The term 'gene' has undergone quite a bit of change in its history, so this isn't really all that surprising in light of this. The term was originally coined (probably by Mendel himself, but I don't remember) to mean roughly "whatever is responsible for the observable results of hybridization experiments" and later, with the advent of molecular biology, came to be shorthand for referring to a molecular structure of a certain kind. It's an interesting question of course, whether those definitions are coextensive (my bet is they aren't) and these latest findings are just evidence of a new conceptual (or at least terminological) shift.
See Stotz and Griffiths "Gene" (2005) (to appear in Cambridge companion to philosophy of biology, and also can be viewed online at http://philsci-archive.pitt.edu/archive/00002494/ )
I agree about PS:T -- definitely one of my all-time favorites. As to my "game of the year" though, I'd have to say Fallout (the first one). I haven't had a lot of time for gaming of late, but I recently picked it up again and started playing, and quickly remembered that it's one of the all-time great CRPGs. Blasting irradiated mutants with my shotgun and loving it...
A lot of other people have made helpful replies, so I'll just add this. Some philosophers of science in the 20th century have made some very interesting arguments about these kinds of possibilities. I'll mention three. First, Rudolph Carnap argues that, strictly speaking, it doesn't make any sense to think of the laws of logic as being true. Rather, they (or some other set of foundational axioms) must be taken as primitive--i.e. accepted as foundational rules, in order for a truth predicate to defined (in the Tarski-style way) in the first place. So, for Carnap, laws of logic are only true within the logical systems of which they are laws. It makes no sense to ask whether they are true independent of any systems of conventions for use of logic (his major work on this is "The Logical Syntax of Language", but be warned, it is dense and technical).
Willard van Orman Quine, a student of Carnap, vehemently disagreed with Carnap on this point (especially in the famous article "Two Dogmas of Empiricism", but also in "Carnap on Logical Truth" and "Truth by Convention"). Quine held that "our beliefs face the tribunal of experience as a corporate body" -- meaning that individual theories are not confirmed and disconfirmed by evidence but, rather, entire systems of belief, including accepted laws of logic, mathematics etc. If there is a conflict between accepted physical theory and observed results, we may give up the theory, or we may give up the observations or, and this is the crazy part, we may give up the system of logic responsible for the conflict. I realize this sounds pretty crazy to just state, so here's one other philosopher to consider...
Hilary Putnam argues, in a paper which I believe is entitled "Is Logic Empirical?", that certain anomalies in quantum mechanics (i.e. conflicts between predictions of the theory and observed results) could easily be obviated if we rejected the logical law of distribution (namely, that (A and B) or (A and C) = (A and (B or C) ). Putnam took this to suggest that perhaps we ought to reject classical logic (of which the law of distribution is a theorem) and replace it with a more restricted logical system, in which not all instances of the distributive law hold. The situation is comparable, Putnam argued, to the situation of pre-Einsteinien physicists with respect to Euclidean geometry--they took Euclid's axioms to be literally true claims about the geometry of physical space, and they were wrong. Similarly, it may be that while classical logic works in certain restricted domains, some of its laws just aren't literally true when they are taken to be claims about physical entities/events.
I hope that makes sense and is perhaps helpful--eggnog and philosophy of logic are not the best pair!
Well, strictly speaking, the set of current (or "final") physical laws cannot be the only consistent set of laws--if that set is consistent, then so are all of its subsets. And for that matter, the empty set is also consistent (and is a subset of every set anyway)... And this brings up a rather interesting point--what I think you want to say here is that the final set of physical laws will turn out to be the maximal consistent set of laws; a set such that it is consistent, but all of its proper supersets (i.e. sets that contain it as a part) are inconsistent.
In light of recent debates in the philosophy of science, this latter position makes sense, as it offers an easy solution to the "adding dwarves" problem of underdetermination. Namely, the problem is supposed to be that, for any theory T that is consistent with some set of data, there are an infinite of other theories that are also consistent with that data. To see this, just consider that we may simply offer a new theory, T*, that is simply all of the sentences of T plus an additional sentence that posits (for instance) some in-principle unobservable entity that is causally inert (e.g. T* says everything that T does, but also says that there are these particles, Occamons, that cannot be observed and do nothing). It's clear that T* is a distinct theory from T, but it's also clear that T is preferable to T*, and the trick is to come up with some principled reason to say why. Well, now if we demand that the "final" theory will be maximal consistent, this problem simply goes away; it's not possible to add any more sentences to this theory and end up with something consistent. Of course, this won't help with the Quinean suggestion that, if push comes to shove, empirical observations may lead us to reject logical assumptions too (e.g. that inconsistent sets of statements can never be true), but that's a puzzle for another day.
Wait, what? Today (12/16/07) it's true that he brushed his teeth on the morning of 12/16/07 and tomorrow it might be false that he brushed his teeth on the morning of 12/16/07? How's that work?
The propositions expressed by the statement 'I brushed my teeth this morning' may be such that today it is true and tomorrow it is false, but that's just because 'this morning' (and 'I' and 'my', for that matter) are indexicals (i.e. terms whose referent depends on the context of utterance) and hence, the propositions are different.
I agree with the parent--I don't see any reason not to include any proofs that people care to submit (as long as they actually are proofs--i.e. aren't fallacious or invalid). Different kinds of proofs are good for different kinds of things. Some proofs provide elegant verification of the result in question (I find many proofs by induction to be of this variety), while others tend to be much more explanatory in character--they can help us "see" why the result has to be the case. These categories are neither exclusive nor exhaustive, but it is often the case that the best explanatory proof is not the best justificatory proof and vice versa.
Another, particular-to-wikipedia, problem is what results one should take as previously proven in presenting proofs on Wikipedia. Strictly speaking, you can't offer proof of something if your proof depends on unproven results. Most textbooks can easily avoid this problem since they (the textbook editors), of course, have complete control over what results are proven in the text. This isn't a problem for justification of results, since no one is using Wikipedia as a platform for proving new theorems (as far as I know anyway), but it could make it more difficult for offered proofs to be helpful, since they may depend on results which themselves are not proved in wikipedia.
I don't mean to pick on you, specifically, but I think it's worth pointing out that the case of Galileo is significantly more complicated than that. Yes, Galileo was famously prosecuted by the church, but many historians now believe that this was based not on Galileo's astrophysical theory itself but, rather, his attempts to employ his theory in the interpretation of certain biblical passages--i.e. Galileo was prosecuted as a heretic not for doing science, but for engaging in questionable exegesis. I'm not claiming that this justifies or excuses the clergy here, but if this account is true, then we don't have a case of "science vs. religion" here; we have a case of religion vs. religion.
In addition to the religious persecution, Galileo's theory faced serious, and justified, scientific skepticism. First, at the time it was introduced, Galileo's heliocentric model was (1) not any simpler than the prevailing Ptolemaic model, (2) not any more accurate than the prevailing model (it made some correct predictions that weren't made by the geocentric picture, but also some incorrect predictions that were made by the geocentric model) and (3) not verifiably superior to the geocentric picture, given the state of measuring instruments. At the time that Galileo's theory was introduced, it must have looked to his contemporaries much like string theory looks to many current physicists: a best just an interesting new theory to consider, and at worst possibly not even a scientific theory at all. Additionally, many of Galileo's contemporaries were suspicious of the telescopes used to obtain some of the data that supposedly confirmed the theory. While this might look stupid or luddite in retrospect, at the time these devices were bleeding edge technology, and, furthermore, their workings were not adequately explained with a theory of optics. Given this, I posit that a bit of scientific skepticism with respect to the results gotten by these instruments was quite justified.
Of course, history has vindicated Galileo but, at the time, his theory didn't really look any better than its leading competitor. I wonder if we would care so much about the case of Galileo if it turned out that, in fact, the earth really is the center of the universe?
Fair enough, but there is plenty of new stuff out now wherein the whole album is a unified work. The old chestnut applies: Much of what was popular back in [insert musical Halcyon days of choice here] was total shit, just as much of what is popular now is total shit. Granted, I wasn't alive at the time, but I'd wager that KC wasn't *that* popular in the 70s (and, furthermore, I there are some other bands with good "album experiences" from the 70s that I *know* weren't popular, like Magma and Gentle Giant, for instance). Finally, to make good on the promise of my opening sentence, I'll suggest that you have a listen to the newest album, Mirrored, by Battles.
That's true, but that's like saying that every car is really nothing but four wheels, an engine and a chassis. Hence, everyone has the "same" car. It's true that everything digital can be encoded using a string binary digits, but it's not the case that this is the *only* relevant information (at a bare minimum, the order of that encoding makes a big difference too, and there's usually a lot more information "supervening" on the binary structure as well).
I agree--that's precisely the problem. If anything, the problem is worse than that since, even if your vote does happen to be in the majority, it's highly unlikely that it actually made a difference. The only situation in which your vote makes a difference, after all, is the situation in which things go differently than they would have had you not voted (or voted differently). And that situation is just when (say) candidate A and candidate B have precisely the same number of votes in favor of each, and your vote breaks the tie. In any other situation (and the odds are incredibly good that the situation will be other than a tie), your vote fails to make a difference.
I like the suggestion of incorporating probabilistic (or some other kind) of degrees into modal logic. Further, I grant your above arguments. In fact, given what you say here:
So given some numeric values x, y, and z, where x x, he chooses green over blue. The irrationality is still as I put forth earlier, in that by liking red he comes to dislike that which he chose red over, instead of remaining the same in that respect and just liking red more; it's just in relative rather than absolute terms now.
I think we're mostly in agreement. Your position, I take it, is that the irrationality involved in retrospectively rationalizing the choice (i.e. coming to dislike blue) is not logically required. I agree. However, and perhaps I misunderstood your first post, I thought you were also claiming that the preference of green over blue in the second choice was also irrational. I agree that it may be irrational, but that's only because it, as it were, inherits irrationality from the original rationalization event. My point was just that, in the situation where the monkey has (1) decided that red, green, and blue are equally valuable and (2) changed his mind about blue and decided that it is less valuable than red it is rational to (3) prefer the green to the blue. (To see this, suppose that, in (2), the monkey changes his mind for some rational reason--say, by testing a blue one and finding that it tastes bad.)
Else they were just typing random symbols and making meaningless noise grunts. But those people arguing they are merely typing random symbols and making meaningless noise grunts aren't demonstrating anything, aren't showing anything to be true or false, aren't differentiating between knowledge and the unknown (unfortunately, much of government funded academia, especially social sciences).
Suppose they were making "meaningless noise grunts". Were these true, or false? Think carefully...
*By definition* being "non-true" makes something false. If you *know* it's "non-true", you *know* it's false. "True" and "False" is strict EITHER/OR full set possibility.
I think you missed the point. There are some cases where truth and falsity just don't apply. Look at a bird. Is that thing true? Of course not, that's a category error; only linguistic entities (e.g. sentences, propositions) can be true or false and birds are not linguistic. Hence, a bird is non-true. But it's not false either, for the same category error reason. Furthermore, it's also not clear that every linguistic entity is either true or false. Consider Russell's famous example sentence: "The present King of France is bald." True, or false? I'm not sure it makes sense to say either, given that there is no present King of France and, hence, that the sentence fails to refer to anything that would make it either true or false. (I should point out that Russell analyses that case by claiming that the sentence is equivalent to the sentence "There is a present King of France, and he is bald" and is thus false, but there have been many serious criticisms and much debate over that analysis.)
Which brings us back on topic. The monkeys in this experiment were given the choice of red and blue and, choosing red but not-choosing blue (i.e. judging good(red) and not-good(blue)), in the same act chose not-blue (taking not-good(blue) to entail good(not-blue)), when they didn't logically have to to so. So later, presented with blue and green, they remained consistant with their earlier opinion that good(not-blue), when if they had been logical earlier they would have just seen a color they had not-chosen and another color they had not-chosen, rather than a color they had not-chosen and a color they had chosen-not.
Everything you say about the relevant modal logics above is clear and accurate; well-said. Now, I'm not sure I agree with your analysis of the case. First, recall that in the experimental setup, we're taking for granted that (at the beginning) equal preference is accorded to each of the red, green, and blue M&M's (might I write P(r)=P(g)=P(b), for convenience?). Given that, in your terms, I think we'd want to say that the monkey judges good(red) & good(green) & good(blue). Now, when we get to the choice event between red and blue, you claim that the monkey judges good(red) and not-good(blue). Notice, though, that the evidence cited doesn't tell us this--we don't know what the monkey is judging *at the time of choice*. Rather, the data indicates that *after* the choice event, the monkey re-evaluates and accords lesser preference to the blue.
Why is this temporal order relevant? Because, understood this way, there's no evidence to support the claim that the monkey erroneously judges that not-good(blue) entails good(not-blue). In this circumstance, this inference actually happens to work, since "not-good" here just indicates decreased preference. So, after the first choice event, the monkey "rationalizes" and we have that P(r)=P(g)>P(b) (since, recall, all were accorded equal preference in the experimental setup). But now, in this extremely limited domain, the inference from not-good(blue) to good(not-blue) actually goes through since all available options other than blue are accorded equally high preference. Given this, it looks like the monkey's preference for the green over the blue in the second choice event is the rational choice, since they now have the following information:
(1) Red is better than blue ("rationalization" after choice-event 1)
(2) Red is exactly as good as green (from equal preference in experimental setup)
(3) Hence, by substitution of "exactly as good as" (which certainly looks like an equivalence relation to me!) green is better than blue.
In sum, I think the "irrationality", if any, is located in the rationalization, and not in any faulty (tacit) modal reasoning since, as I argue, the (generally erroneous) modal inference which you attribute to the monkey actually does work in this limited domain. For all we know, the monkey may not be disposed to make such a modal inference in domains where it will fail (I think this is unlikely, but the point is that this experiment doesn't give us any evidence about this matter one way or the other).
Wow, that's a great point. I hadn't considered that problem before--when you put it that way, NN begins to look not only appealing, but quite pressing as well. That's a rather grim picture of the future of the (NN-less) internet that you paint...
Slashdot Mirror
Hah, well I just RTFA and it turns out that it was Johanssen that coined the term.
The term 'gene' has undergone quite a bit of change in its history, so this isn't really all that surprising in light of this. The term was originally coined (probably by Mendel himself, but I don't remember) to mean roughly "whatever is responsible for the observable results of hybridization experiments" and later, with the advent of molecular biology, came to be shorthand for referring to a molecular structure of a certain kind. It's an interesting question of course, whether those definitions are coextensive (my bet is they aren't) and these latest findings are just evidence of a new conceptual (or at least terminological) shift. See Stotz and Griffiths "Gene" (2005) (to appear in Cambridge companion to philosophy of biology, and also can be viewed online at http://philsci-archive.pitt.edu/archive/00002494/ )
I agree about PS:T -- definitely one of my all-time favorites. As to my "game of the year" though, I'd have to say Fallout (the first one). I haven't had a lot of time for gaming of late, but I recently picked it up again and started playing, and quickly remembered that it's one of the all-time great CRPGs. Blasting irradiated mutants with my shotgun and loving it...
I haven't laughed that hard in a long time--thanks.
A lot of other people have made helpful replies, so I'll just add this. Some philosophers of science in the 20th century have made some very interesting arguments about these kinds of possibilities. I'll mention three. First, Rudolph Carnap argues that, strictly speaking, it doesn't make any sense to think of the laws of logic as being true. Rather, they (or some other set of foundational axioms) must be taken as primitive--i.e. accepted as foundational rules, in order for a truth predicate to defined (in the Tarski-style way) in the first place. So, for Carnap, laws of logic are only true within the logical systems of which they are laws. It makes no sense to ask whether they are true independent of any systems of conventions for use of logic (his major work on this is "The Logical Syntax of Language", but be warned, it is dense and technical).
Willard van Orman Quine, a student of Carnap, vehemently disagreed with Carnap on this point (especially in the famous article "Two Dogmas of Empiricism", but also in "Carnap on Logical Truth" and "Truth by Convention"). Quine held that "our beliefs face the tribunal of experience as a corporate body" -- meaning that individual theories are not confirmed and disconfirmed by evidence but, rather, entire systems of belief, including accepted laws of logic, mathematics etc. If there is a conflict between accepted physical theory and observed results, we may give up the theory, or we may give up the observations or, and this is the crazy part, we may give up the system of logic responsible for the conflict. I realize this sounds pretty crazy to just state, so here's one other philosopher to consider...
Hilary Putnam argues, in a paper which I believe is entitled "Is Logic Empirical?", that certain anomalies in quantum mechanics (i.e. conflicts between predictions of the theory and observed results) could easily be obviated if we rejected the logical law of distribution (namely, that (A and B) or (A and C) = (A and (B or C) ). Putnam took this to suggest that perhaps we ought to reject classical logic (of which the law of distribution is a theorem) and replace it with a more restricted logical system, in which not all instances of the distributive law hold. The situation is comparable, Putnam argued, to the situation of pre-Einsteinien physicists with respect to Euclidean geometry--they took Euclid's axioms to be literally true claims about the geometry of physical space, and they were wrong. Similarly, it may be that while classical logic works in certain restricted domains, some of its laws just aren't literally true when they are taken to be claims about physical entities/events.
I hope that makes sense and is perhaps helpful--eggnog and philosophy of logic are not the best pair!
Well, strictly speaking, the set of current (or "final") physical laws cannot be the only consistent set of laws--if that set is consistent, then so are all of its subsets. And for that matter, the empty set is also consistent (and is a subset of every set anyway)... And this brings up a rather interesting point--what I think you want to say here is that the final set of physical laws will turn out to be the maximal consistent set of laws; a set such that it is consistent, but all of its proper supersets (i.e. sets that contain it as a part) are inconsistent.
In light of recent debates in the philosophy of science, this latter position makes sense, as it offers an easy solution to the "adding dwarves" problem of underdetermination. Namely, the problem is supposed to be that, for any theory T that is consistent with some set of data, there are an infinite of other theories that are also consistent with that data. To see this, just consider that we may simply offer a new theory, T*, that is simply all of the sentences of T plus an additional sentence that posits (for instance) some in-principle unobservable entity that is causally inert (e.g. T* says everything that T does, but also says that there are these particles, Occamons, that cannot be observed and do nothing). It's clear that T* is a distinct theory from T, but it's also clear that T is preferable to T*, and the trick is to come up with some principled reason to say why. Well, now if we demand that the "final" theory will be maximal consistent, this problem simply goes away; it's not possible to add any more sentences to this theory and end up with something consistent. Of course, this won't help with the Quinean suggestion that, if push comes to shove, empirical observations may lead us to reject logical assumptions too (e.g. that inconsistent sets of statements can never be true), but that's a puzzle for another day.
Wait, what? Today (12/16/07) it's true that he brushed his teeth on the morning of 12/16/07 and tomorrow it might be false that he brushed his teeth on the morning of 12/16/07? How's that work?
The propositions expressed by the statement 'I brushed my teeth this morning' may be such that today it is true and tomorrow it is false, but that's just because 'this morning' (and 'I' and 'my', for that matter) are indexicals (i.e. terms whose referent depends on the context of utterance) and hence, the propositions are different.
I agree with the parent--I don't see any reason not to include any proofs that people care to submit (as long as they actually are proofs--i.e. aren't fallacious or invalid). Different kinds of proofs are good for different kinds of things. Some proofs provide elegant verification of the result in question (I find many proofs by induction to be of this variety), while others tend to be much more explanatory in character--they can help us "see" why the result has to be the case. These categories are neither exclusive nor exhaustive, but it is often the case that the best explanatory proof is not the best justificatory proof and vice versa.
Another, particular-to-wikipedia, problem is what results one should take as previously proven in presenting proofs on Wikipedia. Strictly speaking, you can't offer proof of something if your proof depends on unproven results. Most textbooks can easily avoid this problem since they (the textbook editors), of course, have complete control over what results are proven in the text. This isn't a problem for justification of results, since no one is using Wikipedia as a platform for proving new theorems (as far as I know anyway), but it could make it more difficult for offered proofs to be helpful, since they may depend on results which themselves are not proved in wikipedia.
Straw Man.
There are three kinds of lies: lies, damned lies, and statistics. Nice work uncovering that info.
I don't mean to pick on you, specifically, but I think it's worth pointing out that the case of Galileo is significantly more complicated than that. Yes, Galileo was famously prosecuted by the church, but many historians now believe that this was based not on Galileo's astrophysical theory itself but, rather, his attempts to employ his theory in the interpretation of certain biblical passages--i.e. Galileo was prosecuted as a heretic not for doing science, but for engaging in questionable exegesis. I'm not claiming that this justifies or excuses the clergy here, but if this account is true, then we don't have a case of "science vs. religion" here; we have a case of religion vs. religion.
:)
In addition to the religious persecution, Galileo's theory faced serious, and justified, scientific skepticism. First, at the time it was introduced, Galileo's heliocentric model was (1) not any simpler than the prevailing Ptolemaic model, (2) not any more accurate than the prevailing model (it made some correct predictions that weren't made by the geocentric picture, but also some incorrect predictions that were made by the geocentric model) and (3) not verifiably superior to the geocentric picture, given the state of measuring instruments. At the time that Galileo's theory was introduced, it must have looked to his contemporaries much like string theory looks to many current physicists: a best just an interesting new theory to consider, and at worst possibly not even a scientific theory at all. Additionally, many of Galileo's contemporaries were suspicious of the telescopes used to obtain some of the data that supposedly confirmed the theory. While this might look stupid or luddite in retrospect, at the time these devices were bleeding edge technology, and, furthermore, their workings were not adequately explained with a theory of optics. Given this, I posit that a bit of scientific skepticism with respect to the results gotten by these instruments was quite justified.
Of course, history has vindicated Galileo but, at the time, his theory didn't really look any better than its leading competitor. I wonder if we would care so much about the case of Galileo if it turned out that, in fact, the earth really is the center of the universe?
Ok, now mod me off-topic.
Fair enough, but there is plenty of new stuff out now wherein the whole album is a unified work. The old chestnut applies: Much of what was popular back in [insert musical Halcyon days of choice here] was total shit, just as much of what is popular now is total shit. Granted, I wasn't alive at the time, but I'd wager that KC wasn't *that* popular in the 70s (and, furthermore, I there are some other bands with good "album experiences" from the 70s that I *know* weren't popular, like Magma and Gentle Giant, for instance). Finally, to make good on the promise of my opening sentence, I'll suggest that you have a listen to the newest album, Mirrored, by Battles.
...for sufficiently broad values of 'almost'.
And to think that they might have won if it weren't for their silly hats (seriously, a bird's claw on top of your helmet--wtf?).
I know it's a bit off-topic, but thanks for those links!
That's true, but that's like saying that every car is really nothing but four wheels, an engine and a chassis. Hence, everyone has the "same" car. It's true that everything digital can be encoded using a string binary digits, but it's not the case that this is the *only* relevant information (at a bare minimum, the order of that encoding makes a big difference too, and there's usually a lot more information "supervening" on the binary structure as well).
Yeah, did you see the pic in TFA? Note placement of dinosaur head relative to the person standing next to it...
I agree--that's precisely the problem. If anything, the problem is worse than that since, even if your vote does happen to be in the majority, it's highly unlikely that it actually made a difference. The only situation in which your vote makes a difference, after all, is the situation in which things go differently than they would have had you not voted (or voted differently). And that situation is just when (say) candidate A and candidate B have precisely the same number of votes in favor of each, and your vote breaks the tie. In any other situation (and the odds are incredibly good that the situation will be other than a tie), your vote fails to make a difference.
I think you're exactly right--that's an insightful way to look at UI design. I know that's how I'd like *my* UIs to work.
Suppose they were making "meaningless noise grunts". Were these true, or false? Think carefully...
Everything you say about the relevant modal logics above is clear and accurate; well-said. Now, I'm not sure I agree with your analysis of the case. First, recall that in the experimental setup, we're taking for granted that (at the beginning) equal preference is accorded to each of the red, green, and blue M&M's (might I write P(r)=P(g)=P(b), for convenience?). Given that, in your terms, I think we'd want to say that the monkey judges good(red) & good(green) & good(blue). Now, when we get to the choice event between red and blue, you claim that the monkey judges good(red) and not-good(blue). Notice, though, that the evidence cited doesn't tell us this--we don't know what the monkey is judging *at the time of choice*. Rather, the data indicates that *after* the choice event, the monkey re-evaluates and accords lesser preference to the blue.
Why is this temporal order relevant? Because, understood this way, there's no evidence to support the claim that the monkey erroneously judges that not-good(blue) entails good(not-blue). In this circumstance, this inference actually happens to work, since "not-good" here just indicates decreased preference. So, after the first choice event, the monkey "rationalizes" and we have that P(r)=P(g)>P(b) (since, recall, all were accorded equal preference in the experimental setup). But now, in this extremely limited domain, the inference from not-good(blue) to good(not-blue) actually goes through since all available options other than blue are accorded equally high preference. Given this, it looks like the monkey's preference for the green over the blue in the second choice event is the rational choice, since they now have the following information:
(1) Red is better than blue ("rationalization" after choice-event 1)
(2) Red is exactly as good as green (from equal preference in experimental setup)
(3) Hence, by substitution of "exactly as good as" (which certainly looks like an equivalence relation to me!) green is better than blue.
In sum, I think the "irrationality", if any, is located in the rationalization, and not in any faulty (tacit) modal reasoning since, as I argue, the (generally erroneous) modal inference which you attribute to the monkey actually does work in this limited domain. For all we know, the monkey may not be disposed to make such a modal inference in domains where it will fail (I think this is unlikely, but the point is that this experiment doesn't give us any evidence about this matter one way or the other).
Wow, that's a great point. I hadn't considered that problem before--when you put it that way, NN begins to look not only appealing, but quite pressing as well. That's a rather grim picture of the future of the (NN-less) internet that you paint...
I know -- I was just being a smartass.