Does this mean that when most people read the question, they think of marriage as being a binary predicate (i.e. relation): m : X x X --> {true, false} rather than as a unary predicate (i.e. property): m : X --> {true, false} and thus fail to resolve the question? It would be interesting to see what happens if you replace 'married' with 'wearing red trousers' (something which is intuitively a unary predicate rather than a binary predicate). I thought that the 'trick' was that there might be two Lindae in the question, in which case the following comic is pertinent: https://xkcd.com/169/ Best wishes, Adam P. Goucher
Sent: Friday, May 05, 2017 at 10:59 AM From: "Gareth McCaughan" <gareth.mccaughan@pobox.com> To: math-fun@mailman.xmission.com Subject: Re: [math-fun] Brain teaser
On 05/05/2017 09:36, Victor Miller wrote:
Is there some trick? If Linda is married, then the Linda-John pair is the married/not married pair. If Linda isn't married then the Paul/lLinda pair is the one.
The only trick is that a surprising number of people, even rather clever people, get only as far as figuring out that they can't tell whether L&J are married nor whether P&L are married, and conclude that the answer to the question is no.
I think it likely that training in mathematics correlates strongly with answering this correctly. I am not sure how the explanation for this, if I'm right, should be divided between the effect of learning about logic and the effect of being the sort of person who is willing to spend time learning about logic.
-- g
_______________________________________________ math-fun mailing list math-fun@mailman.xmission.com https://mailman.xmission.com/cgi-bin/mailman/listinfo/math-fun