Help Me Trace the Intellectual and Biographical Origins of Signaling Theory
May 14, 2011 9:26 AM Subscribe
A. Michael Spence and David K. Lewis are two highly regarded, contemporary scholars. Spence was the 2001 co-recipient of the Nobel Prize in economics (with George Akerlof and Joseph Stiglitz), given in part for his work on job market signaling. Lewis is considered to be one of the greatest metaphysical philosophers of the 20th century and a few years before Spence he developed his own theory of signaling.
I cannot find any evidence that Spence's work on signaling is directly connected to Lewis's own work on the same subject - no citations, acknowledgements or biographical information that I have found suggests Spence was influenced by Lewis.
I'm trying to collect more information on the links between these two men so I can better understand the historical development of signaling in economics. I am specifically interested in whether Lewis influenced Spence directly or indirectly (as opposed to both men developing their work on signaling due to some common third factor, like "it was just in the air"). Did they independently discover signaling? What is Spence's unique contribution in signaling theory given Lewis's own work on the subject a few years earlier? What role did Schelling play? Below I have produced some timelines and citations, but there's not a lot given my ignorance of Lewis's work.
Note that I am not a philosopher, and before this week, I wasn't familar with Lewis. I have since been told by a friend of mine in the philosophy department that he is probably one of the greatest metaphysical philosophers of the second half of the 20th century. That independently is interesting, but more narrowly, I was intrigued because of his use of game theory to do philosophy. I think I knew some philosophers did that, but I did not know that anyone of his stature did so. I also was fascinated to learn that his dissertation predates Spence's by 3 years, but contains a lot of the properties of signaling (including pooling and separating equilibria as far as I understand) that we often associate with signaling models. But briefly, here's the information that I have, which isn't much.
1. David Lewis studies at Oxford University for one year from 1959-1960 (Wikipedia). While at Oxford, he heard JL Austin, and was persuaded to pursue a career as a philosopher. Who else did he interact with at Oxford, besides Austin? Did he interact with mathematicians, economists or game theorists?
2. A. Michael Spence graduates from Princeton in 1966, and studies at Magdalen College Oxford afterwards. At Princeton, he had majored in philosophy. He studies mathematics at Oxford and graduates in 1967-1968. JL Austin had died in 1960, so Spence would not have had contact with him. But would Spence have also been exposed to the same philosophical ideas about language and communication that had inspired Lewis 7 years earlier? Was Spence showing an interest in communication problems then?
3a. Lewis pursues a PhD in philosophy at Harvard under Quine and finishes in 1967 (wikipedia). For the duration of time that he is producing this dissertation, Spence is either at Princeton (graduated 1966) or Oxford (graduated 1968). Lewis writes a dissertation that incorporates game theoretic concepts and reasoning to analyze social conventions, with a special focus on language. He publishes his dissertation as a book for Harvard University Press entitled Convention in 1969. The dissertation wins the Franklin Matchette Prize for the best book published in philosophy by a philosopher under 40 years old (wikipedia).
3b. Lewis appears to have taken a job at UCLA just before formally defending his dissertation in 1967, because from 1966 to 1970, he is listed in various timelines as an assistant professor at UCLA. In 1970, he takes a job at Princeton, where he remains until his death. Recall that Spence had studied philosophy at Princeton, but had graduated in 1966. So the two young scholars do not overlap.
4. Alan Hajek (here) suggests that Lewis had read and was heavily influenced by Thomas Schelling's 1960 Strategy of Conflict. But signaling is not something usually associated with Schelling's original contributions, though clearly Schelling is working on many problems related to coordination, which is also Lewis's concern.
5. In his Nobel speech, Spence recounts the story of how he personally started working on what ultimately became signaling.
"When I was a graduate student in economics at Harvard, I had the privilege of serving as rapporteur for a faculty seminar in the then-new Kennedy School of Government. Among other distinguished scholars, it included all of my thesis advisers, Kenneth Arrow, Thomas Schelling, and Richard Zeckhauser. In the course of that seminar there were discussions of statistical discrimination and many other subjects that relate to the incompleteness of information in markets. One of my advisers came in one day with the strong suggestion that I read a paper he had just read called “The Market for Lemons” by George Akerlof. I always did what my advisers told me to do and hence followed up immediately. It was quite electrifying. There we all found a wonderfully clear and plausible analysis of the performance characteristics of a market with incomplete and asymmetrically located information. That, combined with my puzzlement about several aspects of the discussion of the consequences of incomplete information in job markets, pretty much launched me on a search for things that I came to call signals, that would carry information persistently in equilibrium from sellers to buyers, or more generally from those with more to those with less information."
In footnote 3, Spence credits Robert Jarvis for creating the terminology "signal" and "indices", but not theory itself ("Signals are things one does that are visible and that are in part designed to communicate. In a sense they are alterable attributes.", fn. 3). And though he mentions Akerlof's lemons paper, Akerlof's paper does not seem to have the germ of the idea of signaling.
And that's about all I have.
Note that I am not a philosopher, and before this week, I wasn't familar with Lewis. I have since been told by a friend of mine in the philosophy department that he is probably one of the greatest metaphysical philosophers of the second half of the 20th century. That independently is interesting, but more narrowly, I was intrigued because of his use of game theory to do philosophy. I think I knew some philosophers did that, but I did not know that anyone of his stature did so. I also was fascinated to learn that his dissertation predates Spence's by 3 years, but contains a lot of the properties of signaling (including pooling and separating equilibria as far as I understand) that we often associate with signaling models. But briefly, here's the information that I have, which isn't much.
1. David Lewis studies at Oxford University for one year from 1959-1960 (Wikipedia). While at Oxford, he heard JL Austin, and was persuaded to pursue a career as a philosopher. Who else did he interact with at Oxford, besides Austin? Did he interact with mathematicians, economists or game theorists?
2. A. Michael Spence graduates from Princeton in 1966, and studies at Magdalen College Oxford afterwards. At Princeton, he had majored in philosophy. He studies mathematics at Oxford and graduates in 1967-1968. JL Austin had died in 1960, so Spence would not have had contact with him. But would Spence have also been exposed to the same philosophical ideas about language and communication that had inspired Lewis 7 years earlier? Was Spence showing an interest in communication problems then?
3a. Lewis pursues a PhD in philosophy at Harvard under Quine and finishes in 1967 (wikipedia). For the duration of time that he is producing this dissertation, Spence is either at Princeton (graduated 1966) or Oxford (graduated 1968). Lewis writes a dissertation that incorporates game theoretic concepts and reasoning to analyze social conventions, with a special focus on language. He publishes his dissertation as a book for Harvard University Press entitled Convention in 1969. The dissertation wins the Franklin Matchette Prize for the best book published in philosophy by a philosopher under 40 years old (wikipedia).
3b. Lewis appears to have taken a job at UCLA just before formally defending his dissertation in 1967, because from 1966 to 1970, he is listed in various timelines as an assistant professor at UCLA. In 1970, he takes a job at Princeton, where he remains until his death. Recall that Spence had studied philosophy at Princeton, but had graduated in 1966. So the two young scholars do not overlap.
4. Alan Hajek (here) suggests that Lewis had read and was heavily influenced by Thomas Schelling's 1960 Strategy of Conflict. But signaling is not something usually associated with Schelling's original contributions, though clearly Schelling is working on many problems related to coordination, which is also Lewis's concern.
5. In his Nobel speech, Spence recounts the story of how he personally started working on what ultimately became signaling.
"When I was a graduate student in economics at Harvard, I had the privilege of serving as rapporteur for a faculty seminar in the then-new Kennedy School of Government. Among other distinguished scholars, it included all of my thesis advisers, Kenneth Arrow, Thomas Schelling, and Richard Zeckhauser. In the course of that seminar there were discussions of statistical discrimination and many other subjects that relate to the incompleteness of information in markets. One of my advisers came in one day with the strong suggestion that I read a paper he had just read called “The Market for Lemons” by George Akerlof. I always did what my advisers told me to do and hence followed up immediately. It was quite electrifying. There we all found a wonderfully clear and plausible analysis of the performance characteristics of a market with incomplete and asymmetrically located information. That, combined with my puzzlement about several aspects of the discussion of the consequences of incomplete information in job markets, pretty much launched me on a search for things that I came to call signals, that would carry information persistently in equilibrium from sellers to buyers, or more generally from those with more to those with less information."
In footnote 3, Spence credits Robert Jarvis for creating the terminology "signal" and "indices", but not theory itself ("Signals are things one does that are visible and that are in part designed to communicate. In a sense they are alterable attributes.", fn. 3). And though he mentions Akerlof's lemons paper, Akerlof's paper does not seem to have the germ of the idea of signaling.
And that's about all I have.
Response by poster: dfriedman - sorry. I wrote and rewrote this probably too many times and the actual question got lost. Is there any evidence that David Lewis influenced Michael Spence's development of signaling theory? I can't find any evidence that Spence was even aware of Lewis's work, but I find plenty of historical overlap between them (noted in the second half timeline-related stuff I post). The only thing I can find is that Lewis had been influenced by Schelling, and Schelling was Spence's thesis adviser. But beyond that, I can't find anything suggesting that Spence was influenced by Lewis's dissertation on signaling. I find plenty about the historical development of signaling in Lewis's work, and who his influences were, but not for Spence (maybe partly because the history of thought writers have not done as much on game theory for the late 20th century?). I mainly find references to Spence's immediate contemporaries in economics, but am now curious because Lewis has a major work on signaling in philosophy just a few years before Spence at Harvard, and Schelling seems like a natural bridge, but I can't find evidence for that. I thought maybe others out there with backgrounds in economics, game theory, and philosophy might know more.
posted by scunning at 9:38 AM on May 14, 2011
posted by scunning at 9:38 AM on May 14, 2011
I recommend sending one or the other or both an email, but much (much) briefer than this and certainly not including the evidence. Just say who you are and that you're interested in the sociology of knowledge and would be grateful if they could tell you a little more about what influenced this particular development and especially whether so-and-so's works X or Y might be in the mix somewhere, directly or indirectly.
posted by Monsieur Caution at 9:53 AM on May 14, 2011
posted by Monsieur Caution at 9:53 AM on May 14, 2011
Given how many simultaneous discoveries occur in science/math, the connection you have outlined in your followup seems quite sufficient to explain this, no? I.e. schelling influencing both of them. Also, don't forget that in that era it could take somewhat longer for ideas to percolate, especially across field boundaries, than it does on internet time.
I recommend sending one or the other or both an email, but much (much) briefer than this and certainly not including the evidence.
David Lewis died in 2001.
posted by advil at 10:04 AM on May 14, 2011 [1 favorite]
I recommend sending one or the other or both an email, but much (much) briefer than this and certainly not including the evidence.
David Lewis died in 2001.
posted by advil at 10:04 AM on May 14, 2011 [1 favorite]
Response by poster: Lewis passed away in 2001. Spence is a Nobel Laureate and now in administration at NYU, so I don't feel comfortable talking to him. "Evidence" is probably the wrong word, btw, because it implies wrongdoing. I mean evidence strictly in the sense as to whether anyone can trace Spence's work on signaling to Lewis directly or indirectly. As I said, they overlap a lot, and seem to be moving in the same circles. At this time in economics there is a ton of work being doing on the information structure of markets, and more broadly, even in evolutionary biology, people are suggesting something like signaling (the handicap principle, for instance, is in a 1975 article in a theoretical biology journal). Spence and Lewis are interested in very different things, too: language vs. job markets. But I'm just trying to understand the history of thought better, as economics and philosophy really don't overlap a lot given each's extreme specialization. So I wondered, maybe Schelling is the missing link - Schelling was almost like a polymath, and being from RAND, he interacted with everybody. But I can't find anything really even offering this up as a possibility. I also can't find anything that even is thinking about signaling theory being simultaneously developed by Lewis and Spence, and so I wondered if maybe I was missing something subtle - maybe the differences between the formal analytical methods are so different that it's not quite correct to say they are the same discovery. They are always both embedded, afterall, in a very specific application to some problem. Spence is doing applied microeconomic theory, as opposed to pure game theory, and Lewis is doing something like that in philosophy. But the discoveries seem remarkably close: close in time, close in space, close in content too. Maybe there were seminars at Harvard around then where these ideas would have been percolating? It's something along these lines that I'm trying to understand better.
posted by scunning at 10:04 AM on May 14, 2011
posted by scunning at 10:04 AM on May 14, 2011
Response by poster: advil - yes, the schelling explanation seems to be it, and probably after today, I'll be content with that hypothesis.
The simultaneous discovery stuff is interesting - it does happen all the time. What makes this kind of weird to me is that there is so little overlap in metaphysics and microeconomics. But when I trace through Lewis's background, and Spence's too, it's almost like they are fellow travelers at least in terms of places and interests. Both at Oxford (but not at the same time, and not in the same departments), both philosophy majors as undergrads (but not in the same place), both at Princeton (but again, one as a professor, one as a student, and not at the same time), both influenced by Schelling. I keep thinking - maybe it's not so much that they are being affected by the same people, as that they are drawn to the same kinds of people, but from different points of view and with different problems in mind. I need to get _Convention_ and see how close it is to Spence's work - I don't know it well at all.
posted by scunning at 10:08 AM on May 14, 2011
The simultaneous discovery stuff is interesting - it does happen all the time. What makes this kind of weird to me is that there is so little overlap in metaphysics and microeconomics. But when I trace through Lewis's background, and Spence's too, it's almost like they are fellow travelers at least in terms of places and interests. Both at Oxford (but not at the same time, and not in the same departments), both philosophy majors as undergrads (but not in the same place), both at Princeton (but again, one as a professor, one as a student, and not at the same time), both influenced by Schelling. I keep thinking - maybe it's not so much that they are being affected by the same people, as that they are drawn to the same kinds of people, but from different points of view and with different problems in mind. I need to get _Convention_ and see how close it is to Spence's work - I don't know it well at all.
posted by scunning at 10:08 AM on May 14, 2011
Spence is the one whose intellectual genealogy you're trying to piece together, and he's alive, right? The Nobel laureates I've exchanged email with in the past were fine with answering brief questions. And evidence does not imply wrong-doing. You're really over-thinking this.
posted by Monsieur Caution at 10:16 AM on May 14, 2011
posted by Monsieur Caution at 10:16 AM on May 14, 2011
Response by poster: Monsieur Caution - thanks. Over-thinking in what way? Over-thinking the history of the thought? Or over-thinking your question to me?
posted by scunning at 10:18 AM on May 14, 2011
posted by scunning at 10:18 AM on May 14, 2011
« Older How do we get rid of red mites on a lovebird? | Women's health says this will get me a beach body Newer »
This thread is closed to new comments.
posted by dfriedman at 9:30 AM on May 14, 2011