical structure. It is certainly true that one could attempt to draw an analogy that only cited structural similarities of the kind implied by using the same modeling techniques (i.e. similarities in mathematical relationships). However, this does not seem to be the common intended use or resulting (perhaps unintendend) implication of drawing such analogies in science. If the structural relationships between mathematical variables is all that is intended to be claimed, then drawing the analogy seems to be a bit misleading and opens the door for misinterpretation in the sense that other similarities might be “smuggled in” by not being explicit about the intended connection between the two systems. In short, analogies usually involve similarities other than mere structural similarity in terms of the underlying mathematics. Using mathematical modeling techniques seems only to require a commitment to this limited connection. Using modeling techniques is necessarily connected to certain mathematical strctural similarities; analogy drawing, in contrast, is not. Indeed, I think analogies are often drawn either because the structural similarities are not well enough understood to be able to separate them out, or because something more is intended to be claimed about the connection between the systems in question.

    Finally, I think these two ways of connecting systems will sometimes be used simultaneously by investigators. Those who draw analogies certainly seem to imply that they beleive there will be something like the strctural similarity implied by using similar modeling techniques. That is, I agree with Randy, that it would seem somewhat odd to draw a rather explicit analogy between two systems that is intended to convey unerlying structural similarities and then maintain that the same modeling techniques would not be useful for capturing some underlying structural similarities between the two. However, it seems that one could consistently use the same modeling techniques without being committed to anything more than the ability of the mathematics to capture a certain structural relationship. Furthermore, one can draw analogies that connect systems in various ways without implying that the same mathematical techniques should be used to study the systems. In the end, using models and drawing analogies will sometimes overlap, but they are importantly different and each is better suited to different types of similarities between scientific domains.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: