There was an interesting paper (a couple of them, actually) presented at the Sixth Midwest Epistemology Workshop hosted by Indiana University Bloomington. The particular paper I have in mind, “The Practical Dogmatist” by Roy Sorensen of Washington in St. Louis, was very intriguing; but it was not all that well received by the IU Bloomington Philosophy Department. My impression is that the paper’s importance was missed, and this impression is supported by some of the questions asked and comments made after Sorensen gave his presentation. It could be that the paper is still in an early state of formation, inchoate in its development, but I think the merit in its line of thought can be seen. It could very well have been the case that few had actually read the paper prior to the presentation.
Among the central themes of the work is the role that particular pieces of meta-information should play in our reasoning processes, that is, the epistemic status of the meta-information. Perhaps I can bring greater clarity (and maybe formulation) to the scenario by suggesting that we consider three epistemic levels of the decision-maker: 1) the formal or semi-formal reasoning level, 2) the level of meta-reasoning, and 3) the epistemic psychological level. Given any piece of information at random and without artifice —I will explain why artifice is important, in a minute—, the decision-maker will, almost invariably, simply plug the data into his or her reasoning system or send it through the appropriate lines of rigorous method. This is the semi-formal or formal level of reasoning, which might include instrumentation, systematic methodology, logic, or mathematics, depending on the specifics and nature of the decision-maker. The lawyer will gauge the constraints on words, assess the codes and laws relevant to the data, and so on. The scientist will use, maybe, statistical analysis, or whatever. A doctor will perform a biopsy, take vitals, and so on. This level of reasoning, in detail, is not too interesting for the purposes of the discussed paper, but it is important for us to lay down the groundwork.
The meta-reasoning level is the level in which intuition, judgment, and so forth, come into play. For example, the lawyer and the doctor filter out, by virtue of experience, all of the data that they deem irrelevant —all of those observables, useless minutiae, that would waste precious time, if one were to examine. Kuhn has a particularly relevant line in The Structure of Scientific Revolutions, where he says that, if a scientist paid attention to every anomaly, he or she would get nothing done. Kuhn is talking about the scientist’s intuitive ability to assess data as being useless. This meta-reasoning level is one in which the non-rigorous lines of reasoning make assessments of data, et cetera, and where decisions are made regarding what will or will not enter the more formal level of reasoning. It is on this level that Sorensen’s puzzle will place its crux. Before, explaining this, we must enter into the epistemic psychological level, which is where the problems arise from.
The major point left out in virtually every epistemology that depends explicitly, in some way, upon the mind is that they lack the psychological component of the admixture. In a few cases, this isn’t important, but, generally, it is. The reason is that humans can only see the world by way of some theory. The world would not be intelligible otherwise. Moreover, the underpinnings of that theory are founded upon assumptions rooted in the subconscious. One could not empty out their Cartesian basket of apples because doing so would require two things: 1) consciousness of each individual atomic component —if such things exist— that comprises the individual’s worldview and 2) a method by which one could enumerate all hypotheses or demonstrate that each component is atomic. This is simply impossible, as far as the science as psychology is understood today. To put all of this in other words, it will most likely be the case that there are things about ourselves (assumptions we have made and so on) which we do not realize undergird our thought, behavior, and things of the like. A large part of the problem for Sorensen’s politician comes from this level, because it is the level in which there is information about the decision-maker that the decision-maker does not know, and this opens up greater avenues of uncertainty.
The packet that Sorensen’s politician receives should concern us for just the above reason, namely, that the lobbyist could be a super psychologist who has been observing the politician for a long time, and who has determined many, many things about the politician that the politician does not know about his or herself. Let’s change the scenario up, for just a minute. Suppose that the politician is a Texas Hold’em poker player under the gun, playing heads-up with a second (evil) poker player. The political informant, now a poker coach, tells the poker player that the dealer is in cahoots with the evil player, is a master psychologist, and has dealt the cards such that, if the cards are looked at, the dealer and evil player will know what two pocket cards have been dealt. How does the poker player deal with this meta-information? Here’s the thing, if the dealer is going to deal one particular card set more frequently, because it produces a psychological tell, then Sorensen’s personal inclination of burning the file is correct, especially if the frequency is completely unknown; that is, the poker player needs to bet without looking at the cards, because an instance of a psychological tell would be catastrophic, game ending essentially. In a variation of this setup, if the dealer is not in cahoots, but the poker coach informs the poker player that the evil poker player is aware of a certain percentage of the all physiological responses to precise combinations of cards, then we have a very interesting and different scenario. I say “interesting and different” because those percentages on the meta-information level, if known precisely, can be worked into the formal level of reasoning —and we are, of course, assuming the cards that come into play will be random. This is a simplified version of Sorensen’s setup, in the sense that we can clearly see how the meta-data, as supplied by the informant, should alter the poker player’s meta-reasoning level. Part of the reason that this scenario is easier to grasp is thanks to the fact that the possible psychological information possessed by the evil poker player (and a possible card dealer in cahoots) would be linked to a quantifiable gauge. From there, we can see how much the poker player would need to know, in order to make a meta-reasoning level decision about how to implement the data, and, on the formal-reasoning level, what way the poker player will implement the information. Let’s be clear, at this point: Sorensen’s puzzle is much more general and much more open-ended than this repackaged version I have produced. Let’s get back to Sorensen’s version, comparing it to the repackaged version.
In juxtaposing the politician and poker-player stories, one should note that the ontological status of the meta-information is unclear in Sorensen’s story of the politician. What I mean by this is that, as Sorensen explained in his presentation, it could quite literally be the case that the envelope has receipts in it; that is, the informant’s information, alone, may be the only information that has been added to the system. In this case, it is unclear whether the lobbyist has any privileged access to information that could produce an effect in the politician’s decision-making process. With respect to the poker-player scenario, we could have added this element, too, but it was an illustrative example, so we did not. This should make clearer why Sorensen is concerned. There is a number of possibilities that the politician, in his scenario, must consider: 1) is the lobbyist a super psychologist, possibly meaning that the lobbyist has picked up on a background assumption that underlies the politicians thoughts and behaviors, which could be exploited; 2) has the lobbyist observed a systematic error in judgment/intuition (on the meta-reasoning level), making it possible for the lobbyist to slip information into the politician’s formal reasoning system that will send it astray; 3) has the lobbyist found an error in the politician’s formal/semi-formal system of reasoning; and there are any number of other concerns, such as whether the lobbyist has figured out how to psyche out the politician, even though the lobbyist has no access to privileged information. I hope this begins to affirm in the mind of the reader that Sorensen’s puzzle is, in fact, truly a puzzle. At this point, I would like to present some feedback to concerns raised at the workshop.
The first question, which struck me as having missed the above-given considerations, was Gary Ebbs’ concern that, if the politician burned the envelope from the lobbyist, or simply never opened it, isn’t that just a case for dogmatism? In other words, Gary’s concern is that the suggestion is closing ourselves to information means just sticking our fingers in our ears, and never allowing new information in, which, I think he would say, is intellectual dishonest and closed-minded. I think what was missed, here, is the possibility of structure in the data that the lobbyist prepared for the politician. Maybe Sorensen didn’t make this clear enough, though I came away with this understanding. A very good example of illustrating how synthetically implemented structure in the data can make all the difference, I have to cite two of my favorite chess books, Bruce Pandolfini’s Traps and Zaps and Traps and Zaps II. In these books, Pandolfini instructs how a B-class to expert chess player can prey upon the mid-level tournament player’s lack of intuition, computing ability, experience, and instinctive hunger for material. He does this by teaching how to set up positions in which it looks as though the expert “trapper” (analogous to the lobbyist) has made a blunder, allowing for the mid-level player (the politican) seize a material advantage. These positions are nothing more than phenomenal representations, a collection of data, in which there is an underlying structure that is typically too complicated for the mid-level player to see through to the danger. Put yourself in the shoes of the mid-level player: The chess coach (informant) tells you that you are playing a good player, so do you seize the material, with the possibility that the expert has made a mistake, or do you pass on the opportunity to potentially win? You can see how the uncertainty that the politician is faced with, regarding the unclear nature of just how good the lobbyist might be, throws a monkey wrench into the decision-making process. What can be said is that Gary’s concern about dogmatism misses the possible structural potency of lobbyist’s information. In principle, Gary is right to worry about such dogmatism, in the day-to-day, but I think there is a definite espionage-like concern, here, which transcends the more mundane concern that watching a news broadcast will persuade your judgment into error. Sorensen was going for a critical instance of judgment, doing his best to make the storyline dramatic to produce the desire effect. Again, the presentation of the plot might have been rough, but the central components were all there, so much so that, at least, I got the idea.
Another concern, which went beyond Gary’s, was the concern brought up by an undergraduate student: Doesn’t the justified burning of the envelope suggest a grounds for legitmating behavior exhibited by Creationists (exempli gratia, the refusal to acknowledge any kind of data that might appear detrimental to position currently held). Again, I think this concern misses the critical drama that was intentionally sewn into the story to make it an epistemological puzzle. Reverting to the previous paragraph, this concern also falls to the wayside.
After Sorensen’s presentation, he was bombarded with comments regarding the artificiality of the time constraint he imposed on the decision-making process of the politician. I thought this was a bit absurd on the part of the participants. In what scenarios in life is there limitless time to exhaustively analyze data? The chess game has time constraints, and so does the Texas Hold’em (if a player takes “too long”); and the lawyer has tremendous time constraints, as well. Again, there is a dramatic element that is necessary to the story, so that it has the sense of urgency of “what do I do?” and imminence to the end of the time in which a decision must be made.
Overall, I think the idea in the paper is fantastic, and the analysis needs expansion. On this latter point, Sorensen is simply opening up for discussion what some possible avenues are for discussing this problem. I certainly don’t endorse his strategy of burning the envelope (then again, I don’t endorse any, yet, because I think more analysis needs doing), but I will say that the vast majority of expert chess players, when playing against IMs and GMs could probably benefit from Sorensen’s advice, in not gobbling up information that looks too good to be true. I should also like to say that, if I was the only one to get Sorensen’s point, and that the entire IU Bloomington Philosophy Department missed Sorensen’s point, the paper probably needs thorough reworking for the sake of general intelligibility. (My readership will feel free to comment upon whether you feel Sorensen’s paper is easily understood or not.) All in all, I am not familiar any other philosopher who has brought forth a similar puzzle, and I enjoyed the ideas presented in it, and the creative manner in which they were presented.