John Vervaeke - Thinking & Reasoning - Lecture 5 (wW7Sgr0z-xc)

https://youtubetranscript.com/?v=wW7Sgr0z-xc

We then went into the next important issue, another indisputable line of argumentation and evidence converging with earlier work. And this was the whole question concerning transfer. And we started by taking a look at the work of Dick and Julio. A lot of noise coming. Alright. And what we saw there was a kind of astonishing, almost counterintuitive failure of transfer between a description of a general attack on the fortress and the dunker radiation problem. We saw a response from that by Lockhart, Lemon and Dick in 1988. A response to the use of a distinction from William James between sagacity and learning, or between procedural processing and propositional processing. We took a look at that in connection with the multiple marriage problem. And we found that what was missing from the original Dick and Julio experiment was procedural similarity. And that procedural similarity seems to be much more important for insight transfer than propositional similarity, suggesting a huge procedural component to insight. We then followed that up with replication. And we saw similar work coming out of Adams et al. and Medium and Beg. And then along the way we talked about how this lines up with independent work coming out of memory about transfer appropriate processing. Which actually showed some academic or theoretical and scholastic integrity on Lockhart’s part because he was actually providing arguments and evidence for transfer appropriate processing. Which was the theory that overturned his view of levels of problems. We then started to brought this all together towards and make connections with Kaplan and Simon by taking a look at the work of Dick and McGarry, the same Dick that did the work with Lockhart. And that was a very counterintuitive proposal that a series of failed attempts at insight problem solving would actually better facilitate spontaneous transfer of solution than previous solutions. And the reasoning behind that was of course the idea that such a bank would be more procedurally similar to the current problem that you’re not solving. And increase the breadth against which you could apply the notice and variance heuristics so as to reformulate your current problem. And that received strong empirical confirmation. So the issue, and then we finished up with another, we returned because we talked about the whole circle, to Lockhart and another is looking at Dick and Lockhart looking back on the work in 1995. And maybe another important theoretical connection about the relationship between insight and overcoming the tyranny of automaticity. And I suggested to you that automaticity is a procedural way of understanding what a constraint is. And that de-automatization will be a procedural way of understanding what constraint relaxation is. And that we also are going to see evidence of mindfulness de-automatizes cognition and that’s how it relaxes constraints and how it affords insight. So not all of those connections have been made. Some of them I’m showing you to predict from what we have done. Of course by now all of these predictions are actually post-dictions. And the philosophy of science and science itself doesn’t know what to do with post-dictions in terms of their epistemological standing. But you have to understand that at one point they were predictions and then the predictions have increasing empirical confirmation. Okay. We then took a look at some important connections between this idea of transfer from the past and an important popular idea about the facilitation of transfer, namely incubation. And then we talked about how popular this notion is and how old it is. It was from Wallace’s book in 1926 and we looked through the four stages and the incubation stages, the one with zero and the other one. And we noted that for all of its popularity by the time of 1995 when Sieffert and Auer were doing their work on incubation, attempts to find a consistent support, empirical support for incubation were not successful. The confirmation was always inconsistent. The effect would seem to appear and disappear. Sieffert and Auer want to argue that they can remedy that situation and they do this in a theoretically sophisticated way. Again, science is driven by experimental competition and theoretical debate and innovation within theoretical debate. They are proposing to add an important extra element to the notion of incubation in order to explain the inconsistency of the results. And that element of course was the role of the environment which had not been discussed in previous work and especially not present in Wallace’s classical formulation of what incubation is. We then looked at the somewhat rhetorically not fraudulent, sneaky, but a little bit like move that they make where they present three perspectives you could have on incubation and insight. One is the business as usual typified by people like Weisberg and Alva and even in a more sophisticated sense by Kaplan and Simon. And then the opposite is the wizard Merlin perspective is supposed to be typified by people like Feynman, a supernatural sort of ability. And then we found out that that wasn’t as accurate as Feynman had been engaging in quite a bit of drama. But he was still Feynman so you know, you have to make that or whatever, make up that as you will. And then of course Sievert and Al placed their position squarely between business as usual and the wizard Merlin perspective and they call it the prepared mind perspective. The prepared mind perspective is based on the opportunistic assimilation hypothesis. The opportunistic assimilation hypothesis. And the opportunistic assimilation hypothesis is that as you try to solve an insight problem and you impass, that creates memory indexing. That information is especially sensitized in memory. And then what is needed, that prepares the mind, but what is needed is some change of context so that the context now contains the relevant information for solving the previously unsolved insight problem. That of course is posed to the standard explanations of what’s going on in incubation. The standard explanations are people are lying, which is more politely called the conscious work hypothesis. And you always have to take that seriously, right? In every psych experiment there is the very little possibility that people are trying to deceive you or please you. That’s what makes psychology so much harder than physics. Imagine if the objects you were studying in physics were constantly trying to deceive you and please you. You couldn’t get basic physics going probably. That would be very, very hard. So you always have to evaluate the study against the qualities of the objects that are being studied. Next is the fatigue dissipation hypothesis, that all that happens in incubation is that fatigue is dissipated and people come back more alert, ready to go. Next is the selective forgetting, and we’ll see how that is going to get some rejuvenation in the work of Siegel, which we’ll talk about today. And then the sort of Freudian, Jungian idea that what’s happening in incubation is the material is passing into the unconscious with its miraculous powers, and genius is doing things in order to bring about the insight. What Seifert and Al did was run a series of experiments showing that simple time away was not a predictor of incubation having any efficacious effects. Instead, what was needed was that the person had to have impasse, this is the prepared part of the prepared mind perspective, and the change in environment had to contain the missing relevant material. When those were the case, then independent of time away, incubation was effective. And when those two factors were not present, incubation was not effective, and that was the reason why we need to provide quite good evidence for the prepared mind perspective and the opportunistic assimilation hypothesis. And I believe that’s where we got to, is that correct? I already reviewed the format for the test and the topic proposal, correct? So there’s no reason to go over that again unless anybody has some pertinent questions that need to be addressed. Alright, so let’s start back at it. Around the same time, in fact in the same anthology on insight, in which Seifert and Al published their work on incubation, Smith, who you’ve already heard mention, Seifert and Al mentioned Smith’s review of previous empirical work on incubation, Smith also published an article in which he proposed an alternative account of what might be going on in incubation, an alternative account. So what Smith argues is he gives an alternative account, a very, well that’s not quite what it is, it’s not exactly a contrary, it’s alternative, it’s not quite clear, we’ll have to get clear about the degree to which it competes with the opportunistic assimilation hypothesis. But let’s discuss it first and then we will go back and talk about how these various positions on incubation stack up against each other and how they’re competing and not competing. So Smith thinks that what’s going on in incubation is the overcoming of memory interference. So he’s giving a different idea of what might be going on in passing, or what the gestaltists earlier on had called fixation. So he thinks what’s going on in Smith is memory interference. And then what’s going on in incubation is you’re overcoming this. So what he’s interested in is he’s interested in phenomena in which you get into what he at this stage of his theorizing called a mental rut. So he proposes something that’s analogous to mental repose is something that’s analogous to what can happen in the tip of the tongue phenomena. Which is important because some of you mentioned this earlier when we were talking about feeling of warmth. I pointed that the feeling of warmth isn’t equivalent to the tip of the tongue, but tip of the tongue seems to involve feeling of warmth being active. Because in tip of the tongue you have a very strong feeling of warmth, a very strong feeling of knowing. You know this and yet you can’t recall it. This happened to me a few days ago. It was a weird experience. I was trying to remember the lead singer from the doors. And I would go, Jim? And my brain would go, Jimi Hendrix? And I’d go, no, that’s clearly wrong. Like in so many ways. And then I’d try again. And I’d go, Jim? And what happened was exactly what he was talking about. So what happens is you get an incorrect recall and then that material is primed. And it’s also primed by your reaction to it because it’s salient. It’s like, no, that’s clearly wrong. So that’s memory indexing. Remember we’ve talked about this? That was wrong. So failure, failure, primed, indexed, super sensitive. And then four seconds later you say, who is the lead singer of the door? And it’s all ready to go. Jimi Hendrix! Because you start with Jimi Hendrix. It’s Jim Morrison by the way. For those of you who know about the doors. So there’s an excellent Jimi Fallon, another Jimi, doing a rendition of the doors, doing the reading rainbows song. It’s on YouTube. Watch it. It’s really good. But you have to watch some Jim Morrison tapes first. Some videos first. So what is going on there is this vicious cycle. So what he means by a mental rut is exactly this self-reinforcing feedback loop. It’s a self-organizing process. Which is really interesting because he’s starting to introduce a dynamical notion of fixation. So his idea is that what you have to do, how incubation works, and of course that’s what I had to do. I mean this is all anecdotal right now and we’ll come back and look at it in experimental evidence. I’m using the anecdotes to try and explain the phenomena. What I had to do was stop trying. Which sounds like SuperZen, right? Stop trying to remember Paguan and then later it will come to you. And so that’s what I did. I stopped trying to remember. And sure enough, what an hour later was right, Jim Morrison. And so the idea is that what happens in incubation is by taking attention away, you reduce the salience of the stimulus. So it drops below a threshold such that the self-organizing cycle ends. So let’s say that again because that was a lot. That was a big word salad. What happens in incubation is you redirect attention so that the salience of the stimuli drops below a threshold so that the self-organization of the mental rock breaks up. You basically are breaking up a self-organizing system by reducing a linchpin aspect of salience. So what we have now are three different potential candidates. Again, it’s not clear what the relationship is with each other so we’re going to spend a bit of time teasing that up. We have three potential candidates, new candidates, for what’s going on in incubation. So aside from the Wallace ones which have failed to get good confirmation, the review by Smith, and have been to some significant degree disconfirmed by Seifert and Gahn. What are the new potential hypotheses for what’s driving incubation? Well the first one we have of course is opportunistic assimilation. So let’s, I’ll number these so I can refer to them by their numbers. Okay, I’m going to slightly switch the term now. We just, Wayne Smith, in 95 he called it mental rut. Later on he refers to mental set transformation. I’m going to use the more recent nomenclature. And then one that was, we mentioned last time, which is, and again, we’re trying to tease apart the relationship between them, that this idea of overcoming automaticity, which we talked about last time as reversing, like reversing the process. I don’t need to do that right now. We’ll talk about that later about relaxing and strengthening. Okay, so those are the three we’ve sort of brought up. Now, we should note that the first two are in theoretical competition. The first two are in theoretical competition. The first, because the first, number one, right, proposes a passive model, whereas the second proposes an active model. There’s things you have to do in order to sort of base your theory on the theory of the past. The second proposes an active model. There’s things you have to do in order to sort of bake up, break up the mental set, the mental rut. You have to redirect attention in some active manner. Also, the first one is highly dependent on the environment. Number one is highly dependent on the environment. In fact, that’s its chief theoretical claim, right? But the second one, not as much. Now, the environment can cause change for number two, but it need not. The environment can cause a change here, but it need not. Now, I would argue that the third one is not in direct competition with the other two, because it’s a higher order theory. Now, I would argue that the third one is not in direct competition with the other two, because it’s a higher order theory. It doesn’t talk about types of processes, it talks about a manner of processing. However, it tends to be more compatible with mental set transformation, because this carries with it, as I’ve argued, the idea of procedural constraints being broken. And what I’m actually going to argue is that all of these views are going to contribute to the answer in a significant way as to what’s going on in incubation and its relationship to insert. All right. So, we have these three contenders. These two are in direct competition. This one’s higher order, but it’s sort of more consonant with two than one, but that’s a loose connection. You’ve got the opposition between one and two. So, there’s been some recent reviews. So, and… And, from 2009, did a meta-analysis of all of the experimental work to date that had been done on incubation. And, Dodds et al., in 2003, did a theoretical review. And, we found that on balance, the evidence points to incubation as existing, but the results are still very mixed. And, that was supported by this meta-analysis. So, as we said last time, but we seem to be needing to return to it, this overall pattern points to the conclusion that although there is a phenomenon at work, the theoretical construct still needs significant work. So, this is a general methodological principle. When you’re getting enough evidence for the existence of a phenomenon, but that evidence remains consistently inconsistent, and you’ll allow me to say that, that is very good evidence that your theoretical construct is in some way malformed. So, Siegel, the person I’m going to talk about, sits in here, but this is basically a review of the evidence that supported this. So, Siegel, 2004, attempts exactly that, attempts to come up with a more consistent and comprehensively applying construct. And, then we’re going to see that there’s going to be a significant response by Smith et al. And, then we’ll follow that up with some really weird stuff involving turning on light bulbs and weird stuff like that. Alright. So, first of all, he very directly wanted to study the connection. So, a lot of the incubation effects are more vaguely about creativity or solving a problem. So, what Siegel wanted to do was more directly study the connection between insight problem solving. And, in order to do that, breaks, you know, we stop working on the problem, only allowed after people had been passed on a problem. In connection with the break, he has what he calls the Attention Withdrawal Hypothesis, which builds on some of the stuff I suggested to you when we looked at Smith’s model. So, what he argues is, and we’ve sort of already seen, yes? Can you repeat the hypothesis you just said? Yes, the Attention Withdrawal Hypothesis. So, he builds on sort of what we’ve been, sort of something we’ve been building towards. The idea that what’s going on in problem formulation, right, is that there is this dynamic projection of salience onto the information. And so, he argues that what you have to do is you have to withdraw attention, because until you do, the chances for assimilating, and that language, of course, is also hard for you to see for now. You can see how he’s trying to bring different ideas together. The chances for assimilating the needed cue are low until attention is withdrawn from the inappropriate organizing assumptions or constraints or formulation. So, the idea is, as long as you’re, it’s almost as long as you’re looking through your problem formulation, that keeps it primed, and it therefore filters and skews what you find relevant and pay attention to. Until you withdraw attention away, right, and then the problem formulation can begin to dissipate. And what that allows is for the reorganization or the organizing principles to be weakened enough so that your salience landscape for the problem is altered. The salience of various cues starts to spontaneously change. So, this is a, so you see what he’s doing here, right? He’s trying to integrate the opportunistic assimilation hypothesis with a way of trying to get out of the mental rutting of Smith, but he’s also bringing back a now very much more revised version of the selective forgetting hypothesis. So, he’s trying to bring three things together here, right? He’s bringing in ideas of assimilation from Seifert and Al, right? He’s bringing in ideas of trying to break up that sort of vicious cycle between your problem formulation and the information that you’re getting through that problem formulation. And he’s making use of the selective forgetting hypothesis. That what happens is, but what he’s doing is he’s not just selective forgetting, he’s talking about how the redirection of attention is the crucial issue. Which may mean that that’s all that needs to be done is the redirection of attention in order to facilitate incubation. So, he’s modifying the selective forgetting hypothesis because he’s integrating it with these other ideas. Yes? Sorry, what was the first theory that he was trying to integrate? The opportunistic assimilation because he’s talking about needing to assimilate cues that you previously couldn’t assimilate. He’s obviously talking about, Michelle, right? Yes. He’s talking about mental set transformation because he’s talking about breaking up the way your problem formulation makes certain things salient that then reinforces the problem formulation in kind of an extended mental rut. That’s okay. Yeah. And it’s like selective forgetting except he’s putting not so much emphasis on forgetting as he is on the redirection of attention. Okay. Is that okay for everybody? Do you see what he’s doing and how he’s trying to sort of draw things together and come up with an alternative explanation? So, then he has a second hypothesis that he calls the returning act hypothesis which sounds like something from a carnival. It’s the returning act. The returning act hypothesis. All right? So, the idea here is the breaks and this is now where he does that transformation I just mentioned. The breaks sole function, this is his argument, the breaks sole function is to direct attention from the inappropriate framing. He’s going to directly challenge the idea of any kind of unconscious work going on. This redirection of attention alters salience and affords cue assimilation. Now, let’s stop here and I’m sorry, I’ll say that again. This redirection of attention alters salience and affords cue assimilation. Now, if this line of argument is right, it’s quite deflationary for the classical model of incubation. Yes? The first thing you said after you mentioned the hypothesis. The returning act hypothesis, the breaks sole function is to direct attention from the inappropriate framing. Is that the missing piece? Yes. Okay, so notice how deflationary this is for the classical model of incubation. Because technically, if this is right, you don’t even need to break. All you need to do is redirect attention in the right way. Moving away is not sufficient. Forgetting might not be enough. There might be a needed redirection of attention. Now, to be fair to you, you have to put the two hypotheses of Siegel together to get the difference from simply detaching from the problem. You had a question? Yes. Could it be that necessary means necessary for the problem reformation? Because if we’re saying that that is not necessary, then nothing is necessary because the problem is supposed to be solvable. But it might be that after you impasse, then the environment is necessary for you to break out of that impasse. Right. And then his point, and Alison’s point, was the Siegfried and Allen experiment didn’t provide evidence for that. It only provided evidence that the environment could help solve, address the reformulation problem. But that’s not the same thing as showing that it’s absolutely necessary for doing the problem reformulation. Because it depends on the degree to which you think that number two might be a significant factor. Because if that vicious cycle is going on, and maybe what’s necessary is just to break out the cycle, then it’s probably the case that in very many instances the environment is not necessary for breaking that up. And that’s what Siegel is going to try and provide evidence for. Is that okay for now? Yeah. I’m trying to not shut it down, but I’m trying to show you how we can sort of keep going with the debate. Yes. So, like just in an example, in the candle, pinning the candles to the bulletin board and then putting them in the box, he’s saying that if you were to see candles in a box, that’s not new information. It’s just making the box salient in a different way. That’s right. Okay. Go on. Check. You’re doing very good work today. You are. Sorry, I mean that sincerely. People are laughing. I think lots of people tell you when you’re not doing a good job, like you can be told when you’re doing a good job. Yes. I have a question. So, would it be possible, like I thought that just because of the way some people are, that they require cheating in their pockets? That’s an excellent question. And you know what? That brings up a whole issue in which there has been an inadequate amount of empirical work done and theoretical reflection done on this, which is the interaction between personality and these cognitive variables. I totally, and one of the interesting things that might come up, right? I think what you’re saying is plausible. Let’s just go with that. It might be that we could come up, if there are significant interactions between personality and these cognitive factors, we might have additional kinds of more objective tests other than self-reports and linguistic usage for assessing personality factors. For example, maybe people who are highly open can do a lot more self-generation of reframing that would break the mental running, whereas perhaps people that are more extroverted might need to go out into the environment in order to, I don’t know, I’m just speculating in that negative sense of speculation, but I think the connection, now there’s been some, but not very much work has been done. And not to the extent where it might be that the connection could afford more, I mean I hesitate to use this, more objective ways of testing for personality. You wanted to follow up on that? I think I have been kind of noticing this trend, not in this class, not in other classes too, but just use a lot of things that we’re doing in science and stuff, is like we’re trying to create a standard for a very broad population. So everyone is so different in their own way, but we’re trying to create what the standard means, and apply it to almost everyone. That’s kind of not exactly correct, because it changes so much depending on the person. It can and it can’t. I mean, so, what’s your name again? Andrew. Andrew. So Andrew is referring to, I mean it’s called the weird problem. So we take people from Western, Western educated, industrialized, rich, democratic, these are the defining characteristics of the people we use in our experiments. And they represent at most 12% of the Earth’s population. Now, let’s be careful. This is an important issue. So I’m not trying to slam you. Remember that science is about trying to get the broadest applying generalizations you can. So it’s not that science is doing something inappropriate to what it’s trying to do. You’re trying to do that. And this is an issue across all of science, because nature doesn’t tell you how important the similarities or the differences are between any two phenomena. And logically, I mean this really technically, any two objects are indefinitely similar. They share an indefinitely large number of properties. Or they’re indefinitely dissimilar. You can find a whole bunch of properties that they don’t share. So that’s the logical situation. By the way, this is a GAN-wide replication is tied to your theoretical argument, because how we decide whether we’re going to count the similarities or the differences is more important is theoretical debate. Because you can’t decide that issue empirically, because you have to make that decision in order to get your empirical data. Am I going to? So the thing is, that needs to be mentioned and brought up. Toy, thank you for being so patient, by the way. It needs to be mentioned and brought up. But on the other hand, it’s not a problem specific to psychology. We like to emphasize the fact, in certain contexts, we like to emphasize the fact that we’re, as individuals, we’re very different from each other, which is, by the way, a very Western thing to do. Right? And it has effects on our cognition. But in other contexts, of course, we like to emphasize our similarities. For example, people tend to think everybody is basically similar in terms of their moral, they should or should be similar in terms of their moral judgments. So you have to be very careful about, even within psychology, we bounce around on how important we give, how much importance we give to similarity and the differences. So I had to give you a very long, sort of philosophically-linked answer, because it was a philosophically-linked question you asked. You made a good point. Thank you. I mean, I give a lot of time and thought to this, because this is kind of a central issue. This is why I advised all of you, at some point, if you want to be good scientists, you should take at least one course on the history and the philosophy of science, just so you spend some time thinking about these deeper kinds of issues. So thank you for letting me. So for the mental set, I thought this was compared to the attention withdrawal because we’re turning up. For the mental set, it’s just you need to stop paying attention, but it doesn’t really matter how you redirect that attention, whereas for the second one, it matters how you redirect it. So I’m curious that in the second one, if you were just good at stopping paying attention or you couldn’t maintain attention for a very long time, or for instance, you had people with ADHD, there would be a difference between the two hypotheses, because someone with ADHD would be really good at stopping paying attention, or they would stop paying attention more often, but they wouldn’t necessarily be able to direct reframe as often, it would be more similar to the majority in terms of one day reframe, but not in how often. Has anyone tested the differences to see? We’re going to take a look at that. We’re going to talk about aspects of the redirection, and there’s sort of a Goldilocks zone of the redirection of attention that is needed to get the facilitation of that. So yes, I think if I understand your question, I’m going to show you, I think, an attempt to address that question. I also want to point out something that you were at least possibly to find, which is the way this is starting to become more crucial, because if you simply redirect your attention but you haven’t overcome automaticity, chances are you will merely repeat. So it’s a redirection that might also involve some kind of de-automatization, which might mean, in fact, again, that incubation isn’t needed at all, perhaps if something like mindfulness was responsible for de-automatization, you would just have to bring mindfulness to bear, and you wouldn’t have to break from the problem at all. So this has the potential to be quite deflationary towards incubation. Yes? Are you gesturing towards an appointed process? I think that is assimilation takes place during the sedation of the breath, but can you complement the process of accommodation in order to open yourself up to new information? Yeah, I am moving towards that. I’m not going to try and complete that argument here, but I’m eventually going to move into that argument when we take a look at the work of Stefan and Nixon and others, and the self-organizing criticality that had that work in insect problem solving. Okay, so I think that’s very precedent on your part. Cheetah? When you’re talking about redirection, how are they knowing what to redirect to? Right. So that’s part of the issue. Now, I think what I’m going to try and argue, and I hope you find it at least possible, that it’s because this is so procedural, it’s not so much the content of the redirection, but the manner in which you redirect attention that’s going to turn out to be crucial. And that’s not just out of my hat, because we’ve been seeing a lot about how much the manner of things and the procedurality is crucial within insect problem solving. Okay, so what were some of the empirical results of the experiment? So what they basically, what he basically did, right, was he had people in pass, and then he had variables, he would vary the duration of the break. And what I’ll do is I’ll show you the results, and that tells you what the variables were. You found that the duration of the break was irrelevant. The duration, purely the duration of the break was irrelevant. You did find that breakers, so weird now, breakers did better than continuous workers. So again, some kind of effect is being replicated again. But interestingly, he found that a demanding task in the break was better than a non-demanding task. So while you’re breaking, if you do a non-demanding task, that’s not as good as doing a demanding task. If you break and do a demanding task, cognitively and intentionally demanding, in the break, you’re much more likely when you return to solve your insect problem, than if you do a non-demanding task. Now notice how this finding, which is pretty robust by the way, really challenges notions about fatigue, dissipation, or unconscious work. Because you’re really putting a lot of your attention and effort into something else that’s a demanding task. Yes? We’re going to have to modify that by the way. We’re going to have to get clear about how much demand is the right effort. Yeah? Can you speak a little more loudly? I’m sure people in that corner can’t hear you. Okay, when we say that this diffuses the idea that there’s unconscious work at play, do we mean more just the idea that there’s some kind of separate subconscious bank of work happening? Yes. Because I mean, if we’re talking about a specifically self-organizing emotional dynamics, looking at it as an unconscious… You have to be careful if you go to the other side, because if that thing can work independently of your conscious effort, then it’s always working, and taking a break would make no difference. So if it’s somehow triggered or activated by a relaxation from problem solving, then why would you do the demanding task? Because you’re in another problem. Presumably all your unconscious problem solving machinery is re-engaged and re-activated. Yeah. Indu? I don’t know if you said the word, but why does the mental transformation, how about just say that the environment has no… I was very careful to say all the claims is the environment is not necessary. It says the environment can make a difference. Okay, because… No, what they mean by environment, to be fair, is they mean what is meant by secrets and all, an alternative, like physical context from the original context in which the problem was posed. So like a different actual environment, not like changing the environment. Yeah. Because then like, you know like the work that Steve and Dick did about entropy and how it can change… That’s right. And that’s going to… yes. And that’s going to line up with what we’re going to talk about here. Exactly. Well done. Why, you guys are all doing really well today. Maybe I should just sit down. Yes, that’s exactly right. So again, but notice how entropy is a higher order manner of the way in which the information is being processed as opposed to any particular content or particular kind of information from the environment. So notice how even in what we’re talking about here, we’re shifting away from what Seifert and Al clearly meant by a relevant piece of content information in the environment. Is that okay? Yes. Now one thing we should notice is there’s a bit of a potential confound because as I’ve mentioned already and some of you sort of agree, you know the issue of how much demand hasn’t been completely reconciled. And there’s procedural similarity by the way from having a demanding problem you’re trying to solve during the break and the fact that you just didn’t solve a demanding problem. And we’ve seen the procedural similarity also enhances transfer. But instead, right, we will come back to that issue, but sorry, not instead. In addition, I want to point out that as this stands, this seems to be good evidence against the opportunistic assimilation hypothesis. Because the problem did not contain, the demanding task did not contain any relevant information to the initial problem. So again, it looks more like procedural aspects are what’s at work here rather than more propositional content aspects. Now I want to pick up on the discussion I have with Indu about some relationship to the environment might still be relevant in some way and to cut and rethink that relationship in more procedural terms. So what I want to suggest to you and give you some evidence for and some argument for is that what you want is another task that has a balance between being demanding, right, sorry, a balance between being too demanding and not demanding enough. Too demanding would be inappropriate priming and interference effects, especially if there’s too much procedural dissimilarity between your intervening task and your original insight problem. Not demanding enough and you’re not going to affect mental sap transformation. Can you run that by one more time? Sorry. So too demanding and not. Right, so too demanding and you’re going to get inappropriate priming and interference effects, especially if there’s too much procedural dissimilarity. And not demanding enough to withdraw, not demanding enough would mean, it’s not demanding enough to withdraw attention and afford mental sap transformation. So what I’m suggesting to you is that the shifting of attention is what’s doing most of the important work here and that that has to be very skillful in a sense in order to properly bring about mental sap dissolution and cue assimilation. So Siegel had a student, he had more than one student, that was a weird name to look at your, one of Siegel’s students was a guy by the name of Kahn and his thesis work was on exactly that. And he did a bunch of studies where he moderated the level of distraction, moderated the level of the distraction, how demanding the distraction was. And he found that what’s optimal for insight problem solving for the incubation effect is a moderate level of distraction, a moderate level of demand. Too little and not much incubation effect, too much and not much incubation effect. So what might be going on there? What’s going on there? It might be that we’re doing something, so this is now a conjecture, oh I’ll answer the question before I go into it. Are you guys asking independent questions or responding to my question? Not asking a question. Pardon me? Asking a question. That’s fine, so we’ll put my question aside and Emma, what’s your question? How do you define moderate? So to be fair to these people, you define it the same way that Siegfried and Malta. You have a lower bound and an upper bound, which you just stipulate and then, right? How much attention and effort people feel they’re having to direct to the distraction? So it’s sort of subjective. In that sense. But across people. Pardon me? But across people. But across people, yeah. So a more accurate description would be intersubjective. Okay. And that would be done prior? Yes. Okay. Yeah. Well, I mean, we could just operationalize the level and see if you get statistical significance for the entire group, which they did. Thomas, you wanted to say something? Yes. Just in terms of manipulation of distraction, then it seems like it would be a really cool spot for experimentally dealing with some of the other questions that were asked earlier about individual differences. Yep. And intentional factors. Yep. Agreed. I think the connection to personality, the connection to non-statistically normal attentional processing like a ADHD, I think these should be explored. Would this be one of those sort of situations where we’re looking at pathological cases to give a lot of clarity, you think, on how we’re going to… It can. The problem with how we looked at pathological cases in the past is we tend to look at them in a very sort of univariable way. We need a much more complex set of variables. And we should also balance the pathologies, and this is part of the part with positive psychology, over against the excellences, to also address the one-sidedness in our investigation of individual differences. We tend to only look at individual differences in a pathological sense. This is a part of the whole point of positive psychology. In terms of that, so what you’re suggesting there is that we should have sort of similar to what I think Pamela was asking about, how do you measure the demand and distraction of all of this? So you’re suggesting that we should have nonlinear measures to these? I think so. I mean, there’s probably interacting variables. I mean, I’m going to present arguments and evidence that you have multiple simultaneous variables interacting, and we’ve already seen that. We’re at least plausibly behind the cognitive leaping phenomena. So yeah, I think it follows that that’s the kind of measure we should be taking now. And we have ways, and this is the point of Stefan and Dixon, we have ways of turning univariate measures into multidimensional state spaces. We’ll talk about that when we do the Stefan and Dixon stuff. But it really messes up your standard sort of Newtonian idea of measurement, that what we’re always measuring is a single variable in a determinate fashion from the environment. We’ve sort of known for a long time that that’s false, but we still keep practicing as if it were true, even though now the math and the physics are getting much beyond that in a very significant way. Is that enough right now? I mean, I have strong suspicions that we’re at the cusp of a huge sort of shift, sort of at the fundamental framework of how we’re doing psychology and cognitive science, and how cognitive science and psychology integrate with neuroscience. It’s probably going to really kick in the bell when I’m retiring, so that’s really good. I can sort of help foreshadow it, and then retire it, and it will look like I was pressing it. And then I won’t be allowed to fail or something like that, predictions to be disconfirmed. I can delude myself for my whole day. Okay. Yes? Hi. Sorry. Why are you apologizing? You’ve been doing good work this class. So, I’m just a little bit confused. So he’s saying that there’s no unconscious processes happening, but… In the Freudian sense, obviously whenever you’re doing anything conscious, there’s unconscious processing going on. But he means in the Wallace sense, where the conscious mind is just turned off, and then the unconscious runs independently and doesn’t stop. But while you’re doing the demanding task, isn’t he saying that your unconscious mind is reformulating the problem? No, it’s not. Because presumably what it’s doing is doing what you were originally doing it. It’s involved in formulating the problem. But it’s still doing something. Right. Which is what I meant with my little thing about, he doesn’t mean that your unconscious isn’t at work. He means that it’s not separate and disconnected from your consciousness, off on its own during its Jungian and Freudian stuff to solve the problem. So it’s using what you’re doing in the demanding problem? Yeah, because most of the time you’re using most of your brain, whatever you’re doing. So when you move to the other… Right. So, I mean, this again is part of the attacking the 10% fable. So, yeah, when you shift to the demanding task, the idea is you’re engaging all of the same kind of processing, precisely because it’s a demanding task. So that processing is now dedicated to the demanding task, and it’s not free to work on the original problem. So what reformulates the problem then? What reformulates the problem is the redirection of attention. So that when you come back, right, the idea is that when you come back, you have a space or a gap in which you can reformulate the problem. Oh, so you don’t come back and the problem is reformulated. No. You come back and you now have the… your other attentional faculties are no… Okay, I got it. I got it. That’s what’s… I no longer primed or interviewing. No, no. That is how the returning act hypothesis is fundamentally different from the unconscious work hypothesis. Okay. So does it matter whether the break question is similar to the… No, in fact, it doesn’t have to… in fact, it didn’t contain any of the relevant… So they’re just completely separate? Right. So what seems to matter… I’m suggesting that what the things we might be concerned about is there’s a potential confound about how much procedural similarity and things like that. Right. Yes. I’m just wondering, good. This topic always gets a lot of discussion. So you’re saying that your unconscious isn’t working on your initial problem while you’re doing the new problem. Then why is there such a thing as having a too demanding break problem? It would be too demanding in that what I’m suggesting is too demanding precisely because it primes so much information in your distracting problem that you can’t shut it off when you go back to your initial problem. Okay. Again, this is how it’s different from the unconscious work hypothesis. So the information doesn’t even have to be relevant though in order for it to be described? That’s right. It’s more and more about the manner, right, which again doesn’t… so if it’s more about the manner than about the content, again, that lines up… sorry, that lines up against the unconscious work hypothesis. And also against the opportunistic assimilation hypothesis. Unless we trivialize the opportunistic assimilation hypothesis by meaning anything that happens external to you, which then of course the environment is always playing a crucial role. But I already had a useful discussion with Indu about that. Well, the distinction is between content versus a procedural difference. Yes. So the task doesn’t have to be relevant, but then it also can be too irrelevant, as I always said. Well, yeah, what I mean is when I said relevant, yeah, the task can’t contain any relevant content information. But what I meant is it can’t be too irrelevant in the sense that it can’t be too procedurally dissimilar. Okay. Because that’ll induce interference. That’s right. Just like, you know, your golf swing can interfere with your hockey swing. Yes. Just a clarification. Were these experiments in length theorizing still on insight problems, or was it a broader spectrum of what kinds of problems that they did? The Siegel work was on insight problem. I’m pretty sure the Kahn work was too, because he was following directly up on Siegel. It’s been a while since I’ve read that dissertation. I think it was a PhD dissertation. I’m pretty sure it was on insight problems. Yes. So Kahn found that the optimal distractor is moderate, that if you look at the other two extremes, so like no task at all versus like… So we know that no task at all in simple time and weight doesn’t do anything. Siegel provided evidence for that. So low distraction didn’t provide much facilitation, and too much didn’t provide much facilitation. How do we know that how much is too much? What if we go all the way to… You’re worried it might be a local minimum situation. Yeah. I don’t know. That’s a good question. So I guess there’s sort of an implicit plausibility argument that the curve seems to be like this. And until we have other reasons for believing, we need an alternative, independent reason why extreme distraction could be facilitative. We have pretty good evidence that extremely low distraction is non-facilitative. So it’s only the high end that we worry about, that we might be in a local minimum, is that issue. I agree that they didn’t directly address that. But again, it seems implausible why it would be a facilitator. Yes. Oh no, wait. Sara, you haven’t asked the question yet. Sorry about that. Could moderate distraction be similar to flow state, so it’s not too difficult and it’s not too boring? Well done. Very well done. So, yes. We’ll take a look at that when we take a look at the flow channel. And you need to get this balance between two demanding… So basically you need it demanding enough to sort of stretch your skills. We’ll come back to that. So that’s an excellent connection. In fact, I’m publishing a book chapter and also submitting a paper for review on the idea. And I’ll argue for that in class about deep connections between insight and flow. So the connection we made I think is a very good one. So wait till we talk about flow. But yes, excellent connection. I just have a question. So why can’t the task be very distinct if you don’t actually take in cues from it? Well, it’s not cues you’re taking in. It’s the sort of way in which you’re sort of patterning your attention and distributing salience. And if that becomes too far removed, right? You want it to be like… That’s why I say too dissimilar. You want it to be somewhat dissimilar, right? Because you need to do some restructuring. But if it’s just too different, it might just interfere when you try to bring it back. In fact… But what if it’s completely different? How is it going to interfere with the problem that you already have? But if it’s completely different, then presumably there’ll be no… There won’t be enough connection between this mental set running and the one you’re returning to. And you’ll just shut this one off and come back to this one. It needs to be similar enough that it will stay activated, but dissimilar enough that it will cause the initial one to be reformulated. But it’s got to get that just right. Because that’s why the thing about it can’t be like the way your golf swing can interfere with your hockey swing. Because the problem with them is they’re similar, but they’re dissimilar in the wrong ways. So your hockey swing doesn’t give you insight into how to improve your golf swing. It just messes it up. So that’s the same as you gain knowledge of the problem that just kind of helps you to see things differently. That’s right. That’s exactly the point. That’s exactly the point. And that’s how it’s different from the opportunistic assimilation hypothesis. Yeah. Go ahead. I was just going to say it’s like verbal overshadowing. And you were arguing about that, like how thinking about verbal processes biases it to one process. So if you’re doing a different task, it would be the same. That’s right. Excellent connection. Everybody’s doing very well. Very. So either it’s completely fortuitous or I have been educating you. I’ll choose the latter because it justifies my existence. Yes? Would you just not be able to reformulate the problem though? What would that mean? I think in one sense that’s true, but I’m not sure if I understand you. You would never get to the correct answer? Sure. You might be epistemically bounded for all kinds of problems. Or if you just keep turning up the dial on the distraction, would there eventually be a time, you know? I think I understand you. If it looks like I’m not, tell me I’m not. But to use some of the language we’ve already had, and I’m using it by analogy. So by analogy. No one is saying that this moderate level of distraction is like an algorithm where it is guaranteed. And by analogy, because we’re not saying any one of these are algorithmic or heuristic, but by analogy, what it does is increase the probability. Right? So that means that… It can fail. Yeah, but it also means that there could be something else that could help you. Or it could be, or I liked your original proposal. That you could just fail. You can just fail. We are epistemically bounded in many ways. Right, that maybe you just something doesn’t allow you to reformulate the problem. It is possible, right, it is possible that we will never come up with the insight that will reconcile relativity and quantum mechanics. It’s possible. But we are humans and we’re… No, no, we’re not in the image of God or anything. We’re just animals, right? And it’s possible that we just can’t get there. And we’ll talk about that when we talk about Bowdoin and the idea that there, you know, There might be ways in which we’re epistemically bounded. But look, you agree that everything else is epistemically bounded. The fact that you have an intelligent brain doesn’t mean your brain is capable of doing everything every problem posed to it. Right? But none of you think that, you know, you certainly don’t… Like the thing on Einstein’s… Sorry, reconciling relativity and quantum mechanics. Chips are really smart. I don’t think any chip is going to come up with a solution to that problem. I don’t think any five-year-old human being is going to come up with a solution to that problem. I’m not proving… You understand what I’m not doing. I’m not proving to you that we’re epistemically bounded. I think it’s very plausible that, like every other organism and earlier stages of ourself, We are also epistemically bounded and there are certain insights we can’t have. The problem with epistemic boundedness is you can’t distinguish between it’s species-based, like everybody will be epistemically bounded because it’s just the way the machinery works, culturally historically based because it has to do with the way in which the culture is limiting the concepts and their salience, or back to your point of degree to which it might be personality or individually differencially based. It might be that certain insights are impossible for me that are possible for you. I mean that seems to come up in relationship difficulties all the time. I mean that’s one of the things you fight against when you’re doing therapy with people. One person is capable of a kind of insight and the other person isn’t. And then that’s really hard when you get to that place in therapy. Steve? No, I was just thinking about this epistetic boundary. Maybe think about a million typewriters from a million monkeys come up and come up with their works of Shakespeare. If you have an infinite amount of time. The thing about us is we don’t. So that’s the main problem. I have a not really relevant question but I’m just curious about it. I noticed that when there’s no incubation in place, when people ruminate, it always leads to mood disorders. So I’m thinking what’s the relationship between emotions and incubation here. Yeah, so also the idea that rumination might be a variation on the mental rut formation. Which is what some people are now arguing. That’s a lot of really interesting connections you just threw out there all at once. That’s really cool. Yeah, because I have problems to relate to it like personally. Like when I try to… When you’re ruminating. Yeah, I ruminate a lot. So I noticed that it’s hard to actually let go the ruminations. And it always let the cycle goes on and on and on. It had to take on a life of its own. Unfortunately we don’t talk about that too much in this course. But I talk a lot about that kind of stuff in 371. Everything good is in 371. This is all just preparation. No, no. We do talk about it more there. I also like the connection you made to both emotion and implicitly towards personality factors driving this. Again, my response is I think these are important connections. And notice how they’re starting to come up. Notice… I’m sorry, I’m complimenting you. I’m not singling you out as a specimen or anything like that. Notice how as we’re shifting to this procedural, self-organizing way of talking about this, we’re starting to make a lot of interesting and potentially fruitful theoretical connections. Which means this is a viable alternative to be considering to the computational framework. Thanks again for the participation. It makes the class, I think, better for everybody. So before we move on to a response to Siegel, I want to sort of take a look at this idea. I’m trying to give it a little bit more plausibility as to why the manner of distraction might be important. So this comes from neural network theory. So just to briefly, for some of you who are unfamiliar with it, we mentioned it briefly when we talked about… not briefly, we talked about it for about 15 minutes or so. When we talked about cognitive leaping. So what happens in a neural network is a neural network is an artificial machine that is patterned somewhat, and we’ll talk more about this a little bit later, on how the brain works. What you have are multiple microprocessors that are basically hooked up to inhibit or excite each other. And what’s interesting about them is they… I mean, at least, I think it’s non-controversial now to say, because this was one of the explicit points of the people who created neural networks, is that neural network machines are, in some sense, theoretically, and it’s, importantly, theoretically different from computation. That was the connection to what was going on in the cognitive leaping. And I mentioned that neural networks are very good at multiple simultaneous constraints, because you have all of these processors working simultaneously together, et cetera. So all I need right now is the idea that neural networks are sort of relevant to what we’re talking about. Now, there’s a problem that neural networks face. One of the things I said to you is they’re really good pattern detectors. They’re really, really good pattern detectors and pattern completers. The problem with that is they tend to be too good. They tend to be too good. So let’s talk about an example of this. You’re trying to set up a neural network to detect tanks. A lot of these examples are military, which is disturbing, by the way. I have been approached by the military, by the U.S. Army, for some of my work. So, yep. Okay. Let’s go on. So what they did was they were showing pictures to the network that had tanks from a particular location. I think the intent was to have a machine that could be sat at this location to keep guard. And then they were also showing pictures of the same location without tanks. And you have to do this. You have to train networks. You don’t program networks. You train them. You give them input. They excite and inhibit each other in a very complex, self-organizing manner. And then put out an output. And then that output is then fed back into the network. And then you change the network. And so you have to train it in all these cycles. So you show me tons and tons of pictures. It takes a long time. And then eventually it looked like the network was really working. You showed a picture with the tank and it would go, tank. Show a picture without the tank. No tank. We didn’t talk, but I’m just… Although you could do that. Now that’s not so hard. And that’s an amazing thing because being able to recognize something like that is very tricky and very difficult. Because the tanks are going to be in all kinds of different orientations and there’s going to be shadows on them. And then you show them a picture. And they’re like, wow, look, it is doing this. Yay! Right? And you gloriously say, soon we will have our own robots. And then when you try it with the real tanks, it didn’t work at all. So why? Initially it was like, what’s going on? And went back in and they found that all the pictures that had been taken with tanks had been taken in the morning. And the pictures had been taken in the afternoon. So you know what the network had learned? It had learned the much easier problem of distinguishing morning from afternoon. And it just happened that in the data, morning correlated with tank and afternoon correlated with no tank. Of course, that doesn’t translate to the real world. You can’t run onto a battlefield at 3 p.m. in the afternoon and say, you can’t have a tank, it’s the afternoon. This is a version of what’s called overfitting to the data. It’s a version of, some of you know this, the poverty stimulus argument. It’s very similar to the poverty stimulus argument in linguistics. It’s the projection problem in the philosophy of science. The same problem has different names. You guys in psychology, you probably notice it more with something like statistical bias, sampling bias. These are all very related phenomena. The idea is, what you can do is the network will pick up on a pattern that overfits the data so that it will not generalize to other contexts. Is that making sense? What the network did was it overfit to the sampling data. It fitted so well, but that actually masked the fact that it couldn’t generalize at all. Do you understand the problem, overfitting to the data? One of the things you have to do in neural network theory, people who do neural networks, is you have to come up with strategies for overcoming overfitting to the data. To do things. One strategy that’s quite successful is called dropout. It sounds counterintuitive, but what you do is you get your network, you have it learning for a while with all of its nodes, and then at some point you just randomly drop out half of the nodes. You really make the network, if you’ll allow me, stupider for a while. Then you let it go back to its processing. If you do dropout, and you’ve got to get it just right, but if you dropout and come back, and dropout and come back in the right way, what you dramatically do is reduce overfitting to the data. If you’ll allow me this language, and I’m trying to use it in quotation and scare quotes, it’s like the stupidity that you introduce with dropouts weakens the ability of the network to pick up on patterns, so that it doesn’t overfit to the data. That actually allows it to generalize. Obviously you don’t want complete dropout, because then the network will just go, and won’t pick up on anything. But if you get dropout just right, you really reduce overfitting to the data. So perhaps, now you understand what I’m not claiming to have established in the thing there, right? You get that, right? What I’m suggesting to you is what might be at work here is something like dropout. In fact, the plausible way I’ve seen what’s going on in an inside realm, bless you, is perhaps that initially your problem formulation is overfitting to the data. You can’t properly transfer, which is another way of saying you can’t properly generalize. So perhaps what moderate distraction is doing is allowing us to try and get that more fine-tuned dropout going. You can’t directly dropout. You can’t go, okay, shut off some of my neurons. But what you could do, right, is perhaps redirect attention in such a way that you get something like the dropout effect. What’s interesting about dropout is it works independent of the content of your problem. It’s purely procedural in nature. Okay, yes? I just wanted to clarify, how exactly do you do dropout to a machine? And what do you do to a machine? Let’s say it has like 50 neurons, 50 nodes, you just say, those 25 hidden units, by the way, 25 of those, just, they get shut off for a while. Okay, I just have to get them open. Okay, I’m not there anymore. By the way, this is also an example of why you don’t want the damn limitless pill to work. If you turned on all your brain, you would overfit to the data like, this is God. You would overfit to the data, like massively overfit to the data. I have a question regarding the node. Would each node be for a specific function or would they all have the same function? No, all they do is excite or inhibit other nodes. Okay, so you wouldn’t create a node for a specific function? No, not at all. In fact, that’s directly what you’re not doing. You don’t want the nodes to be like components in a program. Okay, so each node is a generalized kind of node? That’s right. Okay, so again, that is not a claim to have established anything. What I’m doing is raising the plausibility of how distraction could work by providing a potential mechanism, right, working for it. I’m trying to show you that we’re not in the gestaltist position of just pointing out a phenomena and not having a causal mechanism behind it, not having a theoretical way of bridging to a causal mechanism. That’s what I’m trying to do. Okay, so one thing we could do is to see if that’s the case. Many of you know there is kind of a related issue. We can’t talk about it very much in this course. They talk about it more in 3.15 than in 3.71. And it goes again back to what I said a minute ago in what I said in a previous class. People with higher general intelligence actually show lower cortical activation than higher cortical activation. And that has to do with ideas about the efficiency of the processing as opposed to sheer amount of processing. Yes? Is the reason why dropout for the neural networks is so effective is because it creates more of a small problem? I think so. That would be my interpretation. I think what you’re doing when you’re cycling is something like self-organizing criticality. And what you’re doing is actually implementing self-organizing criticality in the network. And we’re going to talk about what that is. Not everybody would agree with that. But I think it’s a very plausible interpretation of why it works. So yes, I agree with you on that. And that would be the way in which we start building this theoretical and empirical bridge I’m suggesting might be there. Because you could say, well, look, what you’re doing with the neural, what you’re doing with dropout is you’re doing something very much like self-organizing criticality. And we already have evidence over here that if you put self-organizing criticality into your networks, they gain these characteristics. And look, over here you have experimental evidence for self-organizing criticality within the behavior of inside problem solving. Wow, that’s probably not just a coincidence. That’s how I would start making that argument. Well, that’s how I have been making that argument. But more needs to be made. So please, if you’re noting any of this down here, I’m not presenting this as an established thing, right? I’m showing you the theoretical fecundity of the position we’re now bringing up, the lines of theoretical inquiry and empirical research it engenders. Yes? So it was overfitting because in addition to learning present interactions with no tanks, it also learned tank in the, like, it was like what, tank in the morning, no tank at night? Or did it just learn, it could also have just learned like morning light that was easier and then not known the tank, but then it still learned something that’s not overfitting, it’s just not? No, it means it overfits. What it does is it overfits, it tailors its programming to a pattern that only exists in the sample and doesn’t generalize to the population. What I’m saying is that would be like if it was tank and no… No, it’s only picking up on morning and afternoon. But then it’s not that it’s overfit, because I mean you’re saying it’s picking up like a pattern that exists there, but I mean like it’s not, it’s just learning the wrong thing, not that it’s like extending, because if you present two things together, you could learn it, so like if you could tell morning or night irrespective of whether or not there was a tank, but it did learn something that wasn’t just applicable by that data, if you were trying to teach it morning and night, it just was the wrong thing. Right, but the learning was being given to it because it looked like it was answering tank or no tank. So the correlation is a spurious one that the learning mechanism is picking up on, because the correlation only exists in the sample, it doesn’t exist in the real world. But I would hesitate to say that the network was learning the correlation, it’s more that that’s the way the learning, the back propagation was actually affecting the network. But so, the lesson of understanding, that would only make sense if it could, like the two were linked in the learning… They are linked. So if I mean like, if I’m saying to pick a test morning or night without any tanks, like it’s like, oh this is the morning, there’s no tank, this is the night, there’s no tank, then it’s not linking, the network isn’t linking it, it’s just ignoring the tanks, versus having them together, the results I guess, and the… I can understand that the results are not fitting correctly, but I’m saying from the actual network, it’s not about like, if it just learns the wrong thing, it’s not fundamentally different from learning a pattern that’s only present in one thing, but then it only learns the pattern that’s associated with it, or if it just learns the wrong pattern, but not necessarily… It learns a problem, yeah, it learns a pattern that will generalize, so that it can solve the problem that it’s designed to be solving. Okay, so it’s designed to be solving the tanks, but it’s solving a different problem, but it’s still effectively solving that other problem. Right, and that’s important, because overfitting to the data means you do solve a particular kind of problem, you solve a context specific problem. But like, what I’m saying is that you could solve other problems that were day and night related, but it’s just not… So what’s the issue here? I’m sorry, I’m not getting what hangs on this. Like, so then it feels, so maybe I’m… because you’re saying overfit, but you’re saying it’s not generalizing? Yeah, so if I say, how many people in this room thinks student tuition is too high? Put up your hands. Oh look, the majority of Canadians think that student tuition is too high. It’s true of this room, but it’s not true of the world. What I’m saying is it’s not generalizing the tank pattern, but it is generalizing the day and night pattern, so it is having successful generalization and patterning, it’s not… That’s trivial, because every pattern generalizes, you can find something that generalizes from your data set no matter what. No, but if it was… I don’t know that I’m describing this properly. I think that was a simple solution. Well I was thinking like, it’s overfitting in context to the tank, right? That was my… Yeah, she’s quite understanding that, like, it’s overfitting when you put it in context to the main purpose of detecting tanks. It’s not solving the particular problem. Yeah, but what I’m saying is it’s not overfitting if you take it away from it. Yeah, but it’s not… The problem is a network… you can always find some problem that your network is… Like, you can’t make that a criterion, because abstractly there is some problem in problem space that my network is providing the data for. Like, that can’t be your answer, because it’s just trivially true. So that’s what you’re doing, and I don’t… maybe you’re not, but that wouldn’t do much, is what I’m saying. I was just kind of in defense… if you show any actual warning now, it’ll successfully do that. It’s not just a trivial problem. It’s very interesting skill that it has to figure out. Which I know isn’t entirely… it’s not entirely relevant to the case, but it might be relevant to the terminology. No, no, but what you’re saying is… and this is a different point. You’re saying neural networks are really good at picking up patterns. And what it did is pick up this pattern, which is great. That’s whimsical. But I said that from the beginning. Yes, but the point was a terminological one, which was, hey, this network is not just picking up a pattern, it’s picking up a generalization. And so why are we calling the mistake that’s going on overfitting, given that it did successfully generalize? Because, yeah, but the point is it’s not predicting what it needs to predict. Right, but the problem isn’t one of lack of generalization. But we don’t care about the abstract ability to generalize in any way. Because everything, from any finite data set, there’s an infinite number of generalizations. Yeah, I don’t think you guys are actually very much disagreeing, so I tried to identify it in the terminology. Because that seems to be what actually caused this entire discussion. I think I agree with you. Okay, yeah, I was just trying to… So basically it’s become really, really good at differentiating its data, but that’s not particularly what we want it to do. We want it to learn the greater skills. What we want it to do is to be able to do what we need to be able to do, which is generalize from the sample to the data in a way that is relevant to the problem I have. Okay, but it has to… So it’s sort of like someone who’s learned the material really well but can’t apply it in a way. Yeah, I guess. Okay. That would be kind of only ready for the data. Okay. And that would, if I understand you, that would line up with my suggestion that perhaps some aspects of your past are overfitting to the data so that transfer cannot occur. Are we saying the same thing? Okay. Okay, there we go. So what I’m suggesting to you is there might be a mechanism, a non-computational mechanism. Dropout is a non-computational process. You’re just randomly shutting off nodes in a neural network, right? You’re not sort of logically manipulating abstract symbolic propositions according to inferential procedures or anything like that. There might be a non-computational process at work that’s going on in incubation. Again, I am not arguing that I’ve established it. I’m arguing that it’s plausible. And that means the proposal that we’re looking at a non-computational process is plausible. That’s all I’m trying to do. Okay? All right. Let’s keep going because we’re going to go back and forth, back and forth, back and forth. Okay, so now reason work by Dodd, Smith, and Ward around the same time. We’ve seen all these names so far except for Ward in this discussion. 2002. And they’re going to give a little bit more credence back to the opportunistic assimilation hypothesis. But then we’ll see more work that sort of takes it away. And so getting this all consistent is taking a lot of effort. Which means we have now an increasingly good explanation why the empirical work was so difficult to easily produce consistent results for incubation. It seems to be a very complex set of variables interacting in very complex ways. Okay, so what did they do? They directly tested the mental set transformation theory against the opportunistic assimilation. So they directly, that is explicitly and directly how they set up the experiment. They’re going to directly have the two compete, the mental set transformation theory against the opportunistic assimilation. People were given, and I’ve already made a criticism of this so I will remind you of it, but then we will go forward. People were given a rat task. Remember those? Remote associate task. Remember Black, you know, and was being associated with Ward and magic and death. Remember? And although that was, you remember the Schuller-Melcher, although that was predictive of insight ability, it was not the best predictor. Remember? None of you are acknowledging. Do you remember I made this point that although it’s predictive of insight, it’s not the best one, it just tends to get used because it’s easier. So I’m just saying remember that criticism because it’s relevant here. Okay. So what they did is they controlled and manipulated three variables of interest. They controlled and manipulated three variables of interest. So one was the presence or absence of clues during the interval. So for example, if the missing word is Black during the interval, the word Black would actually appear somewhere in the interval. The word is Black during the interval. The word Black would actually appear somewhere in the environment. Q type. Related or the answer? A related Q would be White because White is highly associated with? Black. Black. Right? Right. They tend to be highly associated, very related. So instead of Black being in the environment, White would be there as an explicit thing that you could take note of. Or Black would be there. Is that okay? Do you understand the second variable? Third was the presence or absence of instructions. When people were given instructions, participants were told there would be environmental clues and they should use them. So the hypothesis is that this induces mental set change. People have to do mental set because they have to now sensitize themselves. They have to look for missing information. They have to try to sensitize themselves to something they don’t currently have. Only for answers plus instruction. So first of all, what that means is, right? So take a look at each variable. There is something in the external environment, but it has to be the answer. It can’t be something related, even closely related. What hypothesis is that arguing against, by the way? Again, unconscious work, spreading activation, any associational process going on in the unconscious that’s being ruled out. Okay? Young and Freud again, he had really hard through all of this work, by the way. Okay? And then the presence or absence of the instruction. The answer only helps if a mental set transformation was previously induced in the people. I know this is kind of going to start, this is starting to sound like a Canadian answer. That you need both the mental set transformation and relevant material in the environment, right? But let’s talk a little bit more about it. So notice this, what the results show is it does rule out both unconscious work or spreading activation and a simple version of the opportunistic assimilation hypothesis. It rules out the passive version of the opportunistic assimilation hypothesis. You need more than to simply have impasse in order to assimilate the relevant information. You have to actually induce mental set transformation in order for the assimilation to occur. Subjects have to set themselves to the correct way and encounter directly relevant information for solving the problem. Now, how could we reconcile that with Siegel? I think this strongly suggests that it’s not a matter of producing new content, so much as altering the relevance of the available information. And we’ll see this supported in and sort of further work we’re going to take a look at. So it seems that what incubation involves is attention shifting and mental set transformation that alters problem formulation probably by altering patterns of relevance and salience. I’ll say that again. Incubation evolves probably, I mean, because this is a highly probabilistic convergence argument. It seems probable that incubation involves attention shifting and mental set transformation that alters problem formulation by altering patterns of relevance and salience. Sorry, by altering patterns of salience and relevance. Okay, now we’re trying to pick up on, again, trying to refine this, and to try and get clearer about the role of content and procedure, I want to take a look at a couple more experiments. Try to follow this through. And I think the next experiment, I think it’s an interesting and good experiment. But I also take as an example about how easy it is to confound these two issues. Okay, so, Sleep In, At Out, from 2010. This does get my sort of marks, like a really cool experiment. Sort of being really interesting. So they start from the idea that insight is associated, I’m going to give their argument and then I’m going to respond to it. They say that insight is culturally associated with a light bulb going on, an incandescent light bulb. So you show insight by showing a light bulb going on over somebody’s head. So you show insight by showing a light bulb going on over somebody’s head. And so the idea is, right, could you trigger insight through that cultural association by turning on an incandescent light bulb when somebody’s trying to solve an insight problem? So what they did, I’m going to come back to, I’m going to critically interact with this, but what they did was they had two, well they actually have two contexts and then they have variations on it each, but let’s just talk about the basic experimental setup. You have two contexts, you have people sitting at a table trying to solve insight problems, right? And what you do is you have a confederate, you guys remember what that means right? You mean anybody from the Civil War. You have a confederate come in and there’s two different situations. The confederate says, oh you probably need more light and they turn on an incandescent light bulb on the desk or they turn on a fluorescent light bulb over. And then you carefully control the experiment, which you can do because it’s a lab, so that there is no difference in the amount of illumination provided. You also do pre-tests to rule out that there might be affect or mood things that are at work because of the different kind of lighting. So you control for those variables, which they did. And then what they found was turning on the incandescent light bulb significantly increased the number of insight problems that were solved, whereas turning on the fluorescent light bulb had no similar effect. Yes? Turn the light bulb on the desk. Yeah, because it was like on a lamp. And you may be thinking, and I think you’d be right that there’s a potential confound there and we’ll talk about it. But first of all, does everybody understand the experiment? Because this is freaky incubation, right? This is really freaky incubation. No resting, the freudy unconscious and young aren’t in here at all anymore. You turn on this or you turn on that and boom, this one works and that one doesn’t. That’s freaky incubation, right? Okay, I’m glad you agree with me about that. Yes? So incandescent light bulb on the desk or fluorescent somewhere else? Above. That’s because generally at the time, I mean we have LED lights and they’re taking over, it’s very hard to get sort of fluorescent lamps, desk lamps. Now the reason they chose the fluorescent is they argued that the cultural association isn’t there. So they were arguing that there were sort of conceptual processes, cultural conceptual processes driving insects. But let’s first of all take a step back and sort of rework our way through this and we’ll also notice some important variations in the experiment that tend to undermine that interpretation of the data. First of all, they don’t consider why that metaphor caught on culturally. Why? They don’t consider the fact that there are analogous metaphors in other cultures in which the electric light bulb was not invented. Flashes of light, enlightenment metaphors are cross-cultural. This point to the fact that it’s probably not specific to incandescent light bulbs, right? That instead the metaphor is based on a more pervasive association between a flash of light and insight. And of course we’ve already noted that that flash seems to be associated with an acceleration of feeling of warmth, some change in humidity, some change in temperature, some change in humidity. And probably a dramatic alteration of salience. So something very analogous to what happens to you when a literal flash of light is presented to you is occurring in insight. We already have lots of evidence that this flash is completely independent of whether or not an incandescent light bulb is present. In fact, it sounds to me like they’re getting cause and effect the wrong way around. The metaphor was chosen to describe the effect rather than being something that drives it. First thing you should notice is, right, if it’s the metaphor that’s at work, then the metaphor has to be in place. Mainly the metaphor involves an incandescent light bulb. Interestingly enough, the metaphor has the incandescent light bulb going off above your head. Which is not where it is in the experiment, it’s on the desk. And here’s something freaky. A shaded incandescent lamp, in which you couldn’t even see the light bulb, was almost as effective as a visible incandescent light bulb. Yes? Just to clarify, do they have the moment of insight right? Like maybe a few seconds after the light is turned on or? No, it’s the timing. It’s fairly close. The facilitation I believe was both in terms of number and speed of solving. You just go back and review it. Yes? So if you can’t tell if an incandescent light bulb, isn’t that the same as, is it different from fluorescent then because you can tell it’s fluorescent and you don’t know if the incandescent light bulb is incandescent? I don’t think it has to do with that at all. I think it has to do with the ones on the desk and the ones above your head. Okay. And I’m going to tell you why I think that’s the case. And then I’m going to show you experiments completely independent of this that line up with that interpretation. Yes, Sarah? So why do they, if you’re saying they didn’t put a fluorescent lamp because it was really insistent, why didn’t they just make like a hanging lamp that had incandescent? They should have. It’s just like, then you have. Yeah. So we’ve been talking a lot about procedural similarity, right? I’ve tried to give you, we have independent evidence for something very procedurally similar going on in your attention, your arousal, your metabolism and salience for insight. That’s very procedurally similar to what goes on if you have something in your field of vision go through a sudden change in illumination. Why isn’t it just the, and that lines up better, right? That’s why the covered lamp is almost as effective as the uncovered lamp. Because all you really need is the sudden change of illumination within your focal attention. Because that’s procedurally similar enough to what happens in insight to help put you into the procedural frame of mind necessary for enhanced solving in insight. So instead of this being something that is inconsistent with what we’re seeing about incubation, it’s actually lining up with it very well. Let me try and show you another experiment that adds convergent evidence for the argument I’ve just made. So Thomas and, so in English you can’t have two L’s at the beginning of a name or a word. So how is this pronounced? Jeras. Wow, do that again? Jeras. So I don’t have that phoneme. It’s like, yeah, it’s like Jeras. Is it kind of like a show with a bit of an L at the end? It doesn’t have an, it is, well, in Spanish it’s just called double L or a J and it’s Jeras. Okay, so what she said, I can’t do it. But Lloyd is also double L. Pardon me? At the beginning. It is? I hadn’t thought of that. But that’s not the right sound for that, I probably. It could be Leras. Pardon me? It could be Leras, depending on where he’s from. Yeah, okay, you’re right. I never thought of that. Okay, so you’re right, I’m wrong. I’m wrong on both. I can’t make the sound if it’s Spanish. I was wrong about it not being present in English. So I was just completely wrong all around. But, there we go. Okay, so now back to one of our favorite problems, the Dunker radiation problem. The Dunker radiation problem. Okay, you guys all remember the problem, right? So they had done earlier work in 2007, in which they found that covert eye movement patterns could facilitate solving the Dunker radiation problem. Okay, so how do you do that? What you do is you have a bunch of people trying to solve the Dunker radiation problem, and you’re doing eye tracking on them. You’re doing eye tracking on them. And then what you do is you give a second group of people the Dunker radiation problem, and you say while you’re trying to solve the problem, and think about this being a cognitive load, so it should actually be great performance a bit, right? While you’re trying to solve the problem, I’m going to put an X on the screen, and you have to follow the X around the problem. Do you understand? Now, the second group of people, they’re going to follow the X around the problem. Around the problem. Right? Do you understand? Now, the thing is, what you do is you of course have two groups, you’re just moving the X around, that’s your control group, and you always can tell if you’re in the control group in an experiment because it’s protestantly boring. So when I was a bad undergraduate in that way, whenever that happened to me, I would do my best to become an outlier. I would figure out, oh, I’m in the control group. What are they controlling for? Okay, I’m going to do really bizarre behavior. Which is again why it’s harder to do psychology than physics. Okay, the experimental group, what you’re doing is when they’re tracing the X, they’re repeating the eye movement patterns of the people who solved the Dunker radiation problem spontaneously. Now, isn’t this like completely procedural? And guess what they found? You get significant facilitation on the Dunker radiation problem if people are being made to reproduce the eye movement patterns of people who solved the Dunker radiation problem. That’s since 2007. And then they followed it up because there’s a potential compound here. Right? They wanted to really get clear that this was an attentional thing. Rather than maybe a physiological thing, right? About moving your eyes around. Whatever. So what they did was they did the experiment again, but they did what’s called covert attentional shifting. Covert means you have to keep your perceptual gaze fixed. Like you can keep your perception fixed, and yet you can direct your attention, for example, you can direct your attention to your peripheral awareness even though your perception is fixed. And you of course can track that to make sure that people are staying perceptually fixed. Like on some specific stimulus. You have to keep looking at the red X, but let your attention trace and follow the black circle as it moves around the problem. As you try to solve the Dunker radiation problem. This is so open, like this is so much common to both, but guess what they found? It works. It facilitates solving the Dunker radiation problem. That’s right. This is like a almost purely procedural cube is working. People have no idea why it’s working. One thing you might be thinking, you might be thinking there is at least some content similarity in the idea that what people are doing is they’re sort of tracing out the lines with their attention or with their eye movements. And so they’re sort of diagramming, right, with their attention the solution. And they specifically were able to rule that out. They were specifically able to rule that out. Pardon me? What were they able to rule out? The fact that they were tracing. The way people are moving the attention around in no way looks like the solution to the Dunker radiation problem. They’re not drawing the solution with their attention. Okay. Okay. This is a quote. The alternating relationship between this central tumor in the diagram and multiple outer locations. As long as a participant shifts attentional focus in an alternating manner from the inner tumor to separate outer points, there’s no need for actually in and out movements that enact the path of the lasers, nor does there seem to be a need for internally representing such movements. So there is nothing in the attentional movement that looks like the solution to the Dunker radiation problem. And people have no internal representation of, right, the internal representation of the movement. The representation of, right, coming from the eye movement. This is pretty much purely procedural facilitation. Okay, so there’s, it’s a more abstract relationship. Or another way of putting it, this is not based on sort of visual similarity. This is more abstract attentional in nature. And it has to do, again, with this sort of right kind of destruction and moving and shifting of attention that is in some way facilitating, right, restructuring so that insight occurs. Perhaps by something like dropout, etc., etc. Yes. So people who solved it spontaneously have their eye movements tracking what their attention was making salient. Right. And so they get backwards and have the people who weren’t solving it spontaneously following their eye movements of the other people to shift their attention to make certain things stand out. That’s right. And so you’re running two experiments together. The first was the eye shift and then the second one is they made it just attentional. Okay, keep going. So then you’re paying attention to what’s the manifestation of people’s focus. And so then your focus changes to align with theirs in the way that would cause or facilitate the solution. Right. This complex, multiple simultaneous transformation of salience is the process that’s actually facilitating the insight. Yeah, cool. Cool is the right answer. So I think a more consistent answer between this experiment and the sleepian one is the one I would suggest for how we interpret sleepia. Okay. So we’re getting to sort of a very interesting place about, right, attentional redirection, Okay. So we’re getting to sort of a very interesting place about, right, attentional redirection and the manipulation of salience in a purely procedural manner that is facilitating insight. And along the way I’ve been showing you more and more how, unlike the gestaltus, we might be able to come up with plausible mechanisms for how this occurs and therefore a genuine alternative to the search inference framework is becoming available. Did you have a question? Yeah, just clarification. So there was an experiment run in 2007 and that was the one that wasn’t covert eye movement and then it was one run in 2007. Not covert attention. Oh, sorry. It’s covert eye movement because the people are being made to covertly move their eyes. Right, but I’m just saying that one was one run in 2007. 2007. 2009 is the covert attention. 2007 is the covert eyes. Right, okay, okay, got it. Yes? Another annoying clarification question. For the path that they were having people craft with their eyes, was that one person’s particular path or was it some kind of average of a bunch of people? I think it was the second. An average? Yeah. And how much correlation was there between different people’s eye movements who were able to solve the problem? Is there some kind of pattern there? I don’t know. And that goes towards something that we were talking about earlier. This might be a diagnostic tool for individual differences. And again, if personality, for example, interacts with this cognitive machinery at sort of a very basic level, this could give us very objective ways of testing for certain things that would be significantly more reliable if, this is important if, than the kinds of testing we do now. So, yeah, yeah. But that’s what we want, right? We’re trying to get at the guts. See, I promised you that if we went down this path, we might get at the machinery that gets at the guts of thinking. And it looks like we’re getting there. Back then, Weisberg. 1995. Weisberg. Bom, bom, ba-dum. Okay. 1995. Okay, by 95, he admits that insight exists. Okay. Instead, he shifts and he makes two really, he does, he goes back and makes a philosophical, theoretical critique. What I’m going to do is just lay out that critique and then we will end. Because this is all the material that I’m going to expect you to have processed for the best. Okay? It’s a lot. Come on. It’s a lot? Well, this, I mean, that’s why you’re here, right? You want a lot of like highly integrated, potentially, you know, insight producing information. That’s what an education is, right? You don’t think so? You want, what else? So what’s an education? That’s what it should be. Okay. Oh, so you’re saying that doesn’t often happen under the name of education. I’ll take that as a compliment. Okay. So, and you guys are smart enough. You both made throughout, sorry, both of these points have been made, sorry, I should have turned my sound off. I apologize. Both of these points that Weisberg made have arisen in this class. Both of them have been made. So the first point that Weisberg made, and again, he’s making it philosophically, theoretically, is that we’re going to see what the empirical response is, because science is based on theoretical debate driving experimental competition, and experimental competition driving theoretical debate, right? The first point he made was, he said, there’s an assumption going through all this literature, and again, this point has already been raised in class, that something is intrinsically or inherently an insight problem. Something just is an insight problem. People are saying these are insight problems, and these are not insight problems. And if you’ll allow me to solve for a punch, that’s very problematic. Because, so bear with me, because I mean, this is somewhat metaphysical. There are no problems in the universe. The universe just is. No problems. Because we gave up a long time ago, pretty much with Galileo and the discovery of inertial motion in the middle of the 16th century, the idea that the universe acts on purpose. The universe does not act on purpose, and therefore, by definition, it cannot face problems, because it is not seeking goals. And if you’re not seeking goals, there’s no way you can have a problem. There’s no problems in the physics of things. So no set of physical objects is intrinsically a problem or not. This is actually an important existential point to remind yourself every now and then. I mean, this was one of the crucial claims of Stoicism. And if you’re interested in Stoicism and how it has impacted directly on clinical psychology and on studies of rationality, there’s a course you can take where we talk about it. It’s called 371. But this is one of the central claims of Stoicism. We forget that there are no problems in the world. Second point, whether or not something is an insight problem… So first of all, problems don’t just objectively exist on their own. Secondly, whether or not something is approached… I’m going to hide a discussion we’re going to have in the future under this vague term, approached. Whether something is approached or appropriated as an insight problem or not depends on the knowledge and expertise of the person involved. Whether or not something is approached or appropriated as an insight problem or not depends on the expertise and the knowledge of the person involved. Something that for me might require insight might not require insight from you. So just to give you an obvious example, you know what is no longer an insight problem for me because I’ve been doing this for a very long time? The 9-dot problem. I have no insight experience whatsoever when that problem is brought to me. You’re going to put it on my tombstone. I came into one class one time and there was a box of Rice Krispies on the table. And I thought, this is a weird bribe. I don’t even really like Rice Krispies. And then I turned around the box and on the back of the box there was a 9-dot problem. That’s why the student had brought it in for me. It’s not an insight problem for me at all. And that’s reportedly one of the hardest insight problems. And I’m not claiming any special ability. The point of claiming is familiarity and experience and expertise. None of you find this to be an insight problem anymore. Take a four-year-old. They can count. Here’s five candies. They’re all the same shape and color. Here’s five candies. What row do you want? What do they pick? Second one. Because they centrate. They find one variable super salient. Remember, we talked about this. And at some point they have the insight. In fact, there’s some talk, some work that, for example, one of my students, Jim Sunde is doing about trying to interconnect the machinery of insight and the machinery when you move from one Piagetian stage to another. But are any of you having an insight experience right now? Initially, I was going to go for the bottom row. I was stuck there. And then, oh, right. No, right? There’s no difference. Yes? It’s not an insight problem anymore. So there’s strong physiological response to solving an insight problem. That’s right. And presumably not when you already know the problem. Or even when you’re solving an incremental problem in which you don’t know the answer. Yeah. Has anybody studied whether in between, so if you used to know the problem, but you’ve sort of forgotten it, and then you find it in your head, you go, aha. But it’s not quite that feeling of, look, I’m a genius. It’s just an, aw, I wish I didn’t know the answer, because then I could have figured it out. Well, that’s the feeling I get. So is there a physiological associate there? I think there is. I don’t know if anybody’s ever studied exactly the phenomena you described. I mean, there’s some good songs about that. Bob Cedar said, I wish I didn’t know now what I didn’t know then. In the same as songs. And you’ve all had that experience, eh? I wish I didn’t know now what I didn’t know then. Okay. But I don’t think anybody studied it that specifically. It would be interesting. It would be interesting. And it’s interesting to know, I would like to broaden your question. I would like there to be more specific and dedicated work trying to get at, what is it that the phenomenology of the insight, the insight experience, is specifically tracking? We had some initial good work with feeling of warmth in the metabolic cell, but that sort of faded into the background as sort of these other kinds of experiments took prominence. I’m really intrigued by the fact that there’s something like a flash that goes on in the insight experience, and that seems to be somewhat similar to altering your state of consciousness. Okay. So the first point is, nothing is objectively an insight problem. And it looks like people have been assuming that things are objectively insight problems or not insight problems. And if that distinction isn’t objective, then the experiments that have been run on the basis of that distinction could be called into question. So you see what Weisberg is doing, eh? He’s doing to other people’s work what Dolmanovsky and others did to him. And he’s showing, I can play the theoretical philosophy game too. He’s good at it. He’s very good. Very good. I mean, this is a, and that’s why when this point was brought up, I think it was initially brought up like, oh, I acknowledge it, and said, yeah, this is a great point. We’re going to come back to it. Okay. What’s the second point? He said, even if we could somehow manage the distinction, there’s a second problematic assumption in the literature. And this point was also brought up. I’m sorry, what point? Did I just kind of just throw it over my head? I haven’t said it yet. Sorry, Steve, did you ask me to say what the second point is? No, yeah. What you just, basically repeat what you just said. So the second point is even if we had some way of managing the relationship between being an insight problem or not, there is a second faulty assumption. The assumption is that we have an exhaustive dichotomy. Things are either insight problems or non-insight problems. What’s the independent justification for that idea? Isn’t it possible that many problems are hybrid problems? Many problems are a combination of insight and incremental problem solving. Isn’t that at least possible? Yes. If we can somehow manage the distinction, is that to say an objective distinction? No, it’s if we can… Is there any way of… If we find a way of… You can’t just sort of throw away the fact that there’s all of the… Well, perhaps we would have to. But is there a way of re-construing the distinction such that we don’t have to invalidate all of the previous experimental work? Okay. Would that include an account of identifying when relative to a person something is an insight problem? Something like that. Okay. Yes. We’re going to talk about how you answer both of these assumptions. The thing is, let’s say, but Weisberg says, even if you somehow can answer the problem of your first assumption, there’s a second assumption that’s still problematic, which is the assumption that you have two exhaustive and internally homogeneous classes. There’s all these insight problems, all these non-insight problems, the classes are somehow homogeneous, and there aren’t any mixed problems that are hybrid problems. And of course, again, if it is the case that it’s not the exhaustive distinction that people are assuming, then a lot of our experimental results are now confounded. This is again why simply gathering the data isn’t good enough. So, what we have is we have two significant challenges. After the test, we’re going to come back and look at whole new lines of investigation that are going to actually end up converging with what we’ve been looking at. That’s going to take work. But what’s interesting is now we’re going to be involving neuroscience in some of these lines of investigation. And these lines of investigation are explicitly designed to respond to Weisberg’s 1995 criticisms. We’ll then see that Weisberg will return with FLEC in 2004 with one more set of significant criticisms. And then we’ll take a look at those responses as well. Yes, Chloe? If you redefine insights and come up with, you solved assumption one, you solved assumption two, if we use this definition of insight, isn’t that also not the same insight you’ve been using before? So, is it that all sorts of things that may be not overlapping, so you can’t even use the… Maybe not. Is it automatically followed? Is that a direct implication? Well, if there’s two… If you come up with one and it’s different, you can’t automatically assume that it’s applicable, so you have to justify that you could do it. Or it could be that the answers between the two actually turn out to reinforce a conclusion that you can accept all the previous experiments. And that’s what I’m going to argue actually happens. Pardon me? You know, the sea weaners, the sea… Oh, you were interested in the final scorecard? So, I think there’s a very powerful response to both, there’s powerful responses to both of the assumptions. And that’s because, and what that basically does is allow people to sort of move forward. The 2004 thing, I think, brings up another issue which actually requires us to go back and do something I promised to do, although I’ve been foreshadowing it along the way, show you why the construct of verbal overshadowing is a mongrel. So, does he win or lose? He advances science significantly, so he wins. Does he lose in the sense that he isn’t able to eradicate insight? I don’t think he wants to do that anymore. He’s interesting. As this part of the career sort of took this, he started writing books on how creativity didn’t exist. So, that seems to be what he does. And I’m a lot more sympathetic about him about that. We’re going to talk about that later in the course. I don’t, I’m not that convinced there is such a thing, other than in a trivial sense of making stuff happen in your environment. There is such a thing as creativity in any studyable way. Or, here’s a more, maybe here’s a more precise way of putting it, which we’ll investigate. When I say the word creativity, and I’m being rigorous in a way that we can scientifically investigate, am I saying anything other than insight? Am I? Yes. What am I saying? Oh, you were so easy with the first answer, but when I ask you what it is beyond insight, you go, brrrr brrr. Because in insight, the thing that people can solve inside in the end is because they are inherently competent of solving the problem. So, they are not creating something new, they’re just finding something that’s already there they didn’t find before. So, when someone is creative, they weren’t inherently capable of creating it? You’re making something, like it’s like a combination of old stuff, you get something that’s never existed before. Whereas the solution to insight… Really? Complete? I mean that sounds like romantic propaganda to me. Never existed before? Give me an example of somebody making something that never existed before. Light bulbs. Well, actually that’s not true, right? We already had flammable lamps, and then we went to an incandescent electric lamp. So, you can say flammable lamps. Pardon me? When you make something new, it’s not the same as finding… I think what she’s saying is when you have an insight problem, it’s a problem, so there’s a solution. Creativity is not like, oh, you had a creative problem, so you came up with the solution. So, you don’t think that when artists are trying to do something creative, they see that they’re trying to solve problems? They’re trying to alter the salience of the phenomena? But the problem is not defined. I’m not thinking of doing this, oh yes, I have a problem and there’s like one way to do it, because there’s multiple things that I can arrive at that would be equal in something that you’d be creating. Like if someone’s like, I’m going to draw a face, they could draw multiple faces, they could do it multiple ways, and none is inherently better than the other. I think if you look at the definition of creativity based on culturally how we perceive it and really what it means, those are not necessarily the same thing. So, picking one and then to use in terms of contrasting it to something else, creativity is not something that… So, if it’s not solving problems, how do you distinguish it from just generating random crap? Let’s do the thing is… No, no, I listen to you, answer my question. If you’re not solving problems, how do you distinguish creativity from just generating random crap? Yes, and what I’m saying is that culturally we have a distinction, but you could argue that there isn’t a distinction and that generating random things is being creative. Then that’s not an important phenomenon to study anymore, right? But like… Because anything could generate random crap. So, if I want a psychological phenomenon, I want the interesting notion of creativity, and then you seem to admit that that collapses back into insight, because unless I’m solving problems, there’s no way of distinguishing creativity from just generating random crap. Because if you don’t solve, if you don’t give people new ways of thinking about things, alternative ways of reformulating issues in their own life, they don’t care about what you’ve done. I don’t totally believe what I’m saying, by the way. What I’m trying to do is provoke you to realize that Weisberg has a point. And how, as we get into this, again, talking about, if you’ll allow me to stretch a term, the metaphysics of insight, we start to get into another topic we’re going to take a look at in this course, which is namely creativity. But that’s after the test. So go home and just worry about insight right now. Okay, so I’ll see you guys next week for the test.