Re: [agi] How do we know we don't know?
Mike: I wrote my last email in a rush. Basically what I was trying to explain is precisely the basis of what you call creative process in understanding words. I simplified the whole thing a lot because I did not even consider the various layers of mappings - mappings of mappings and so on. What you say is correct, the word art-cop will invoke various ideas, amongst which art - which in turn will evoke art-exhibit, painting, art-criticism, and whatever else you want to add. The word cop in analogy will evoke a series of concepts, and those concepts themselves will evoke more concepts and so on. Now obviously if there were no 'measuring' system to how much concepts are evoked among each other, this process would go no-where. But fortunately there is such a measuring process and simplifying things a lot, it consists of excitatory and inhibitory synapses, as well as overall disturbance or 'noise' which after so and so many transitions will make the signal lose its significance (i.e. become random for practical purposes). Hope this is not too confusing. I'm not that great at explaining my ideas with words :) Jim Vlad: that is a difficult question because it depends a lot on your database. Actually Marko Rodriguez has attempted this in a program that uses a database of related words from the University of South Florida. This program is able to understand very simple analogies such as Which word of the second list best fits in the first list? bear, cow, dog, tiger: turtle, carp, parrot, lion Obviously this program is very limited. If you just need to just search words correspondence, I'd go with Vlad's suggestion. Otherwise there is a lot to be implemented, in terms of layers, inhibitory vs excitatory connections, concept from stimuli and so on..What strikes me in AGI is that so many researchers try to build an AGI with the presupposition that everything should be built in already, the machine should be able to resolve tasks from day 1 - just like in AI. That's like expecting a new born baby to talk about political issues! It's easy to forget that the database we have in our brains, upon which we make decisions, selections, creations, and so on.. is incredibly large.. in fact it took a life-time to assembe. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
On Fri, Aug 1, 2008 at 12:16 PM, Valentina Poletti [EMAIL PROTECTED] wrote: Jim Vlad: that is a difficult question because it depends a lot on your database. Actually Marko Rodriguez has attempted this in a program that uses a database of related words from the University of South Florida. This program is able to understand very simple analogies such as Which word of the second list best fits in the first list? bear, cow, dog, tiger: turtle, carp, parrot, lion Obviously this program is very limited. If you just need to just search words correspondence, I'd go with Vlad's suggestion. Otherwise there is a lot to be implemented, in terms of layers, inhibitory vs excitatory connections, concept from stimuli and so on..What strikes me in AGI is that so many researchers try to build an AGI with the presupposition that everything should be built in already, the machine should be able to resolve tasks from day 1 - just like in AI. That's like expecting a new born baby to talk about political issues! It's easy to forget that the database we have in our brains, upon which we make decisions, selections, creations, and so on.. is incredibly large.. in fact it took a life-time to assembe. Hi Valentina, I'm not quite sure what you mean here, but to be on the safe side, did you internalize warnings given in e.g. ( http://www.overcomingbias.com/2008/07/detached-lever.html ), ( Drew McDermott's Artificial Intelligence Meets Natural Stupidity ), ( http://www.overcomingbias.com/2007/11/artificial-addi.html )? I tried to describe the physical origins of this disconnect between the shallowness of properties and tags we use to reason about real-world objects and exquisite level of detail in objects themselves in ( http://causalityrelay.wordpress.com/2008/07/06/rules-of-thumb/ ), ( http://causalityrelay.wordpress.com/2008/07/17/flow-of-reality/ ), ( http://causalityrelay.wordpress.com/2008/07/20/precise-answers/ ). My argument wasn't about word-matching per se, but about the high-level characterization of the process of reasoning/perception/recall/problem-solving. You can't paint a word Intelligence in big letters and expect it to start doing intelligent things. -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.com/ --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
This is how I explain it: when we perceive a stimulus, word in this case, it doesn't reach our brain as a single neuron firing or synapse, but as a set of already processed neuronal groups or sets of synapses, that each recall various other memories, concepts and neuronal group. Let me clarify this. In the example you give, the wod artcop might reach us as a set of stimuli: art, cop, mediu-sized word, word that begins with a, and so on. All these connect activate various maps in our memory, and if something substantial is monitored at some point (going with Richard's theory of the monitor, I don't have other references of this actually), we form a response. This is more obvious in the case of sight - where an image is first broken into various compontents that are separately elaborated: colours, motion, edges, shapes, etc. - and then further sent to the upper parts of the memory where they can be associated to higher level concepts. If any of this is not clear let me know, instead of adding me to your kill-lists ;-P Valentina On 7/31/08, Mike Tintner [EMAIL PROTECTED] wrote: Vlad: I think Hofstadter's exploration of jumbles ( http://en.wikipedia.org/wiki/Jumble ) covers this ground. You don't just recognize the word, you work on trying to connect it to what you know, and if set of letters didn't correspond to any word, you give up. There's still more to word recognition though than this. How do we decide what is and isn't, may or may not be a word? A neologism? What may or may not be words from: cogrough dirksilt thangthing artcop coggourd cowstock or fomlepaung or whatever? --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com -- A true friend stabs you in the front. - O. Wilde Einstein once thought he was wrong; then he discovered he was wrong. For every complex problem, there is an answer which is short, simple and wrong. - H.L. Mencken --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Vladimir Nesov wrote: I think Hofstadter's exploration of jumbles ( http://en.wikipedia.org/wiki/Jumble ) covers this ground. You don't just recognize the word, you work on trying to connect it to what you know, and if set of letters didn't correspond to any word, you give up. This establishes deep similarity between problem-solving, perception and memory, and poses deliberative reasoning as iterative application of reflexive perception-steps. If you think the question and it gives you an answer, you can act on it. If it doesn't, the context in which you thought the question, deliberative program starting the request, will produce I don't know... response. It's probably as simple as that: a higher level of organization, not fundamental to the structure of mind, learned behavior. Agreed: Hofstadter's Jumbo system was inspirational to me when i read it in 1986/7, and that idea of relaxation is exactly what was behind the descriptions that I gave, earlier in this thread, of systems that tried to do recognition and question answering by constraint relaxation. Richard Loosemore --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Mike, Valentina was referring to a remark I made (and shouldn't have -- just on general principles) about her making my *personal* kill-list thanks to the LOL she left regarding Richard Loosemore's original reply to the post that started this thread. I should have taken a time out before I opened my big fingers. Had I done so, I would have found out (as I did through subsequent exchanges with Richard) that his comments were based on a misunderstanding. He thought what I was calling the list of things we don't know was a list of all things not known. It wasn't. I was referring to the list of things we know we don't know. I take full responsibility for creating this misunderstanding through sloppy writing/editing. Anyhow, I took Richard's initial comments the wrong way (probably because I'm as insecure as the next person). Valentina's message got read in that context. The misunderstanding has all been worked out now, so there was really no reason for all the initial drama. Valentina: if you're reading this, I apologize for overreacting. I re-read your post after I'd calmed down and realized that you did add a brief comment on Richard's reply. You didn't just pile on. I look forward to hearing more about your views on building an AGI. I'm happy to see this thread has generated some interesting side discussions. I'm here to learn and, occasionally, see what people who give a lot of time and thought to this subject think of my whacky ideas. Cheers, Brad Mike Tintner wrote: Er no, I don't believe in killing people :) I'm not quite sure what you're what getting at. I was just trying to add another layer of complexity to the brain's immensely multilayered processing. Our processing of new words/word combinations shows that there is a creative aspect to this processing - it isn't just matching. Some of this might be done by standard verbal associations/ semantic networks - e.g. yes IMO artcop could be a word for, say, art critic - cops police, and art can be seen as being policed - I may even have that last expression in memory. But in other cases, the processing may have to be done by imaginative association/drawing - dirksilt could just conceivably be a word, if I imagine some dirk/dagger-like tool being used on silt, (doesn't make much sense but conceivable for my brain) - I doubt that such reasoning could be purely verbal. Valentina: This is how I explain it: when we perceive a stimulus, word in this case, it doesn't reach our brain as a single neuron firing or synapse, but as a set of already processed neuronal groups or sets of synapses, that each recall various other memories, concepts and neuronal group. Let me clarify this. In the example you give, the wod artcop might reach us as a set of stimuli: art, cop, mediu-sized word, word that begins with a, and so on. All these connect activate various maps in our memory, and if something substantial is monitored at some point (going with Richard's theory of the monitor, I don't have other references of this actually), we form a response. This is more obvious in the case of sight - where an image is first broken into various compontents that are separately elaborated: colours, motion, edges, shapes, etc. - and then further sent to the upper parts of the memory where they can be associated to higher level concepts. If any of this is not clear let me know, instead of adding me to your kill-lists ;-P On 7/31/08, *Mike Tintner* [EMAIL PROTECTED] wrote: Vlad: I think Hofstadter's exploration of jumbles ( http://en.wikipedia.org/wiki/Jumble ) covers this ground. You don't just recognize the word, you work on trying to connect it to what you know, and if set of letters didn't correspond to any word, you give up. There's still more to word recognition though than this. How do we decide what is and isn't, may or may not be a word? A neologism? What may or may not be words from: cogrough dirksilt thangthing artcop coggourd cowstock or fomlepaung or whatever? *agi* | Archives https://www.listbox.com/member/archive/303/=now https://www.listbox.com/member/archive/rss/303/ | Modify https://www.listbox.com/member/?; Your Subscription [Powered by Listbox] http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Wow! The civility level on this list is really bottoming out . . . . along with any sort of scientific grounding. I have to agree with both Valentina and Richard . . . . since they are supported by scientific results while others are merely speculating without basis. Experimental (imaging) evidence shows that known words will strongly activate some set of neurons when heard. Unknown words with recognizable parts/features will also activate some other set of neurons when heard, possibly allowing the individual to puzzle out the meaning even if the word has never been heard before. Totally unknown words will not strongly activate any neurons -- except subsequently (i.e. on a delay) some set of HUH? neurons. If you wish, you can consider this to be an analogue of a massively parallel search carried out by the subconscious but it's really just an automatic operation. Recognized word == activated neurons bringing it's meaning forward through spreading activation. Totally unrecognized word == no activated neurons which is then interpreted as I don't know this word. Ed's response (which you praised), while a nice fanciful story that might work in another universe, is *not* supported by any evidence and is contra-indicated by a reasonable amount of experimental evidence. - Original Message - From: Brad Paulsen [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Tuesday, July 29, 2008 7:33 PM Subject: Re: [agi] How do we know we don't know? Valentina, Well, the LOL is on you. Richard failed to add anything new to the two previous responses that each posited linguistic surface feature analysis as being responsible for generate the feeling of not knowing with that *particular* (and, admittedly poorly-chosen) example query. This mechanism will, however, apply to only a very tiny number of cases. In response to those first two replies (not including Richard's), I apologized for the sloppy example and offered a new one. Please read the entire thread and the new example. I think you'll find Richard's and your explanation will fail to address how the new example might generate the feeling of not knowing. Cheers, Brad Valentina Poletti wrote: lol.. well said richard. the stimuli simply invokes no signiticant response and thus our brain concludes that we 'don't know'. that's why it takes no effort to realize it. agi algorithms should be built in a similar way, rather than searching. Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I note that even in the simplest word-recognition neural nets that I built and studied in the 1990s, activation of a nonword proceeded in a very different way than activation of a word: it would have been easy to build something to trigger a this is a nonword neuron. Is there some type of AI formalism where nonword recognition would be problematic? Richard Loosemore *agi* | Archives https://www.listbox.com/member/archive/303/=now https://www.listbox.com/member/archive/rss/303/ | Modify https://www.listbox.com/member/?; Your Subscription [Powered by Listbox] http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
On Wed, Jul 30, 2008 at 9:50 AM, Mark Waser [EMAIL PROTECTED] wrote: Wow! The civility level on this list is really bottoming out . . . . along with any sort of scientific grounding. Experimental (imaging) evidence shows that known words will strongly activate some set of neurons when heard. Unknown words with recognizable parts/features will also activate some other set of neurons when heard, possibly allowing the individual to puzzle out the meaning even if the word has never been heard before. Totally unknown words will not strongly activate any neurons -- except subsequently (i.e. on a delay) some set of HUH? neurons. Well, your imaging evidence is part imaging and part imagining since no one knows what the imaging is actually showing. I think it is commonly believed that the imaging techniques show blood flow into areas of the brain, and this is (reasonably in my view) taken as evidence of neural activity. Ok, but what kind of thinking is actually going on and how extensive are the links that don't have enough wow factor for repeatable experiments researchers to issue as a press release. So if you are going to claim that you're speculations are superiorly grounded, I would like to see some research that shows that unknown words will not strongly activate any neurons. Take your time, I am only asking a question, not challenging you to fantasy combat. Jim Bromer --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Brad Paulsen wrote: Richard Loosemore wrote: Brad Paulsen wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept or event is a product of the Danger, Will Robinson!, reptilian part of our brain. When we don't know we don't know something we react with a feeling of not knowing as a survival response. Then, having survived, we put the thing not known at the head of our list of things I don't know. As long as that thing is in this list it explains how we can come to the feeling of not knowing it so quickly. Of course, keeping a large list of things I don't know around is probably not a good idea. I suspect such a list will naturally get smaller through atrophy. You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. And... Another intuition tells me that the list of things I don't know, might generate a certain amount of cognitive dissonance the resolution of which can only be accomplished by seeking out new information (i.e., learning)? If so, does this mean that such a list in an AGI could be an important element of that AGI's desire to learn? From a functional point of view, this could be something as simple as a scheduled background task that checks the things I don't know list occasionally and, under the right circumstances, pings the AGI with a pang of cognitive dissonance from time to time. So, what say ye? Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I note that even in the simplest word-recognition neural nets that I built and studied in the 1990s, activation of a nonword proceeded in a very different way than activation of a word: it would have been easy to build something to trigger a this is a nonword neuron. Is there some type of AI formalism where nonword recognition would be problematic? Richard Loosemore Richard, You seem to have decided my request for comment was about word (mis)recognition. It wasn't. Unfortunately, I included a misleading example in my initial post. A couple of list members called me on it immediately (I'd expect nothing less from this group -- and this was a valid criticism duly noted). So far, three people have pointed out that a query containing an un-common (foreign, slang or both) word is one way to quickly generate the feeling of not knowing. But, it is just that: only one way. Not all feelings of not knowing are produced by linguistic analysis of surface features. In fact, I would guess that the vast majority of them are not so generated. Still, some are and pointing this out was a valid contribution (perhaps that example was fortunately bad). I don't think my query is a no-brainer to answer (unless you want to make it one) and your response, since it contained only another flavor of the previous two responses, gives me no reason whatsoever to change my opinion. Please take a look at the revised example in this thread. I don't think it has the same problems (as an example) as did the initial example. In particular, all of the words are common (American English) and the syntax is valid. Well, no, I did understand
Re: [agi] How do we know we don't know?
Brad Paulsen wrote: Valentina, Well, the LOL is on you. Richard failed to add anything new to the two previous responses that each posited linguistic surface feature analysis as being responsible for generate the feeling of not knowing with that *particular* (and, admittedly poorly-chosen) example query. This mechanism will, however, apply to only a very tiny number of cases. In response to those first two replies (not including Richard's), I apologized for the sloppy example and offered a new one. Please read the entire thread and the new example. I think you'll find Richard's and your explanation will fail to address how the new example might generate the feeling of not knowing. Brad, Isn't this response, as well as the previous response directed at me, just a little more annoyed-sounding than it needs to be? Both Valentina and I (and now Mark Waser also) have simply focused on the fact that it is relatively trivial to build mechanisms that monitor the rate at which the system is progressing in its attempt to do a recognition operation, and then call it as a not known if the progress rate is below a certain threshold. In particular, you did suggest the idea of a system keeping lists of things it did not know, and surely it is not inappropriate to give a good-naturedly humorous response to that one? So far, I don't see any of us making a substantial misunderstanding of your question, nor anyone being deliberately rude to you. Richard Loosemore Valentina Poletti wrote: lol.. well said richard. the stimuli simply invokes no signiticant response and thus our brain concludes that we 'don't know'. that's why it takes no effort to realize it. agi algorithms should be built in a similar way, rather than searching. Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I note that even in the simplest word-recognition neural nets that I built and studied in the 1990s, activation of a nonword proceeded in a very different way than activation of a word: it would have been easy to build something to trigger a this is a nonword neuron. Is there some type of AI formalism where nonword recognition would be problematic? Richard Loosemore *agi* | Archives https://www.listbox.com/member/archive/303/=now https://www.listbox.com/member/archive/rss/303/ | Modify https://www.listbox.com/member/?; Your Subscription[Powered by Listbox] http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Jim Bromer wrote: On Wed, Jul 30, 2008 at 9:50 AM, Mark Waser [EMAIL PROTECTED] wrote: Wow! The civility level on this list is really bottoming out . . . . along with any sort of scientific grounding. Experimental (imaging) evidence shows that known words will strongly activate some set of neurons when heard. Unknown words with recognizable parts/features will also activate some other set of neurons when heard, possibly allowing the individual to puzzle out the meaning even if the word has never been heard before. Totally unknown words will not strongly activate any neurons -- except subsequently (i.e. on a delay) some set of HUH? neurons. Well, your imaging evidence is part imaging and part imagining since no one knows what the imaging is actually showing. I think it is commonly believed that the imaging techniques show blood flow into areas of the brain, and this is (reasonably in my view) taken as evidence of neural activity. Ok, but what kind of thinking is actually going on and how extensive are the links that don't have enough wow factor for repeatable experiments researchers to issue as a press release. So if you are going to claim that you're speculations are superiorly grounded, I would like to see some research that shows that unknown words will not strongly activate any neurons. Take your time, I am only asking a question, not challenging you to fantasy combat. There are many such studies. It appears that activation is strong in semantic areas when words are involved, but only in the phoneme-grapheme mapping areas when nonwords are involved. If something in the brain is monitoring the strength of activation in the semantic area, it would be able to extract a feeling of knowing signal. There are also studies of feeling of knowing in episodic memory, and also work on being able to distinguish syntactically correct sentences from incorrect ones. The common thread in all of these studies is that gross differences of processing can be found between recognition of known and unknown items, whether those items be at the word level or higher. And the common interpretation of these results seems to be that strong activation occurs in the expected area when the item is known which means that it would be easy for the system to conclude that the *absence* of such strong activation can be taken to mean that the item is not known. I have copied one example of such a paper below (title and abstract only). This is about the word-nonword case, but as I have said before, the interpretation does generalize easily to higher cases such as recognizing that you do not know the answer to a question. Richard Loosemore Neural Correlates of Lexical Access During Visual Word Recognition, Binder, J.R., McKiernan, K.A., Parsons, M.E. , Westbury, C.F., Possing, E.T., Kaufman, J.N., Buchanan, L.J., Cogn. Neurosci..2003; 15: 372-393 People can discriminate real words from nonwords even when the latter are orthographically and phonologically word-like, presumably because words activate specific lexical and/or semantic information. We investigated the neural correlates of this identification process using event-related functional magnetic resonance imaging (fMRI). Participants performed a visual lexical decision task under conditions that encouraged specific word identification: Nonwords were matched to words on orthographic and phonologic characteristics, and accuracy was emphasized over speed. To identify neural responses associated with activation of nonsemantic lexical information, processing of words and nonwords with many lexical neighbors was contrasted with processing of items with no neighbors. The fMRI data showed robust differences in activation by words and word-like nonwords, with stronger word activation occurring in a distributed, left hemisphere network previously associated with semantic processing, and stronger nonword activation occurring in a posterior inferior frontal area previously associated with grapheme-to-phoneme mapping. Contrary to lexicon-based models of word recognition, there were no brain areas in which activation increased with neighborhood size. For words, activation in the left prefrontal, angular gyrus, and ventrolateral temporal areas was stronger for items without neighbors, probably because accurate responses to these items were more dependent on activation of semantic information. The results show neural correlates of access to specific word information. The absence of facilitatory lexical neighborhood effects on activation in these brain regions argues for an interpretation in terms of semantic access. Because subjects performed the same task throughout, the results are unlikely to be due to task-specific attentional, strategic, or expectancy effects. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed:
Re: [agi] How do we know we don't know?
Abram, The syntactic surface feature argument makes a good, but rather narrow, addition to the list of mechanisms that can engender a feeling of not knowing. The interesting part is that, as someone who speaks Norwegian, using that word in the example didn't set off phonological feature alarms for me. Non-Norwegian speakers picked it up right off. Your argument for semantic (i.e., meaning) features lacks concrete examples so it is difficult to tell exactly what you mean. Based on your general argument, I would conclude that it requires a content search of some sort and, therefore, falls under one of the mechanisms posited in my initial post. Cheers, Brad Abram Demski wrote: I think the same sort of solution applies to the world series case; the only difference is that it is semantic features that fail to combine, rather than syntactic. In other words, there are either zero associations or none with the potential to count as an answer. --Abram On Tue, Jul 29, 2008 at 7:51 PM, Brad Paulsen [EMAIL PROTECTED] wrote: Matt, I confess, I'm not sure I understand your response. It seems to be a variant of the critique made by three people early-on in this thread based on the misleading example query in my original post. These folks noted that an analysis of linguistic surface features (i.e., the word fomlepung would not sound right to an English speaking query recipient) could account for the feeling of not knowing. And they were right. For queries of that type (i.e., queries that contained foreign, slang or uncommon words). I apologized for that first example and provided an improved query (one that has valid English syntax and uses common English words -- so it will pass linguistic surface feature analysis). To wit: Which team won the 1924 World Series? Cheers, Brad Matt Mahoney wrote: This is not a hard problem. A model for data compression has the task of predicting the next bit in a string of unknown origin. If the string is an encoding of natural language text, then modeling is an AI problem. If the model doesn't know, then it assigns a probability of about 1/2 to each of 0 and 1. Probabilities can be easily detected from outside the model, regardless of the intelligence level of the model. -- Matt Mahoney, [EMAIL PROTECTED] --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Sure, search is at the root of all processing, be it human or AI. How we each go about the search, and how efficient we are at the task are different, and what exactly we are searching for, and exponential explosion. But some type of search is done, whether we are consciously aware of our brains doing the search or not. Given a bit of context information about the question should allow us to use some heuristics to look at a smaller area of knowledge bases in our brains, or in a computer's memory. Having a list of things we dont know is nonsensical as has been pointed out, when it comes to individual terms, but something like a aggregate estimate of knowledge known could be computed. I myself know a little about baseball say 10%, but baseball history and world series statistics would be more like 0.1% James Ratcliff --- On Tue, 7/29/08, Brad Paulsen [EMAIL PROTECTED] wrote: James, So, you agree that some sort of search must take place before the feeling of not knowing presents itself? Of course, realizing we don't have a lot of information results from some type of a search and not a separate process (at least you didn't posit any). Thanks for your comments! Cheers Brad James Ratcliff wrote: It is fairly simple at that point, we have enough context to have a very limited domain world series - baseball 1924 answer is a team, so we can do a lookup in our database easily enough, or realize that we really dont have a lot of information about baseball in our mindset. And for the other one, it would just be a strait term match. James Ratcliff ___ James Ratcliff - http://falazar.com Looking for something... --- On *Mon, 7/28/08, Brad Paulsen /[EMAIL PROTECTED]/* wrote: From: Brad Paulsen [EMAIL PROTECTED] Subject: Re: [agi] How do we know we don't know? To: agi@v2.listbox.com Date: Monday, July 28, 2008, 4:12 PM Jim Bromer wrote: On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. Brad My guess that initial recognition must be based on the surface features of an input. If this is true, then that could suggest that our initial recognition reactions are stimulated by distinct components (or distinct groupings of components) that are found in the surface input data. Jim Bromer Hmmm. That particular query may not have been the best example since, to a non-Norwegian speaker, the phonological surface feature of that statement alone could account for the feeling of not knowing. In other words, the word fomlepung just doesn't sound right. Good point. But, that may only explain how we know we don't know strange sounding words. Let's try another example: Which team won the 1924 World Series? Cheers, Brad --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com *agi* | Archives https://www.listbox.com/member/archive/303/=now https://www.listbox.com/member/archive/rss/303/ | Modify https://www.listbox.com/member/?; Your Subscription [Powered by Listbox] http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member
Re: [agi] How do we know we don't know?
Richard Loosemore wrote: Brad Paulsen wrote: Richard Loosemore wrote: Brad Paulsen wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept or event is a product of the Danger, Will Robinson!, reptilian part of our brain. When we don't know we don't know something we react with a feeling of not knowing as a survival response. Then, having survived, we put the thing not known at the head of our list of things I don't know. As long as that thing is in this list it explains how we can come to the feeling of not knowing it so quickly. Of course, keeping a large list of things I don't know around is probably not a good idea. I suspect such a list will naturally get smaller through atrophy. You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. And... Another intuition tells me that the list of things I don't know, might generate a certain amount of cognitive dissonance the resolution of which can only be accomplished by seeking out new information (i.e., learning)? If so, does this mean that such a list in an AGI could be an important element of that AGI's desire to learn? From a functional point of view, this could be something as simple as a scheduled background task that checks the things I don't know list occasionally and, under the right circumstances, pings the AGI with a pang of cognitive dissonance from time to time. So, what say ye? Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I note that even in the simplest word-recognition neural nets that I built and studied in the 1990s, activation of a nonword proceeded in a very different way than activation of a word: it would have been easy to build something to trigger a this is a nonword neuron. Is there some type of AI formalism where nonword recognition would be problematic? Richard Loosemore Richard, You seem to have decided my request for comment was about word (mis)recognition. It wasn't. Unfortunately, I included a misleading example in my initial post. A couple of list members called me on it immediately (I'd expect nothing less from this group -- and this was a valid criticism duly noted). So far, three people have pointed out that a query containing an un-common (foreign, slang or both) word is one way to quickly generate the feeling of not knowing. But, it is just that: only one way. Not all feelings of not knowing are produced by linguistic analysis of surface features. In fact, I would guess that the vast majority of them are not so generated. Still, some are and pointing this out was a valid contribution (perhaps that example was fortunately bad). I don't think my query is a no-brainer to answer (unless you want to make it one) and your response, since it contained only another flavor of the previous two responses, gives me no reason whatsoever to change my opinion. Please take a look at the revised example in this thread. I don't think it has the same problems (as an example) as did the initial example. In particular, all of the words are common (American English) and the syntax is valid.
Re: [agi] How do we know we don't know?
Brad Paulsen wrote: Richard Loosemore wrote: Brad Paulsen wrote: Richard Loosemore wrote: Brad Paulsen wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept or event is a product of the Danger, Will Robinson!, reptilian part of our brain. When we don't know we don't know something we react with a feeling of not knowing as a survival response. Then, having survived, we put the thing not known at the head of our list of things I don't know. As long as that thing is in this list it explains how we can come to the feeling of not knowing it so quickly. Of course, keeping a large list of things I don't know around is probably not a good idea. I suspect such a list will naturally get smaller through atrophy. You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. And... Another intuition tells me that the list of things I don't know, might generate a certain amount of cognitive dissonance the resolution of which can only be accomplished by seeking out new information (i.e., learning)? If so, does this mean that such a list in an AGI could be an important element of that AGI's desire to learn? From a functional point of view, this could be something as simple as a scheduled background task that checks the things I don't know list occasionally and, under the right circumstances, pings the AGI with a pang of cognitive dissonance from time to time. So, what say ye? Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I note that even in the simplest word-recognition neural nets that I built and studied in the 1990s, activation of a nonword proceeded in a very different way than activation of a word: it would have been easy to build something to trigger a this is a nonword neuron. Is there some type of AI formalism where nonword recognition would be problematic? Richard Loosemore Richard, You seem to have decided my request for comment was about word (mis)recognition. It wasn't. Unfortunately, I included a misleading example in my initial post. A couple of list members called me on it immediately (I'd expect nothing less from this group -- and this was a valid criticism duly noted). So far, three people have pointed out that a query containing an un-common (foreign, slang or both) word is one way to quickly generate the feeling of not knowing. But, it is just that: only one way. Not all feelings of not knowing are produced by linguistic analysis of surface features. In fact, I would guess that the vast majority of them are not so generated. Still, some are and pointing this out was a valid contribution (perhaps that example was fortunately bad). I don't think my query is a no-brainer to answer (unless you want to make it one) and your response, since it contained only another flavor of the previous two responses, gives me no reason whatsoever to change my opinion. Please take a look at the revised example in this thread. I don't think it has the same problems (as an example) as did the initial example. In particular, all of the words are common (American English) and
Re: [agi] How do we know we don't know?
Richard, Someone who can throw comments like Isn't this a bit of a no-brainer? and Keeping lists of 'things not known' is wildly, outrageously impossible, for any system! at people should expect a little bit of annoyance in return. If you can't take it, don't dish it out. Your responses to my initial post so far have been devoid of any real substantive evidence or argument for the opinions you have expressed therein. Your initial reply correctly identified an additional mechanism that two other list members had previously reported (that surface features could raise the feeling of not knowing without triggering an exhaustive memory search). As I pointed out in my response to them, this observation was a good catch but did not, in any way, show my ideas to be no-brainers or wildly, outrageously impossible. In that reply, I posted a new example query that contained only common American English words and was syntactically valid. If you want to present an evidence-based or well-reasoned argument why you believe my ideas are meritless, then let's have it. Pejorative adjectives, ad hominem attacks and baseless opinions don't impress me much. As to your cheerleader, she's just made my kill-list. The only thing worse than someone who slings unsupported opinions around like they're facts, is someone who slings someone else's unsupported opinions around like they're facts. Who is Mark Waser? Cheers, Brad Richard Loosemore wrote: Brad Paulsen wrote: Valentina, Well, the LOL is on you. Richard failed to add anything new to the two previous responses that each posited linguistic surface feature analysis as being responsible for generate the feeling of not knowing with that *particular* (and, admittedly poorly-chosen) example query. This mechanism will, however, apply to only a very tiny number of cases. In response to those first two replies (not including Richard's), I apologized for the sloppy example and offered a new one. Please read the entire thread and the new example. I think you'll find Richard's and your explanation will fail to address how the new example might generate the feeling of not knowing. Brad, Isn't this response, as well as the previous response directed at me, just a little more annoyed-sounding than it needs to be? Both Valentina and I (and now Mark Waser also) have simply focused on the fact that it is relatively trivial to build mechanisms that monitor the rate at which the system is progressing in its attempt to do a recognition operation, and then call it as a not known if the progress rate is below a certain threshold. In particular, you did suggest the idea of a system keeping lists of things it did not know, and surely it is not inappropriate to give a good-naturedly humorous response to that one? So far, I don't see any of us making a substantial misunderstanding of your question, nor anyone being deliberately rude to you. Richard Loosemore Valentina Poletti wrote: lol.. well said richard. the stimuli simply invokes no signiticant response and thus our brain concludes that we 'don't know'. that's why it takes no effort to realize it. agi algorithms should be built in a similar way, rather than searching. Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I note that even in the simplest word-recognition neural nets that I built and studied in the 1990s, activation of a nonword proceeded in a very different way than activation of a word: it would have been easy to build something to trigger a this is a nonword neuron. Is there some type of AI formalism where nonword recognition would be problematic? Richard Loosemore *agi* | Archives https://www.listbox.com/member/archive/303/=now https://www.listbox.com/member/archive/rss/303/ | Modify https://www.listbox.com/member/?; Your Subscription[Powered by Listbox] http://www.listbox.com --- agi Archives:
Re: [agi] How do we know we don't know?
People can discriminate real words from nonwords even when the latter are orthographically and phonologically word-like, presumably because words activate specific lexical and/or semantic information. http://cat.inist.fr/?aModele=afficheNcpsidt=14733408 Categories like noun and verb represent the basic units of grammar in all human languages, and the retrieval of categorical information associated with words is an essential step in the production of grammatical speech. Studies of brain-damaged patients suggest that knowledge of nouns and verbs can be spared or impaired selectively; however, the neuroanatomical correlates of this dissociation are not well understood. We used event-related functional MRI to identify cortical regions that were active when English-speaking subjects produced nouns or verbs in the context of short phrases http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1360518 Neuroimaging and lesion studies suggest that processing of word classes, such as verbs and nouns, is associated with distinct neural mechanisms. Such studies also suggest that subcategories within these broad word class categories are differentially processed in the brain. Within the class of verbs, argument structure provides one linguistic dimension that distinguishes among verb exemplars, with some requiring more complex argument structure entries than others. This study examined the neural instantiation of verbs by argument structure complexity: one-, two-, and three-argument verbs. http://portal.acm.org/citation.cfm?id=1321140.1321142coll=dl= The neural basis for verb comprehension has proven elusive, in part because of the limited range of verb categories that have been assessed. In the present study, 16 healthy young adults were probed for the meaning associated with verbs of MOTION and verbs of COGNITION. We observed distinct patterns of activation for each verb subcategory: MOTION verbs are associated with recruitment of left ventral temporal-occipital cortex, bilateral prefrontal cortex and caudate, whereas COGNITION verbs are associated with left posterolateral temporal activation. These findings are consistent with the claim that the neural representations of verb subcategories are distinct http://cat.inist.fr/?aModele=afficheNcpsidt=13451551 Neural processing of nouns and verbs: the role of inflectional morphology http://csl.psychol.cam.ac.uk/publications/04_Tyler_Neuropsychologia.pdf Others: http://cercor.oxfordjournals.org/cgi/content/abstract/12/9/900 http://www3.interscience.wiley.com/journal/99520773/abstract?CRETRY=1SRETRY=0 http://www.jneurosci.org/cgi/content/abstract/22/7/2936 - Original Message - From: Jim Bromer [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Wednesday, July 30, 2008 12:15 PM Subject: Re: [agi] How do we know we don't know? On Wed, Jul 30, 2008 at 9:50 AM, Mark Waser [EMAIL PROTECTED] wrote: Wow! The civility level on this list is really bottoming out . . . . along with any sort of scientific grounding. Experimental (imaging) evidence shows that known words will strongly activate some set of neurons when heard. Unknown words with recognizable parts/features will also activate some other set of neurons when heard, possibly allowing the individual to puzzle out the meaning even if the word has never been heard before. Totally unknown words will not strongly activate any neurons -- except subsequently (i.e. on a delay) some set of HUH? neurons. Well, your imaging evidence is part imaging and part imagining since no one knows what the imaging is actually showing. I think it is commonly believed that the imaging techniques show blood flow into areas of the brain, and this is (reasonably in my view) taken as evidence of neural activity. Ok, but what kind of thinking is actually going on and how extensive are the links that don't have enough wow factor for repeatable experiments researchers to issue as a press release. So if you are going to claim that you're speculations are superiorly grounded, I would like to see some research that shows that unknown words will not strongly activate any neurons. Take your time, I am only asking a question, not challenging you to fantasy combat. Jim Bromer --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Brad, --- On Wed, 7/30/08, Brad Paulsen [EMAIL PROTECTED] wrote: As to your cheerleader, she's just made my kill-list. The only thing worse than someone who slings unsupported opinions around like they're facts, is someone who slings someone else's unsupported opinions around like they're facts. I have to say I think you're over-reacting here a bit. Obviously you're free to do whatever, but to place someone on your kill list for supporting someone who disagrees with you seems awfully thin-skinned to me. I thought her post was a valid contribution, and not merely cheerleading, because I didn't really understand Richard's point about neural nets until I read her post. Terren --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Brad, Go back and look at Richard's e-mail again. His statement that Keeping lists of 'things not known' is wildly, outrageously impossible, for any system *WAS* supported by a brief but very clear evidence-based *and* well-reasoned argument that should have made it's truth *very* obvious to someone with sufficient background. Just because you don't understand why something is true doesn't change it from a fact to an opinion. Richard is generally very good in clearly and accurately distinguishing between what is a generally-accepted fact and what is his guestimate or opinion is his e-mails. - Original Message - From: Brad Paulsen [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Wednesday, July 30, 2008 4:14 PM Subject: Re: [agi] How do we know we don't know? Richard, Someone who can throw comments like Isn't this a bit of a no-brainer? and Keeping lists of 'things not known' is wildly, outrageously impossible, for any system! at people should expect a little bit of annoyance in return. If you can't take it, don't dish it out. Your responses to my initial post so far have been devoid of any real substantive evidence or argument for the opinions you have expressed therein. Your initial reply correctly identified an additional mechanism that two other list members had previously reported (that surface features could raise the feeling of not knowing without triggering an exhaustive memory search). As I pointed out in my response to them, this observation was a good catch but did not, in any way, show my ideas to be no-brainers or wildly, outrageously impossible. In that reply, I posted a new example query that contained only common American English words and was syntactically valid. If you want to present an evidence-based or well-reasoned argument why you believe my ideas are meritless, then let's have it. Pejorative adjectives, ad hominem attacks and baseless opinions don't impress me much. As to your cheerleader, she's just made my kill-list. The only thing worse than someone who slings unsupported opinions around like they're facts, is someone who slings someone else's unsupported opinions around like they're facts. Who is Mark Waser? Cheers, Brad Richard Loosemore wrote: Brad Paulsen wrote: Valentina, Well, the LOL is on you. Richard failed to add anything new to the two previous responses that each posited linguistic surface feature analysis as being responsible for generate the feeling of not knowing with that *particular* (and, admittedly poorly-chosen) example query. This mechanism will, however, apply to only a very tiny number of cases. In response to those first two replies (not including Richard's), I apologized for the sloppy example and offered a new one. Please read the entire thread and the new example. I think you'll find Richard's and your explanation will fail to address how the new example might generate the feeling of not knowing. Brad, Isn't this response, as well as the previous response directed at me, just a little more annoyed-sounding than it needs to be? Both Valentina and I (and now Mark Waser also) have simply focused on the fact that it is relatively trivial to build mechanisms that monitor the rate at which the system is progressing in its attempt to do a recognition operation, and then call it as a not known if the progress rate is below a certain threshold. In particular, you did suggest the idea of a system keeping lists of things it did not know, and surely it is not inappropriate to give a good-naturedly humorous response to that one? So far, I don't see any of us making a substantial misunderstanding of your question, nor anyone being deliberately rude to you. Richard Loosemore Valentina Poletti wrote: lol.. well said richard. the stimuli simply invokes no signiticant response and thus our brain concludes that we 'don't know'. that's why it takes no effort to realize it. agi algorithms should be built in a similar way, rather than searching. Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I
Re: [agi] How do we know we don't know?
Brad, I just wrote a long, point-by-point response to this, but on reflection I am not going to send it. Instead, I would like to echo Terren Suydam's comment and say that I think that you have overreacted here, because in my original reply to you I had not the slightest intention of insulting you or your ideas. The opening remark, for example, was meant to suggest that the QUESTION you posed was a no-brainer (as in, easily answerable), not that your ideas were brainless. You will note that there was a smiley in the post, and it started with a question, not a declaration (Isn't this a bit of a no-brainer?...). Throughout, I have simply been trying to explain that there is a general strategy for solving your initial question - a strategy quite well known to many people - which applies to all versions of the question, whether they be at the lexical level or the semantic level. Valentina, it seems to me, was reacting to the humorous example I gave, not mocking you personally. Certainly, if you feel that I insulted you I am quite willing to apologize for what (from my point of view) was an accident of prose style. Richard Loosemore Brad Paulsen wrote: Richard, Someone who can throw comments like Isn't this a bit of a no-brainer? and Keeping lists of 'things not known' is wildly, outrageously impossible, for any system! at people should expect a little bit of annoyance in return. If you can't take it, don't dish it out. Your responses to my initial post so far have been devoid of any real substantive evidence or argument for the opinions you have expressed therein. Your initial reply correctly identified an additional mechanism that two other list members had previously reported (that surface features could raise the feeling of not knowing without triggering an exhaustive memory search). As I pointed out in my response to them, this observation was a good catch but did not, in any way, show my ideas to be no-brainers or wildly, outrageously impossible. In that reply, I posted a new example query that contained only common American English words and was syntactically valid. If you want to present an evidence-based or well-reasoned argument why you believe my ideas are meritless, then let's have it. Pejorative adjectives, ad hominem attacks and baseless opinions don't impress me much. As to your cheerleader, she's just made my kill-list. The only thing worse than someone who slings unsupported opinions around like they're facts, is someone who slings someone else's unsupported opinions around like they're facts. Who is Mark Waser? Cheers, Brad Richard Loosemore wrote: Brad Paulsen wrote: Valentina, Well, the LOL is on you. Richard failed to add anything new to the two previous responses that each posited linguistic surface feature analysis as being responsible for generate the feeling of not knowing with that *particular* (and, admittedly poorly-chosen) example query. This mechanism will, however, apply to only a very tiny number of cases. In response to those first two replies (not including Richard's), I apologized for the sloppy example and offered a new one. Please read the entire thread and the new example. I think you'll find Richard's and your explanation will fail to address how the new example might generate the feeling of not knowing. Brad, Isn't this response, as well as the previous response directed at me, just a little more annoyed-sounding than it needs to be? Both Valentina and I (and now Mark Waser also) have simply focused on the fact that it is relatively trivial to build mechanisms that monitor the rate at which the system is progressing in its attempt to do a recognition operation, and then call it as a not known if the progress rate is below a certain threshold. In particular, you did suggest the idea of a system keeping lists of things it did not know, and surely it is not inappropriate to give a good-naturedly humorous response to that one? So far, I don't see any of us making a substantial misunderstanding of your question, nor anyone being deliberately rude to you. Richard Loosemore Valentina Poletti wrote: lol.. well said richard. the stimuli simply invokes no signiticant response and thus our brain concludes that we 'don't know'. that's why it takes no effort to realize it. agi algorithms should be built in a similar way, rather than searching. Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously
Re: [agi] How do we know we don't know?
Brad, Ah,, perhaps there has been a failure of communication - it sounded (rightly or wrongly) from your original post, like your things I don't know list was being used DURING the process of perception/ categorization, and so was key to producing the I don't know this feeling. That was hard to understand and accept. If you're just saying that AFTER our brain has failed to recognize something, it effectively (as you discuss) stores those failures on an I don't know list, that is unobjectionable. James, Someone ventured the *opinion* that keeping such a list of things I don't know was nonsensical, but I have yet to see any evidence or well-reasoned argument backing that opinion. So, it's just an opinion. One with which I, obviously, do not agree. There are two stages of not knowing. The first is when the agent doesn't know it doesn't know something. It's clueless. This can be such a dangerous stage to be in that one can imagine the agent might be equipped with a knee-jerk type reaction, which evinces itself in a variety of ways. One of those ways could be to promote this thing it didn't know it didn't know to the next stage of not knowing by storing it (subconsciously, most likely) in a list of things I know I don't know. I use the term list generically. I don't argue that the human brain maintains knowledge in list structures or that this would necessarily be the way this information is stored in an AGI agent). I fail to see how saving this type of information in memory is any different from saving any other type of information. It's a positive fact about the world as that world relates to the individual human (or AGI agent). The first way having such a list might help is in optimizing memory search. The next time the agent encounters a thing not known on this list, it won't have to perform an exhaustive search of things it knows to come to the feeling of not knowing. It's right there on the (comparatively short) list of things it doesn't know (which would be searched first, of course). In addition, if the agent's experience in the world results in repeated hits on a particular item in this list, this could be a factor in producing the desire to learn that is such a characteristic behavior of our species. Once the thing is known, it is, of course, removed from the not known list. If a thing on the list is not encountered again for a long period of time, it might just fall off the list. Both of these characteristics of such a list would work, subconsciously, to keep the list both small and relevant. Cheers, Brad James Ratcliff wrote: Sure, search is at the root of all processing, be it human or AI. How we each go about the search, and how efficient we are at the task are different, and what exactly we are searching for, and exponential explosion. But some type of search is done, whether we are consciously aware of our brains doing the search or not. Given a bit of context information about the question should allow us to use some heuristics to look at a smaller area of knowledge bases in our brains, or in a computer's memory. Having a list of things we dont know is nonsensical as has been pointed out, when it comes to individual terms, but something like a aggregate estimate of knowledge known could be computed. I myself know a little about baseball say 10%, but baseball history and world series statistics would be more like 0.1% James Ratcliff --- On *Tue, 7/29/08, Brad Paulsen /[EMAIL PROTECTED]/* wrote: James, So, you agree that some sort of search must take place before the feeling of not knowing presents itself? Of course, realizing we don't have a lot of information results from some type of a search and not a separate process (at least you didn't posit any). Thanks for your comments! Cheers Brad James Ratcliff wrote: It is fairly simple at that point, we have enough context to have a very limited domain world series - baseball 1924 answer is a team, so we can do a lookup in our database easily enough, or realize that we really dont have a lot of information about baseball in our mindset. And for the other one, it would just be a strait term match. James Ratcliff ___ James Ratcliff - http://falazar.com Looking for something... --- On *Mon, 7/28/08, Brad Paulsen /[EMAIL PROTECTED]/* wrote: From: Brad Paulsen [EMAIL PROTECTED] Subject: Re: [agi] How do we know we don't know? To: agi@v2.listbox.com Date: Monday, July 28, 2008, 4:12 PM Jim Bromer wrote: On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're
Re: [agi] How do we know we don't know?
Yes ok, this is needed, but was a bit different than what was being discussed earlier, thank you for the clarification. ___ James Ratcliff - http://falazar.com Looking for something... --- On Wed, 7/30/08, Brad Paulsen [EMAIL PROTECTED] wrote: From: Brad Paulsen [EMAIL PROTECTED] Subject: Re: [agi] How do we know we don't know? To: agi@v2.listbox.com Date: Wednesday, July 30, 2008, 3:47 PM James, Someone ventured the *opinion* that keeping such a list of things I don't know was nonsensical, but I have yet to see any evidence or well-reasoned argument backing that opinion. So, it's just an opinion. One with which I, obviously, do not agree. There are two stages of not knowing. The first is when the agent doesn't know it doesn't know something. It's clueless. This can be such a dangerous stage to be in that one can imagine the agent might be equipped with a knee-jerk type reaction, which evinces itself in a variety of ways. One of those ways could be to promote this thing it didn't know it didn't know to the next stage of not knowing by storing it (subconsciously, most likely) in a list of things I know I don't know. I use the term list generically. I don't argue that the human brain maintains knowledge in list structures or that this would necessarily be the way this information is stored in an AGI agent). I fail to see how saving this type of information in memory is any different from saving any other type of information. It's a positive fact about the world as that world relates to the individual human (or AGI agent). The first way having such a list might help is in optimizing memory search. The next time the agent encounters a thing not known on this list, it won't have to perform an exhaustive search of things it knows to come to the feeling of not knowing. It's right there on the (comparatively short) list of things it doesn't know (which would be searched first, of course). In addition, if the agent's experience in the world results in repeated hits on a particular item in this list, this could be a factor in producing the desire to learn that is such a characteristic behavior of our species. Once the thing is known, it is, of course, removed from the not known list. If a thing on the list is not encountered again for a long period of time, it might just fall off the list. Both of these characteristics of such a list would work, subconsciously, to keep the list both small and relevant. Cheers, Brad James Ratcliff wrote: Sure, search is at the root of all processing, be it human or AI. How we each go about the search, and how efficient we are at the task are different, and what exactly we are searching for, and exponential explosion. But some type of search is done, whether we are consciously aware of our brains doing the search or not. Given a bit of context information about the question should allow us to use some heuristics to look at a smaller area of knowledge bases in our brains, or in a computer's memory. Having a list of things we dont know is nonsensical as has been pointed out, when it comes to individual terms, but something like a aggregate estimate of knowledge known could be computed. I myself know a little about baseball say 10%, but baseball history and world series statistics would be more like 0.1% James Ratcliff --- On *Tue, 7/29/08, Brad Paulsen /[EMAIL PROTECTED]/* wrote: James, So, you agree that some sort of search must take place before the feeling of not knowing presents itself? Of course, realizing we don't have a lot of information results from some type of a search and not a separate process (at least you didn't posit any). Thanks for your comments! Cheers Brad James Ratcliff wrote: It is fairly simple at that point, we have enough context to have a very limited domain world series - baseball 1924 answer is a team, so we can do a lookup in our database easily enough, or realize that we really dont have a lot of information about baseball in our mindset. And for the other one, it would just be a strait term match. James Ratcliff ___ James Ratcliff - http://falazar.com Looking for something... --- On *Mon, 7/28/08, Brad Paulsen /[EMAIL PROTECTED]/* wrote: From: Brad Paulsen [EMAIL PROTECTED] Subject: Re: [agi] How do we know we don't know? To: agi@v2.listbox.com Date: Monday, July 28, 2008, 4:12 PM Jim Bromer wrote: On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're
Re: [agi] How do we know we don't know?
Richard, I just finished reading and replying to your post preceding this one (I guess). Your tone and approach in that post was more like what I expected from you. I'm not going to get in a pissing match about what I should or should not take personally. That will generate only heat not light. Peace, OK? Cheers, Brad P.S. I will review Valentina's post to see if I misunderstood it. When I originally read it, it sure looked like piling on to me. P.S. Terren: I reserve the right to put anyone in my personal kill-list. I don't have to justify my reasons. If I choose to not read the posts of a particular list member, and that person turns up on Time Magazine's cover ten years from now, well... my loss. Right? Richard Loosemore wrote: Brad, I just wrote a long, point-by-point response to this, but on reflection I am not going to send it. Instead, I would like to echo Terren Suydam's comment and say that I think that you have overreacted here, because in my original reply to you I had not the slightest intention of insulting you or your ideas. The opening remark, for example, was meant to suggest that the QUESTION you posed was a no-brainer (as in, easily answerable), not that your ideas were brainless. You will note that there was a smiley in the post, and it started with a question, not a declaration (Isn't this a bit of a no-brainer?...). Throughout, I have simply been trying to explain that there is a general strategy for solving your initial question - a strategy quite well known to many people - which applies to all versions of the question, whether they be at the lexical level or the semantic level. Valentina, it seems to me, was reacting to the humorous example I gave, not mocking you personally. Certainly, if you feel that I insulted you I am quite willing to apologize for what (from my point of view) was an accident of prose style. Richard Loosemore Brad Paulsen wrote: Richard, Someone who can throw comments like Isn't this a bit of a no-brainer? and Keeping lists of 'things not known' is wildly, outrageously impossible, for any system! at people should expect a little bit of annoyance in return. If you can't take it, don't dish it out. Your responses to my initial post so far have been devoid of any real substantive evidence or argument for the opinions you have expressed therein. Your initial reply correctly identified an additional mechanism that two other list members had previously reported (that surface features could raise the feeling of not knowing without triggering an exhaustive memory search). As I pointed out in my response to them, this observation was a good catch but did not, in any way, show my ideas to be no-brainers or wildly, outrageously impossible. In that reply, I posted a new example query that contained only common American English words and was syntactically valid. If you want to present an evidence-based or well-reasoned argument why you believe my ideas are meritless, then let's have it. Pejorative adjectives, ad hominem attacks and baseless opinions don't impress me much. As to your cheerleader, she's just made my kill-list. The only thing worse than someone who slings unsupported opinions around like they're facts, is someone who slings someone else's unsupported opinions around like they're facts. Who is Mark Waser? Cheers, Brad Richard Loosemore wrote: Brad Paulsen wrote: Valentina, Well, the LOL is on you. Richard failed to add anything new to the two previous responses that each posited linguistic surface feature analysis as being responsible for generate the feeling of not knowing with that *particular* (and, admittedly poorly-chosen) example query. This mechanism will, however, apply to only a very tiny number of cases. In response to those first two replies (not including Richard's), I apologized for the sloppy example and offered a new one. Please read the entire thread and the new example. I think you'll find Richard's and your explanation will fail to address how the new example might generate the feeling of not knowing. Brad, Isn't this response, as well as the previous response directed at me, just a little more annoyed-sounding than it needs to be? Both Valentina and I (and now Mark Waser also) have simply focused on the fact that it is relatively trivial to build mechanisms that monitor the rate at which the system is progressing in its attempt to do a recognition operation, and then call it as a not known if the progress rate is below a certain threshold. In particular, you did suggest the idea of a system keeping lists of things it did not know, and surely it is not inappropriate to give a good-naturedly humorous response to that one? So far, I don't see any of us making a substantial misunderstanding of your question, nor anyone being deliberately rude to you. Richard Loosemore Valentina Poletti wrote:
Re: [agi] How do we know we don't know?
Richard Loosemore wrote: Brad Paulsen wrote: James, Someone ventured the *opinion* that keeping such a list of things I don't know was nonsensical, but I have yet to see any evidence or well-reasoned argument backing that opinion. So, it's just an opinion. One with which I, obviously, do not agree. Please be clear about what was intended by my remarks. I *now* have an explicit, episodic memory of confronting the question Who won the world series in 1954, and as a result of that episode that occured today, I have the explicit knowledge that I do not know the answer. Having that kind of explicit knowledge of lack-of-knowledge is not problematic at all. The only thing that seems implausible is that IN GENERAL we try to answer questions by first looking up explicit elements that encode the fact that we do not know the answer. As a general strategy this must, surely, be deeply implausible, for the reasons that I originally gave, which centered on the fact that the sheer quantity of unknowns would be overwhelming for any system. For almost every one of the potentially askable questions that would elicit, in me, a response of I do not know, there would not be any such episode. Similarly, it would be clearly implausible for the cognitive system to spend its time making lists of things that it did not know. If that is not an example of an obviously implausible mechanism, then I do not know what would be. Ah. Now we're getting somewhere! I do *not* (and did not) propose that we keep a list of all the things unknown in memory. Nor did I propose some background task that would maintain or add to such a list. That would be ...wildly, outrageously impossible, for any system! Maybe, instead of assuming the worse (that I could be so ignorant as to propose such a list), you might have asked for some clarification? The list of things I don't know is, by definition, a list of things I know I don't know. How could I *possibly* know about things I don't know I don't know? The list I propose contains ONLY those things we know we don't know. Such a list is, in my opinion, completely manageable and, indeed, helpful information to have around. When we first encounter a completely novel object or event we will have to search (percolate, whatever) for it in memory and come up empty (however you want to define that). It is then, and *only* then, that we put this knowledge (or meta-knowledge) on the things (I know) I don't know list. This list can be consulted before performing a search of all memory to determine if there's a need to do such an exhaustive search. If the thing we're trying to remember is on the things (I know) I don't know list, we can very quickly report the feeling of not knowing. Otherwise, we have to do the exhaustive (however you define that) search of things we do know and come up empty. Such a list can also be used by subconscious processes to power our desire to learn. Presumably, we experience cognitive dissonance when we feel there's something we know nothing about and want to resolve that feeling. How? By learning. Once learned, the thing falls off the things (I know) I don't know list. Similarly, if an item is on the list for a long time, it will naturally fall off the list (the use it or lose it principle). Both of these natural actions will work, I believe, to keep this list quite small. Sometimes (well, don't ask my ex) I can be a bit thick. I know you're all surprised to hear that, but... It just dawned on me that much of the uproar here may have been caused by a miscommunication (gee, where have we heard of that happening before?). I may have used the term things we don't know to denote the things we know we don't know list. If so, please accept my apologies. Having played with these questions for a long time, this *important* distinction apparently became lost to me and I began to assume it self-evident that a things we don't know list would have had to come into being as the result of our encounters with those things when they were things we didn't know we didn't know (and, therefore, could not be in any list of knowledge we had -- we are clueless about these things until we encounter them). If that's the case, let me (finally) be clear: the list I am talking about in the human or AGI agent's memory is a list of THINGS I KNOW I DON'T KNOW. In the first (misleading) example I gave, the word fomlepung would be on that list after the query containing it had resulted in the I don't know answer (how that determination is made is really a minor point for this discussion). In the second example query I gave, the Which team won the 1924 World Series? would also, after eliciting the I don't know response, find its way onto this list. This was not merely an opinion, it was a reasoned argument, illustrated by an example of a nonword that clearly belonged to a vast class of nonwords. Well, be
Re: [agi] How do we know we don't know?
Brad Paulsen wrote: Richard Loosemore wrote: Brad Paulsen wrote: Richard Loosemore wrote: Brad Paulsen wrote: Richard Loosemore wrote: Brad Paulsen wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept or event is a product of the Danger, Will Robinson!, reptilian part of our brain. When we don't know we don't know something we react with a feeling of not knowing as a survival response. Then, having survived, we put the thing not known at the head of our list of things I don't know. As long as that thing is in this list it explains how we can come to the feeling of not knowing it so quickly. Of course, keeping a large list of things I don't know around is probably not a good idea. I suspect such a list will naturally get smaller through atrophy. You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. And... Another intuition tells me that the list of things I don't know, might generate a certain amount of cognitive dissonance the resolution of which can only be accomplished by seeking out new information (i.e., learning)? If so, does this mean that such a list in an AGI could be an important element of that AGI's desire to learn? From a functional point of view, this could be something as simple as a scheduled background task that checks the things I don't know list occasionally and, under the right circumstances, pings the AGI with a pang of cognitive dissonance from time to time. So, what say ye? Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I note that even in the simplest word-recognition neural nets that I built and studied in the 1990s, activation of a nonword proceeded in a very different way than activation of a word: it would have been easy to build something to trigger a this is a nonword neuron. Is there some type of AI formalism where nonword recognition would be problematic? Richard Loosemore Richard, You seem to have decided my request for comment was about word (mis)recognition. It wasn't. Unfortunately, I included a misleading example in my initial post. A couple of list members called me on it immediately (I'd expect nothing less from this group -- and this was a valid criticism duly noted). So far, three people have pointed out that a query containing an un-common (foreign, slang or both) word is one way to quickly generate the feeling of not knowing. But, it is just that: only one way. Not all feelings of not knowing are produced by linguistic analysis of surface features. In fact, I would guess that the vast majority of them are not so generated. Still, some are and pointing this out was a valid contribution (perhaps that example was fortunately bad). I don't think my query is a no-brainer to answer (unless you want to make it one) and your response, since it contained only another flavor of the previous two responses, gives me no reason whatsoever to change my opinion. Please take a look at the revised example in this thread. I don't think it has the same problems (as an example) as did the initial example. In particular,
Re: [agi] How do we know we don't know?
Brad Paulsen wrote: Richard Loosemore wrote: Brad Paulsen wrote: James, Someone ventured the *opinion* that keeping such a list of things I don't know was nonsensical, but I have yet to see any evidence or well-reasoned argument backing that opinion. So, it's just an opinion. One with which I, obviously, do not agree. Please be clear about what was intended by my remarks. I *now* have an explicit, episodic memory of confronting the question Who won the world series in 1954, and as a result of that episode that occured today, I have the explicit knowledge that I do not know the answer. Having that kind of explicit knowledge of lack-of-knowledge is not problematic at all. The only thing that seems implausible is that IN GENERAL we try to answer questions by first looking up explicit elements that encode the fact that we do not know the answer. As a general strategy this must, surely, be deeply implausible, for the reasons that I originally gave, which centered on the fact that the sheer quantity of unknowns would be overwhelming for any system. For almost every one of the potentially askable questions that would elicit, in me, a response of I do not know, there would not be any such episode. Similarly, it would be clearly implausible for the cognitive system to spend its time making lists of things that it did not know. If that is not an example of an obviously implausible mechanism, then I do not know what would be. Ah. Now we're getting somewhere! I do *not* (and did not) propose that we keep a list of all the things unknown in memory. Nor did I propose some background task that would maintain or add to such a list. That would be ...wildly, outrageously impossible, for any system! Maybe, instead of assuming the worse (that I could be so ignorant as to propose such a list), you might have asked for some clarification? The list of things I don't know is, by definition, a list of things I know I don't know. How could I *possibly* know about things I don't know I don't know? The list I propose contains ONLY those things we know we don't know. Such a list is, in my opinion, completely manageable and, indeed, helpful information to have around. When we first encounter a completely novel object or event we will have to search (percolate, whatever) for it in memory and come up empty (however you want to define that). It is then, and *only* then, that we put this knowledge (or meta-knowledge) on the things (I know) I don't know list. This list can be consulted before performing a search of all memory to determine if there's a need to do such an exhaustive search. If the thing we're trying to remember is on the things (I know) I don't know list, we can very quickly report the feeling of not knowing. Otherwise, we have to do the exhaustive (however you define that) search of things we do know and come up empty. Such a list can also be used by subconscious processes to power our desire to learn. Presumably, we experience cognitive dissonance when we feel there's something we know nothing about and want to resolve that feeling. How? By learning. Once learned, the thing falls off the things (I know) I don't know list. Similarly, if an item is on the list for a long time, it will naturally fall off the list (the use it or lose it principle). Both of these natural actions will work, I believe, to keep this list quite small. These are all interesting questions, in a way, but they involve a way of doing AI that I find ... problematic ... for other reasons. I would have many questions about whether the maintenance and deployment of such a list would actually be as viable as you imply, but that is very much a practical question specific to that type of AI. The more general issue of whether the system keeps meta knowledge of that sort is something that we completely agree on: whichever way it uses it, it certainly does keep it for at least a while. Sometimes (well, don't ask my ex) I can be a bit thick. I know you're all surprised to hear that, but... It just dawned on me that much of the uproar here may have been caused by a miscommunication (gee, where have we heard of that happening before?). I may have used the term things we don't know to denote the things we know we don't know list. If so, please accept my apologies. Having played with these questions for a long time, this *important* distinction apparently became lost to me and I began to assume it self-evident that a things we don't know list would have had to come into being as the result of our encounters with those things when they were things we didn't know we didn't know (and, therefore, could not be in any list of knowledge we had -- we are clueless about these things until we encounter them). If that's the case, let me (finally) be clear: the list I am talking about in the human or AGI agent's memory is a list of THINGS I KNOW I DON'T KNOW.
Re: [agi] How do we know we don't know?
lol.. well said richard. the stimuli simply invokes no signiticant response and thus our brain concludes that we 'don't know'. that's why it takes no effort to realize it. agi algorithms should be built in a similar way, rather than searching. Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I note that even in the simplest word-recognition neural nets that I built and studied in the 1990s, activation of a nonword proceeded in a very different way than activation of a word: it would have been easy to build something to trigger a this is a nonword neuron. Is there some type of AI formalism where nonword recognition would be problematic? Richard Loosemore --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
RE: [agi] How do we know we don't know?
I believe the human brain, in addition to including the controller for a physical robot, includes the controller of a thought robot which includes pushing much of the brain through learned or instinctual mental behaviors. My understanding is that much of higher level function of this thought controller is largely in the prefrontal cortex, basil ganglia, thalamic loop. I am guessing that answering a query such as What does word_X (in this case fomlepung) mean? is a type of learned behavior. The thought robot is consciously aware of the query task and the idea that as a query, its task is to search for recollection of the word fomlepung and its associations. I think the search is generated by a consciously broadcasting a pattern looking for a match for fomlepung to the appropriate areas of the brain. Although much of the spreading activation done in response to this conscious activation is, itself, in the subconscious, the thought robot task of answering a query is focusing attention on the query and any feedback from it indicating a possible answer. This could be done by looking for feedback from cortical activations to the thalamus that are in synchrony with the query pattern, tuning into them, and testing them so see if any of them are a desired match. When the conscious task of query answering does not get feedback indicating an answer, the conscious pre-frontal process engaged in query is aware of that lack of desired feedback and, thus, the human in whose mind the process is taking place is conscious that he/she doesn't know (or at least can recall) the meaning of the word. Conscious feelings of not knowing can arise in other contexts besides answering a what does word_X mean query. In some of them, subconscious processes might, for various reasons, promote a failure to match a subconscious query or task up to the consciousness. For example, a sub-subconscious pattern completion process, in say high level perception or in cognition, might draws activation to itself, pushing its activation into semi-conscious or conscious attention, both because its activation pattern is beginning to better match an emotionally weighted patterns that direct more activation energy back to it, and because there is a missing a piece of information necessary for that valuable match to be made. The brain may have learned by evolution or individual experience that such information would be more likely found if the much greater spreading activation resources of semi-conscious or conscious attention could be utilized for conducting the search for such missing information. This causes a greater search to be made for such information, and if the information is not found quickly, could cause even more attention to be allocated to the search, pushing the search and its failure into clear conscious awareness. Ed Porter -Original Message- From: Abram Demski [mailto:[EMAIL PROTECTED] Sent: Monday, July 28, 2008 4:25 PM To: agi@v2.listbox.com Subject: Re: [agi] How do we know we don't know? It seems like you have some valid points, but I cannot help but point out a problem with your question. It seems like any system for pattern recognition and/or prediction will have a sensible I Don't Know state. An algorithm in a published paper might suppress this in an attempt to give as reasonable an output as is possible in all situations, but it seems like in most such cases it would be easy to add. Therefore, where is the problem? Yet, I follow your comments and to an extent agree... the feeling when I don't know something could possibly be related to animal fear (though I am not sure), and the second time I encounter the same thing is certainly different (because I remember the previous not-knowing, so I at least have that info for context this time). But I think the issue might nonetheless be non-fundamental, because algorithms typically can easily report their not knowing. --Abram On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept
Re: [agi] How do we know we don't know?
On Tuesday 29 July 2008 03:08:55 am Valentina Poletti wrote: lol.. well said richard. the stimuli simply invokes no signiticant response and thus our brain concludes that we 'don't know'. that's why it takes no effort to realize it. agi algorithms should be built in a similar way, rather than searching. Unhhh that *IS* a kind of search. It's a shallowly truncated breadth-first search, but it's a search. Compare that with the the words right on the tip of my tongue phenomenon. In that case you get sufficient response that you become aware of a search going on, and you even know that the result *should* be positive. You just can't find it. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
This is not a hard problem. A model for data compression has the task of predicting the next bit in a string of unknown origin. If the string is an encoding of natural language text, then modeling is an AI problem. If the model doesn't know, then it assigns a probability of about 1/2 to each of 0 and 1. Probabilities can be easily detected from outside the model, regardless of the intelligence level of the model. -- Matt Mahoney, [EMAIL PROTECTED] --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Richard Loosemore wrote: Brad Paulsen wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept or event is a product of the Danger, Will Robinson!, reptilian part of our brain. When we don't know we don't know something we react with a feeling of not knowing as a survival response. Then, having survived, we put the thing not known at the head of our list of things I don't know. As long as that thing is in this list it explains how we can come to the feeling of not knowing it so quickly. Of course, keeping a large list of things I don't know around is probably not a good idea. I suspect such a list will naturally get smaller through atrophy. You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. And... Another intuition tells me that the list of things I don't know, might generate a certain amount of cognitive dissonance the resolution of which can only be accomplished by seeking out new information (i.e., learning)? If so, does this mean that such a list in an AGI could be an important element of that AGI's desire to learn? From a functional point of view, this could be something as simple as a scheduled background task that checks the things I don't know list occasionally and, under the right circumstances, pings the AGI with a pang of cognitive dissonance from time to time. So, what say ye? Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I note that even in the simplest word-recognition neural nets that I built and studied in the 1990s, activation of a nonword proceeded in a very different way than activation of a word: it would have been easy to build something to trigger a this is a nonword neuron. Is there some type of AI formalism where nonword recognition would be problematic? Richard Loosemore Richard, You seem to have decided my request for comment was about word (mis)recognition. It wasn't. Unfortunately, I included a misleading example in my initial post. A couple of list members called me on it immediately (I'd expect nothing less from this group -- and this was a valid criticism duly noted). So far, three people have pointed out that a query containing an un-common (foreign, slang or both) word is one way to quickly generate the feeling of not knowing. But, it is just that: only one way. Not all feelings of not knowing are produced by linguistic analysis of surface features. In fact, I would guess that the vast majority of them are not so generated. Still, some are and pointing this out was a valid contribution (perhaps that example was fortunately bad). I don't think my query is a no-brainer to answer (unless you want to make it one) and your response, since it contained only another flavor of the previous two responses, gives me no reason whatsoever to change my opinion. Please take a look at the revised example in this thread. I don't think it has the same problems (as an example) as did the initial example. In particular, all of the words are common (American English) and the syntax is valid. Cheers, Brad
Re: [agi] How do we know we don't know?
James, So, you agree that some sort of search must take place before the feeling of not knowing presents itself? Of course, realizing we don't have a lot of information results from some type of a search and not a separate process (at least you didn't posit any). Thanks for your comments! Cheers Brad James Ratcliff wrote: It is fairly simple at that point, we have enough context to have a very limited domain world series - baseball 1924 answer is a team, so we can do a lookup in our database easily enough, or realize that we really dont have a lot of information about baseball in our mindset. And for the other one, it would just be a strait term match. James Ratcliff ___ James Ratcliff - http://falazar.com Looking for something... --- On *Mon, 7/28/08, Brad Paulsen /[EMAIL PROTECTED]/* wrote: From: Brad Paulsen [EMAIL PROTECTED] Subject: Re: [agi] How do we know we don't know? To: agi@v2.listbox.com Date: Monday, July 28, 2008, 4:12 PM Jim Bromer wrote: On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. Brad My guess that initial recognition must be based on the surface features of an input. If this is true, then that could suggest that our initial recognition reactions are stimulated by distinct components (or distinct groupings of components) that are found in the surface input data. Jim Bromer Hmmm. That particular query may not have been the best example since, to a non-Norwegian speaker, the phonological surface feature of that statement alone could account for the feeling of not knowing. In other words, the word fomlepung just doesn't sound right. Good point. But, that may only explain how we know we don't know strange sounding words. Let's try another example: Which team won the 1924 World Series? Cheers, Brad --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com *agi* | Archives https://www.listbox.com/member/archive/303/=now https://www.listbox.com/member/archive/rss/303/ | Modify https://www.listbox.com/member/?; Your Subscription [Powered by Listbox] http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Valentina, Well, the LOL is on you. Richard failed to add anything new to the two previous responses that each posited linguistic surface feature analysis as being responsible for generate the feeling of not knowing with that *particular* (and, admittedly poorly-chosen) example query. This mechanism will, however, apply to only a very tiny number of cases. In response to those first two replies (not including Richard's), I apologized for the sloppy example and offered a new one. Please read the entire thread and the new example. I think you'll find Richard's and your explanation will fail to address how the new example might generate the feeling of not knowing. Cheers, Brad Valentina Poletti wrote: lol.. well said richard. the stimuli simply invokes no signiticant response and thus our brain concludes that we 'don't know'. that's why it takes no effort to realize it. agi algorithms should be built in a similar way, rather than searching. Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I note that even in the simplest word-recognition neural nets that I built and studied in the 1990s, activation of a nonword proceeded in a very different way than activation of a word: it would have been easy to build something to trigger a this is a nonword neuron. Is there some type of AI formalism where nonword recognition would be problematic? Richard Loosemore *agi* | Archives https://www.listbox.com/member/archive/303/=now https://www.listbox.com/member/archive/rss/303/ | Modify https://www.listbox.com/member/?; Your Subscription [Powered by Listbox] http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Ed, Thanks for the response. I'm going to read it a couple more times to make sure I didn't miss anything. But, on first read, looks good! Thanks for taking the time to comment in such detail! Cheers, Brad Ed Porter wrote: I believe the human brain, in addition to including the controller for a physical robot, includes the controller of a thought robot which includes pushing much of the brain through learned or instinctual mental behaviors. My understanding is that much of higher level function of this thought controller is largely in the prefrontal cortex, basil ganglia, thalamic loop. I am guessing that answering a query such as What does word_X (in this case fomlepung) mean? is a type of learned behavior. The thought robot is consciously aware of the query task and the idea that as a query, its task is to search for recollection of the word fomlepung and its associations. I think the search is generated by a consciously broadcasting a pattern looking for a match for fomlepung to the appropriate areas of the brain. Although much of the spreading activation done in response to this conscious activation is, itself, in the subconscious, the thought robot task of answering a query is focusing attention on the query and any feedback from it indicating a possible answer. This could be done by looking for feedback from cortical activations to the thalamus that are in synchrony with the query pattern, tuning into them, and testing them so see if any of them are a desired match. When the conscious task of query answering does not get feedback indicating an answer, the conscious pre-frontal process engaged in query is aware of that lack of desired feedback and, thus, the human in whose mind the process is taking place is conscious that he/she doesn't know (or at least can recall) the meaning of the word. Conscious feelings of not knowing can arise in other contexts besides answering a what does word_X mean query. In some of them, subconscious processes might, for various reasons, promote a failure to match a subconscious query or task up to the consciousness. For example, a sub-subconscious pattern completion process, in say high level perception or in cognition, might draws activation to itself, pushing its activation into semi-conscious or conscious attention, both because its activation pattern is beginning to better match an emotionally weighted patterns that direct more activation energy back to it, and because there is a missing a piece of information necessary for that valuable match to be made. The brain may have learned by evolution or individual experience that such information would be more likely found if the much greater spreading activation resources of semi-conscious or conscious attention could be utilized for conducting the search for such missing information. This causes a greater search to be made for such information, and if the information is not found quickly, could cause even more attention to be allocated to the search, pushing the search and its failure into clear conscious awareness. Ed Porter -Original Message- From: Abram Demski [mailto:[EMAIL PROTECTED] Sent: Monday, July 28, 2008 4:25 PM To: agi@v2.listbox.com Subject: Re: [agi] How do we know we don't know? It seems like you have some valid points, but I cannot help but point out a problem with your question. It seems like any system for pattern recognition and/or prediction will have a sensible I Don't Know state. An algorithm in a published paper might suppress this in an attempt to give as reasonable an output as is possible in all situations, but it seems like in most such cases it would be easy to add. Therefore, where is the problem? Yet, I follow your comments and to an extent agree... the feeling when I don't know something could possibly be related to animal fear (though I am not sure), and the second time I encounter the same thing is certainly different (because I remember the previous not-knowing, so I at least have that info for context this time). But I think the issue might nonetheless be non-fundamental, because algorithms typically can easily report their not knowing. --Abram On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly
Re: [agi] How do we know we don't know?
On Tuesday 29 July 2008 04:12:27 pm Brad Paulsen wrote: Richard Loosemore wrote: Brad Paulsen wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept or event is a product of the Danger, Will Robinson!, reptilian part of our brain. When we don't know we don't know something we react with a feeling of not knowing as a survival response. Then, having survived, we put the thing not known at the head of our list of things I don't know. As long as that thing is in this list it explains how we can come to the feeling of not knowing it so quickly. Of course, keeping a large list of things I don't know around is probably not a good idea. I suspect such a list will naturally get smaller through atrophy. You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. And... Another intuition tells me that the list of things I don't know, might generate a certain amount of cognitive dissonance the resolution of which can only be accomplished by seeking out new information (i.e., learning)? If so, does this mean that such a list in an AGI could be an important element of that AGI's desire to learn? From a functional point of view, this could be something as simple as a scheduled background task that checks the things I don't know list occasionally and, under the right circumstances, pings the AGI with a pang of cognitive dissonance from time to time. So, what say ye? Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I note that even in the simplest word-recognition neural nets that I built and studied in the 1990s, activation of a nonword proceeded in a very different way than activation of a word: it would have been easy to build something to trigger a this is a nonword neuron. Is there some type of AI formalism where nonword recognition would be problematic? Richard Loosemore Richard, You seem to have decided my request for comment was about word (mis)recognition. It wasn't. Unfortunately, I included a misleading example in my initial post. A couple of list members called me on it immediately (I'd expect nothing less from this group -- and this was a valid criticism duly noted). So far, three people have pointed out that a query containing an un-common (foreign, slang or both) word is one way to quickly generate the feeling of not knowing. But, it is just that: only one way. Not all feelings of not knowing are produced by linguistic analysis of surface features. In fact, I would guess that the vast majority of them are not so generated. Still, some are and pointing this out was a valid contribution (perhaps that example was fortunately bad). I don't think my query is a no-brainer to answer (unless you want to make it one) and your response, since it contained only another flavor of the previous two responses, gives me no reason whatsoever to change my opinion. Please take a look at the revised example in this thread. I don't think it has the same problems (as an example) as did the initial example. In
[agi] How do we know we don't know?
All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept or event is a product of the Danger, Will Robinson!, reptilian part of our brain. When we don't know we don't know something we react with a feeling of not knowing as a survival response. Then, having survived, we put the thing not known at the head of our list of things I don't know. As long as that thing is in this list it explains how we can come to the feeling of not knowing it so quickly. Of course, keeping a large list of things I don't know around is probably not a good idea. I suspect such a list will naturally get smaller through atrophy. You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. And... Another intuition tells me that the list of things I don't know, might generate a certain amount of cognitive dissonance the resolution of which can only be accomplished by seeking out new information (i.e., learning)? If so, does this mean that such a list in an AGI could be an important element of that AGI's desire to learn? From a functional point of view, this could be something as simple as a scheduled background task that checks the things I don't know list occasionally and, under the right circumstances, pings the AGI with a pang of cognitive dissonance from time to time. So, what say ye? Cheers, Brad --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. Brad My guess that initial recognition must be based on the surface features of an input. If this is true, then that could suggest that our initial recognition reactions are stimulated by distinct components (or distinct groupings of components) that are found in the surface input data. Jim Bromer --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
It seems like you have some valid points, but I cannot help but point out a problem with your question. It seems like any system for pattern recognition and/or prediction will have a sensible I Don't Know state. An algorithm in a published paper might suppress this in an attempt to give as reasonable an output as is possible in all situations, but it seems like in most such cases it would be easy to add. Therefore, where is the problem? Yet, I follow your comments and to an extent agree... the feeling when I don't know something could possibly be related to animal fear (though I am not sure), and the second time I encounter the same thing is certainly different (because I remember the previous not-knowing, so I at least have that info for context this time). But I think the issue might nonetheless be non-fundamental, because algorithms typically can easily report their not knowing. --Abram On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept or event is a product of the Danger, Will Robinson!, reptilian part of our brain. When we don't know we don't know something we react with a feeling of not knowing as a survival response. Then, having survived, we put the thing not known at the head of our list of things I don't know. As long as that thing is in this list it explains how we can come to the feeling of not knowing it so quickly. Of course, keeping a large list of things I don't know around is probably not a good idea. I suspect such a list will naturally get smaller through atrophy. You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. And... Another intuition tells me that the list of things I don't know, might generate a certain amount of cognitive dissonance the resolution of which can only be accomplished by seeking out new information (i.e., learning)? If so, does this mean that such a list in an AGI could be an important element of that AGI's desire to learn? From a functional point of view, this could be something as simple as a scheduled background task that checks the things I don't know list occasionally and, under the right circumstances, pings the AGI with a pang of cognitive dissonance from time to time. So, what say ye? Cheers, Brad --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
I think I decided pretty quickly that I don't know any words starting with foml. I don't know if this is a clue On 7/28/08, Abram Demski [EMAIL PROTECTED] wrote: It seems like you have some valid points, but I cannot help but point out a problem with your question. It seems like any system for pattern recognition and/or prediction will have a sensible I Don't Know state. An algorithm in a published paper might suppress this in an attempt to give as reasonable an output as is possible in all situations, but it seems like in most such cases it would be easy to add. Therefore, where is the problem? Yet, I follow your comments and to an extent agree... the feeling when I don't know something could possibly be related to animal fear (though I am not sure), and the second time I encounter the same thing is certainly different (because I remember the previous not-knowing, so I at least have that info for context this time). But I think the issue might nonetheless be non-fundamental, because algorithms typically can easily report their not knowing. --Abram On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept or event is a product of the Danger, Will Robinson!, reptilian part of our brain. When we don't know we don't know something we react with a feeling of not knowing as a survival response. Then, having survived, we put the thing not known at the head of our list of things I don't know. As long as that thing is in this list it explains how we can come to the feeling of not knowing it so quickly. Of course, keeping a large list of things I don't know around is probably not a good idea. I suspect such a list will naturally get smaller through atrophy. You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. And... Another intuition tells me that the list of things I don't know, might generate a certain amount of cognitive dissonance the resolution of which can only be accomplished by seeking out new information (i.e., learning)? If so, does this mean that such a list in an AGI could be an important element of that AGI's desire to learn? From a functional point of view, this could be something as simple as a scheduled background task that checks the things I don't know list occasionally and, under the right circumstances, pings the AGI with a pang of cognitive dissonance from time to time. So, what say ye? Cheers, Brad --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
In fact, had you asked me right away, I'd have said I don't know any words starting with fom. But on some reflection I was able to think of foment. Certainly there's some kind of habituation/reflex involved here, where a word falls off in familiarity as you sound it out. I don't know about other kinds of not knowing. :| On 7/28/08, Eric Burton [EMAIL PROTECTED] wrote: I think I decided pretty quickly that I don't know any words starting with foml. I don't know if this is a clue On 7/28/08, Abram Demski [EMAIL PROTECTED] wrote: It seems like you have some valid points, but I cannot help but point out a problem with your question. It seems like any system for pattern recognition and/or prediction will have a sensible I Don't Know state. An algorithm in a published paper might suppress this in an attempt to give as reasonable an output as is possible in all situations, but it seems like in most such cases it would be easy to add. Therefore, where is the problem? Yet, I follow your comments and to an extent agree... the feeling when I don't know something could possibly be related to animal fear (though I am not sure), and the second time I encounter the same thing is certainly different (because I remember the previous not-knowing, so I at least have that info for context this time). But I think the issue might nonetheless be non-fundamental, because algorithms typically can easily report their not knowing. --Abram On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept or event is a product of the Danger, Will Robinson!, reptilian part of our brain. When we don't know we don't know something we react with a feeling of not knowing as a survival response. Then, having survived, we put the thing not known at the head of our list of things I don't know. As long as that thing is in this list it explains how we can come to the feeling of not knowing it so quickly. Of course, keeping a large list of things I don't know around is probably not a good idea. I suspect such a list will naturally get smaller through atrophy. You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. And... Another intuition tells me that the list of things I don't know, might generate a certain amount of cognitive dissonance the resolution of which can only be accomplished by seeking out new information (i.e., learning)? If so, does this mean that such a list in an AGI could be an important element of that AGI's desire to learn? From a functional point of view, this could be something as simple as a scheduled background task that checks the things I don't know list occasionally and, under the right circumstances, pings the AGI with a pang of cognitive dissonance from time to time. So, what say ye? Cheers, Brad --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. Does it email you when this occurs? xD On 7/28/08, Eric Burton [EMAIL PROTECTED] wrote: In fact, had you asked me right away, I'd have said I don't know any words starting with fom. But on some reflection I was able to think of foment. Certainly there's some kind of habituation/reflex involved here, where a word falls off in familiarity as you sound it out. I don't know about other kinds of not knowing. :| On 7/28/08, Eric Burton [EMAIL PROTECTED] wrote: I think I decided pretty quickly that I don't know any words starting with foml. I don't know if this is a clue On 7/28/08, Abram Demski [EMAIL PROTECTED] wrote: It seems like you have some valid points, but I cannot help but point out a problem with your question. It seems like any system for pattern recognition and/or prediction will have a sensible I Don't Know state. An algorithm in a published paper might suppress this in an attempt to give as reasonable an output as is possible in all situations, but it seems like in most such cases it would be easy to add. Therefore, where is the problem? Yet, I follow your comments and to an extent agree... the feeling when I don't know something could possibly be related to animal fear (though I am not sure), and the second time I encounter the same thing is certainly different (because I remember the previous not-knowing, so I at least have that info for context this time). But I think the issue might nonetheless be non-fundamental, because algorithms typically can easily report their not knowing. --Abram On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept or event is a product of the Danger, Will Robinson!, reptilian part of our brain. When we don't know we don't know something we react with a feeling of not knowing as a survival response. Then, having survived, we put the thing not known at the head of our list of things I don't know. As long as that thing is in this list it explains how we can come to the feeling of not knowing it so quickly. Of course, keeping a large list of things I don't know around is probably not a good idea. I suspect such a list will naturally get smaller through atrophy. You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. And... Another intuition tells me that the list of things I don't know, might generate a certain amount of cognitive dissonance the resolution of which can only be accomplished by seeking out new information (i.e., learning)? If so, does this mean that such a list in an AGI could be an important element of that AGI's desire to learn? From a functional point of view, this could be something as simple as a scheduled background task that checks the things I don't know list occasionally and, under the right circumstances, pings the AGI with a pang of cognitive dissonance from time to time. So, what say ye? Cheers, Brad --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription:
Re: [agi] How do we know we don't know?
Jim Bromer wrote: On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. Brad My guess that initial recognition must be based on the surface features of an input. If this is true, then that could suggest that our initial recognition reactions are stimulated by distinct components (or distinct groupings of components) that are found in the surface input data. Jim Bromer Hmmm. That particular query may not have been the best example since, to a non-Norwegian speaker, the phonological surface feature of that statement alone could account for the feeling of not knowing. In other words, the word fomlepung just doesn't sound right. Good point. But, that may only explain how we know we don't know strange sounding words. Let's try another example: Which team won the 1924 World Series? Cheers, Brad --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Brad Paulsen wrote: All, Here's a question for you: What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. It could be that your brain keeps a list of things I don't know. I tend to think this is the case, but it doesn't explain why your brain can react so quickly with the feeling of not knowing when it doesn't know it doesn't know (e.g., the very first time it encounters the word fomlepung). My intuition tells me the feeling of not knowing when presented with a completely novel concept or event is a product of the Danger, Will Robinson!, reptilian part of our brain. When we don't know we don't know something we react with a feeling of not knowing as a survival response. Then, having survived, we put the thing not known at the head of our list of things I don't know. As long as that thing is in this list it explains how we can come to the feeling of not knowing it so quickly. Of course, keeping a large list of things I don't know around is probably not a good idea. I suspect such a list will naturally get smaller through atrophy. You will probably never encounter the fomlepung question again, so the fact that you don't know what it means will become less and less important and eventually it will drop off the end of the list. And... Another intuition tells me that the list of things I don't know, might generate a certain amount of cognitive dissonance the resolution of which can only be accomplished by seeking out new information (i.e., learning)? If so, does this mean that such a list in an AGI could be an important element of that AGI's desire to learn? From a functional point of view, this could be something as simple as a scheduled background task that checks the things I don't know list occasionally and, under the right circumstances, pings the AGI with a pang of cognitive dissonance from time to time. So, what say ye? Isn't this a bit of a no-brainer? Why would the human brain need to keep lists of things it did not know, when it can simply break the word down into components, then have mechanisms that watch for the rate at which candidate lexical items become activated when this mechanism notices that the rate of activation is well below the usual threshold, it is a fairly simple thing for it to announce that the item is not known. Keeping lists of things not known is wildly, outrageously impossible, for any system! Would we really expect that the word ikrwfheuigjsjboweonwjebgowinwkjbcewijcniwecwoicmuwbpiwjdncwjkdncowk- owejwenowuycgxnjwiiweudnpwieudnwheudxiweidhuxehwuixwefgyjsdhxeiowudx- hwieuhyxweipudxhnweduiweodiuweydnxiweudhcnhweduweiducyenwhuwiepixuwe- dpiuwezpiweudnzpwieumzweuipweiuzmwepoidumw is represented somewhere as a word that I do not know? :-) I note that even in the simplest word-recognition neural nets that I built and studied in the 1990s, activation of a nonword proceeded in a very different way than activation of a word: it would have been easy to build something to trigger a this is a nonword neuron. Is there some type of AI formalism where nonword recognition would be problematic? Richard Loosemore --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
It is fairly simple at that point, we have enough context to have a very limited domain world series - baseball 1924 answer is a team, so we can do a lookup in our database easily enough, or realize that we really dont have a lot of information about baseball in our mindset. And for the other one, it would just be a strait term match. James Ratcliff ___ James Ratcliff - http://falazar.com Looking for something... --- On Mon, 7/28/08, Brad Paulsen [EMAIL PROTECTED] wrote: From: Brad Paulsen [EMAIL PROTECTED] Subject: Re: [agi] How do we know we don't know? To: agi@v2.listbox.com Date: Monday, July 28, 2008, 4:12 PM Jim Bromer wrote: On Mon, Jul 28, 2008 at 2:58 PM, Brad Paulsen [EMAIL PROTECTED] wrote: All, What does fomlepung mean? If your immediate (mental) response was I don't know. it means you're not a slang-slinging Norwegian. But, how did your brain produce that feeling of not knowing? And, how did it produce that feeling so fast? Your brain may have been able to do a massively-parallel search of your entire memory and come up empty. But, if it does this, it's subconscious. No one to whom I've presented the above question has reported a conscious feeling of searching before having the conscious feeling of not knowing. Brad My guess that initial recognition must be based on the surface features of an input. If this is true, then that could suggest that our initial recognition reactions are stimulated by distinct components (or distinct groupings of components) that are found in the surface input data. Jim Bromer Hmmm. That particular query may not have been the best example since, to a non-Norwegian speaker, the phonological surface feature of that statement alone could account for the feeling of not knowing. In other words, the word fomlepung just doesn't sound right. Good point. But, that may only explain how we know we don't know strange sounding words. Let's try another example: Which team won the 1924 World Series? Cheers, Brad --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
Re: [agi] How do we know we don't know?
Well, I can put the components of who won the 1924 world series and determine that I don't know the answer (unless it was the Yankees since I think Babe Ruth might have done something significant in 1924). However, the fact that I was able to interpret the sentence without seeming to search for the answer to what it means suggests that there were some underlying processes at work. Of course there were extensive underlying processes, so the possibility that we have various lists of things that we know we don't know along with lists of things that we think we do know is a possibility. But for a feasible AI I think we have to find a way to compact all that information in a way that would make it accessible for quick location of information. Maybe it could be done through generalizations and the like. The point that we know it was a baseball team is very important because it might help us to delineate some of the processes of thinking with the hope of finding feasible ways to do this might be done in an AI program. Jim Bromer On Mon, Jul 28, 2008 at 5:23 PM, James Ratcliff [EMAIL PROTECTED] wrote: It is fairly simple at that point, we have enough context to have a very limited domain world series - baseball 1924 answer is a team, so we can do a lookup in our database easily enough, or realize that we really dont have a lot of information about baseball in our mindset. And for the other one, it would just be a strait term match. James Ratcliff ___ James Ratcliff - http://falazar.com Looking for something... --- On Mon, 7/28/08, Brad Paulsen [EMAIL PROTECTED] wrote: Hmmm. That particular query may not have been the best example since, to a non-Norwegian speaker, the phonological surface feature of that statement alone could account for the feeling of not knowing. In other words, the word fomlepung just doesn't sound right. Good point. But, that may only explain how we know we don't know strange sounding words. Let's try another example: Which team won the 1924 World Series? Cheers, Brad --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=108809214-a0d121 Powered by Listbox: http://www.listbox.com
