Being a person who likes neat mental hierarchies, Hume's Problem of Induction and the implications of Gödel's Incompleteness Theorems (particularly the unprovability and unfalsifiability [in a logical sense] of certain parts of axiomatic systems–namely mathematics and Aristotelian logic) have emotionally nagged at me ever since I learned about them. In a recent episode of "Rationally Speaking", a podcast which I enjoy immensely, I was reminded of Quine's Web of Belief, which involves various belief strands with various thicknesses, but wherein the snapping of one strand does not necessarily lead to the collapse of the whole web. This is like an analogy I've heard, wherein all empirical claims are accompanied by a permanent asterisk, which is essentially saying "There is no ultimate basis for this." Even despite my desire for a rigid hierarchical system of bases, this answer is still somewhat satiating. This is especially the case, since the concept of an ultimate basis may be incoherent in any case, and this answer essentially says, "We don't have it and we also don't need it." All of this is to say that even on deep levels, the best probable view of the universe will be Bayesian in some sense. This means that plausibility is unavoidable when reasoning about the world.
What I'm saying is not that logicians should arrive at their conclusions through Bayesian reasoning, merely that there's always some level that involves plausibility. That doesn't mean that there aren't epistemic levels where same-level plausibility shouldn't be considered, however.
An example that pops to mind is that of the court and criminal justice systems. In the United States and many other parts of the Western world, a very important aspect of the criminal justice system is "innocent until proven guilty", or Presumption of Innocence. I will say that while I don't think that this is particularly epistemically sound, I agree wholeheartedly that this is how the actual execution of law should be carried out. I think that the empirical position that should be taken in a situation devoid of evidence is one of agnosticism, or a lack of knowledge; however, this should be treated as the equivalent of innocence under the law, and that is, I think, the intent behind the law. This is a perfect situation where the plausibility of guilt should not really be considered on the level of criminal justice. There should be hard and conclusive evidence demonstrating guilt before that ruling is administered, and that evidence should be conclusive from the starting point of neutrality, no matter how high the prior statistical plausibility is (consider, say, a case involving an individual belonging to an of the ethnic minorities that commit much more of a type of crime on average). The plausibility comes into play with the epistemic problems of evidence-gathering in the first place, but it is used only extremely sparingly when considering guilt on the actual level of criminal justice.
Similarly, I actually think that this should be the goal in social reasoning as well. By this, I mean the assumption of guilt on a social level, or that equivalent. In this case, I don't think that it should actually be illegal to consider plausibility in the Bayesian sense, but I do think that it should be a societal goal to eliminate at least the extremes in this case. I care much more about avoiding the creation of a social pariah from an "innocent" person than the missing of a "guilty" person.
In some senses, these are anti-utilitarian arguments, and I understand that various strains of utilitarianism can be very valuable even in social settings. I wouldn't, for example, want to seriously risk ruining a million lives just because I didn't feel that my evidence for guilt wasn't absolute and thus didn't justify treating a hypothetical person as being guilty. I don't think it's as simple as a calculation, however. There are values other than sheer maximization of intelligent happiness and minimization of intelligent suffering that must be weighed. I guess I'll have to go with the better-feeling, yet vaguer "flourishing" versus "suffering".
For this very reason, discussions about social justice on individual and even lager scales can be very difficult to have. Part of the problem will inevitably come from conflicting values. Unfortunately, morals and ethical axioms aren't objective, no matter how hard some people try to claim that they are. Even putting generally unreasonably moral systems like religions aside, some people will simply disagree that we should value the freedom of the innocent more than the imprisonment of the guilty. I find that this opinion often comes from a somewhat immature (in my opinion!) reciprocal justice view of the world, but that's not necessarily the case. There are many perfectly valid, naturalistic ethical views that conflict. That doesn't mean that progress can't be made, however. I think it should behoove us (as intelligent, sentient beings) to engage in critical discourse and to come to the best consensus that we can.
What I'm saying is not that logicians should arrive at their conclusions through Bayesian reasoning, merely that there's always some level that involves plausibility. That doesn't mean that there aren't epistemic levels where same-level plausibility shouldn't be considered, however.
An example that pops to mind is that of the court and criminal justice systems. In the United States and many other parts of the Western world, a very important aspect of the criminal justice system is "innocent until proven guilty", or Presumption of Innocence. I will say that while I don't think that this is particularly epistemically sound, I agree wholeheartedly that this is how the actual execution of law should be carried out. I think that the empirical position that should be taken in a situation devoid of evidence is one of agnosticism, or a lack of knowledge; however, this should be treated as the equivalent of innocence under the law, and that is, I think, the intent behind the law. This is a perfect situation where the plausibility of guilt should not really be considered on the level of criminal justice. There should be hard and conclusive evidence demonstrating guilt before that ruling is administered, and that evidence should be conclusive from the starting point of neutrality, no matter how high the prior statistical plausibility is (consider, say, a case involving an individual belonging to an of the ethnic minorities that commit much more of a type of crime on average). The plausibility comes into play with the epistemic problems of evidence-gathering in the first place, but it is used only extremely sparingly when considering guilt on the actual level of criminal justice.
Similarly, I actually think that this should be the goal in social reasoning as well. By this, I mean the assumption of guilt on a social level, or that equivalent. In this case, I don't think that it should actually be illegal to consider plausibility in the Bayesian sense, but I do think that it should be a societal goal to eliminate at least the extremes in this case. I care much more about avoiding the creation of a social pariah from an "innocent" person than the missing of a "guilty" person.
In some senses, these are anti-utilitarian arguments, and I understand that various strains of utilitarianism can be very valuable even in social settings. I wouldn't, for example, want to seriously risk ruining a million lives just because I didn't feel that my evidence for guilt wasn't absolute and thus didn't justify treating a hypothetical person as being guilty. I don't think it's as simple as a calculation, however. There are values other than sheer maximization of intelligent happiness and minimization of intelligent suffering that must be weighed. I guess I'll have to go with the better-feeling, yet vaguer "flourishing" versus "suffering".
For this very reason, discussions about social justice on individual and even lager scales can be very difficult to have. Part of the problem will inevitably come from conflicting values. Unfortunately, morals and ethical axioms aren't objective, no matter how hard some people try to claim that they are. Even putting generally unreasonably moral systems like religions aside, some people will simply disagree that we should value the freedom of the innocent more than the imprisonment of the guilty. I find that this opinion often comes from a somewhat immature (in my opinion!) reciprocal justice view of the world, but that's not necessarily the case. There are many perfectly valid, naturalistic ethical views that conflict. That doesn't mean that progress can't be made, however. I think it should behoove us (as intelligent, sentient beings) to engage in critical discourse and to come to the best consensus that we can.
No comments:
Post a Comment