Intro: To entertain the thought corresponding to the proposition 'The Beaujolais is better than the Cabernet Shiraz' you need to be (mentally) representing, at least: i) the beaujolais, ii) the cabernet shiraz, iii) some predicate BETTER_THAN with two arguments; and possibly the whole thing iv) BETTER_THAN (the beaujolais, the cabernet shiraz). Sam discusses this idea in his 2001 paper for CSS.
Question: What could in principle have truth conditions? At least, I am sure propositions do (they'd better!). How about MRs? MRs (i-iii) above seem not to 'accept' truth conditions, they can surely mis-represent (the beaujolais is actually the cabernet shiraz, but I inadvertently switched glasses). This fact, of course, does not necessarily imply that they are true or false, does it?. How about Mr (iv)? Is MR (iv) the same thing as the proposition? If not, what's the difference?
Buncha terminological spagetti. Not all MRs have truth conditions, especially not those coming from imagery. Granted, some people I read did bite the bullet and went in looking for logical representations within imagery and others extended imagery to include truth-toting propositions. In MneMonic (my attempt at architecturizing cognition =), everything's imitation. That is, context-bound matching to imperfectly retrieved analogical bases.
But in the case of MR truth values a la MITECS, I betcha your (iv) choice is the one proposition that has truth values. Their analogy, if I recall well, is made with morphology. Words don't have meaning in and of themselves, only as part of sentences, and sentences can (supposedly) be broken down to atomic propositions that have truth values... Just the same, predicates and [other] slot fillers can't get truth values by themselves. I say it all depends on the complexity of the MR.
Let's take an example: I tell you 'ball' (which is a noun), and ask you to answer with True or False. Chances are you'll call me crazy. But if you're a Muslim extremist and I tell you 'christianity' (which is also a noun), and ask you to answer True or False, my working hypothesis is that you'll have a very strong opinion with a very low RT.
Only a proposition or something that has the same degree of articulation as a proposition (e.g., Rob's favourite example, a subsentential assertion, i.e., something less than a sentence that still says something about something) can be true or false. Individual lexical items can fail in other ways -- misidentify, fail to refer -- but they cannot 'say what is not so', as Aristotle already put it, because they do no say.
MRs can be true or false when they have the requisite structure. E.g., one can have false beliefs, false memories. Whether this structure *simply is* propositional structure is, of course, just the biggest issue in Cog Sci right now, being the central issue between connectionists and physical symbol system theorists, at least PSSs of the sententialist stripe.
The following is just a glimpse of why I think propositional logic is overrated, and why I think that if one is willing to accept truth values, that one should accept them for any kind of mental representation.
Andy: Overrated for what? For determining what follows from a proposition or group of propositions, what else is there?
Radu: Playing the only game in town again, eh? Well, I say there are other games to be played. Here are some of the ones I looked at during the course on Thinking, Reasoning and Problem Solving:
Back to my glimpse. Aristotle did not consider the lack of something as a negation thereof. And neither would common sense allow us to. As a result, some people do not include inhibitory links in their cognitive architectures (whether logic, semantic or connectionist). But don't you consider that obvious omissions and choice among alternatives results in a more complex logical pattern?
Andy: What does this have to do with Leo's question, or with truth and falsity for that matter?
Radu: I was suggesting that truth conditions are unfit to describe any form of real-life thinking beyond formal thinking on artificial toy-worlds. The paradigm(s) of formal logic, by imposing a binary true-false straight jacket on thinking has done lots of harm:
When people stop thinking in terms of propositions and start thinking in terms of MRs, logic breaks down and can be replaced with more resource-intensive, but also more accurate exploratory methods. Methods like those used by mathematicians who explore a given ISpace (e.g. Geometry and subsets thereof), for a solution to a given problem.
Andy: Saying something false is not necessary negating anything.
Radu: I disagree. Saying any particular thing at all, making any statement negates everything that can be said that conflicts with that statement or contradicts it. For example "this particular door is white" negates "this particular door is black" and many other similar propositions.
E.g. by saying "this door is white" you are not only assigning a truth value to this proposition, but also affecting the truth values of other ones, like "this door is blue", "this is a door", "there are doors", etc.
Andy: True, but what does this have to do with the paragraph on Aristotle.
Radu: Is it clearer now?
|[ Current Projects ][ Current Projects ][ Portfolio ][ Pastime ][ Pseudonyms ][ CV ]|