1 00:00:03,170 --> 00:00:12,210 Final. Okay. 2 00:00:22,050 --> 00:00:25,410 So now we come to humans argument concerning induction. 3 00:00:26,750 --> 00:00:33,620 And you will find on the reverse of your handout a diagram. 4 00:00:37,620 --> 00:00:41,130 Which outlines the argument of treaties. One, three, six. 5 00:00:42,360 --> 00:00:45,390 Now, I'm not actually going to be going through this in detail. 6 00:00:46,450 --> 00:00:54,760 But I've put it there because I think when you read through one, three, six, it will be helpful to understand how the argument is structured. 7 00:00:55,540 --> 00:01:06,520 As you can see, it's quite a complex argument. I'm understanding how this relates to discussions of human induction can sometimes be difficult. 8 00:01:07,960 --> 00:01:14,660 We'll see why. Well, there are three versions of this famous argument. 9 00:01:15,440 --> 00:01:18,410 The first is in treaties one, three, six. 10 00:01:19,550 --> 00:01:26,450 But it seems to me, at any rate, that HUME in the treaties does not really quite appreciate the significance of this argument. 11 00:01:27,340 --> 00:01:36,250 He slots it in here, as we've seen as a detour. It's fitting into a structure whose aim is to find the origin of the idea of cause. 12 00:01:37,120 --> 00:01:42,130 He's clearly very interested in probable inference, probable belief. 13 00:01:43,590 --> 00:01:48,920 But he doesn't seem to realise what a jewel he's hit on with this argument. 14 00:01:51,090 --> 00:01:58,920 Now in the abstract, which was probably written only eight months or so later, it wasn't published till 1740. 15 00:02:00,220 --> 00:02:03,490 But it was probably written around September, October 1739. 16 00:02:04,720 --> 00:02:09,670 We got a striking change. The abstract as a whole is very condensed. 17 00:02:11,100 --> 00:02:19,830 But the argument concerning induction in the abstract is something like three quarters, as long as the argument concerning induction in the treaties. 18 00:02:20,640 --> 00:02:25,469 And that's rather surprising when you take a large work like the first two 19 00:02:25,470 --> 00:02:29,550 books of the treaties and you shrink it to something as small as the abstract. 20 00:02:29,910 --> 00:02:35,250 You would expect that the argument concerning induction would reduce to just a paragraph or two. 21 00:02:35,760 --> 00:02:43,750 Very far from it, it occupies centre stage. And when we get to the inquiry published in 1748, 22 00:02:44,470 --> 00:02:51,250 the the argument concerning induction is yet more full and occupies section four 23 00:02:52,180 --> 00:02:57,730 and one can see most of the rest of the inquiry or significant chunk of it anyway, 24 00:02:58,120 --> 00:03:01,720 as spelling out the consequences of Hume's theory of induction. 25 00:03:03,970 --> 00:03:07,900 So I think in the case of induction, if we want to understand what HUME is about, 26 00:03:08,470 --> 00:03:13,090 we do need to pay careful attention to the inquiry as well as the treaties. 27 00:03:16,250 --> 00:03:21,920 So much of what I say here will be derived from the inquiry argument rather than the treatise argument. 28 00:03:21,920 --> 00:03:29,990 But I will be saying things about how they link. So here's a very brief overview of the argument. 29 00:03:31,280 --> 00:03:38,450 Suppose we say we see A followed by B. Again and again and again. 30 00:03:39,700 --> 00:03:44,440 Next time we see an A, what happens? 31 00:03:45,410 --> 00:03:51,650 We expect to be to follow. Why do we do that? Well, is it a priori insight? 32 00:03:52,550 --> 00:03:58,670 No, because a priori we can't know anything about the causal effects that they will have. 33 00:04:01,290 --> 00:04:07,350 It must be based then on extrapolating into the future the associations that we've observed. 34 00:04:08,750 --> 00:04:15,680 Since we can't a priori I know the table calls B it must be our experience of being followed by B, 35 00:04:15,920 --> 00:04:22,850 which leads us to make this inference, and that involves extrapolation from past to future, from observe to unobserved. 36 00:04:24,650 --> 00:04:30,650 Okay, so what grounds can we give for extrapolating from observed to unobserved? 37 00:04:32,170 --> 00:04:40,130 Is it based on sensation? No. When we perceive objects, that doesn't tell us anything about their causes or effects. 38 00:04:40,150 --> 00:04:45,470 Only experience does. What about logical intuition? 39 00:04:46,510 --> 00:04:50,470 Is it self-evident that what's happened in the past will happen in the future? 40 00:04:51,430 --> 00:04:55,120 No. We can easily imagine it not happening in the future. 41 00:04:56,320 --> 00:04:59,530 That also shows it can't be demonstratively proved. 42 00:05:00,190 --> 00:05:05,320 We can't give a demonstrative argument that what's happened in the past will continue in the future. 43 00:05:06,320 --> 00:05:09,680 Because if we could, it wouldn't be conceivable that anything else would happen. 44 00:05:09,890 --> 00:05:14,210 But we can easily conceive of different things happening, see one billiard ball moving towards another. 45 00:05:14,780 --> 00:05:22,490 It's very easy to conceive that the second billiard ball won't move or that it'll jump into the air or disappear or all sorts of things. 46 00:05:23,800 --> 00:05:27,160 So the can't be a demonstrative argument to prove. 47 00:05:28,260 --> 00:05:32,070 The next billiard ball will behave in the same way as all the billiard bulls before. 48 00:05:33,930 --> 00:05:41,430 What about probable reasoning, moral reasoning, reasoning concerning matter of fact, what we now call inductive reasoning? 49 00:05:42,750 --> 00:05:53,750 Well. This whole argument has been for the conclusion that such reasoning depends on the assumption that the future will resemble the past. 50 00:05:54,960 --> 00:05:59,340 So clearly we cannot use probable reasoning to draw that inference. 51 00:06:04,570 --> 00:06:12,010 So what I've given there is an outline of the argument as it appears in the inquiry. 52 00:06:14,430 --> 00:06:23,430 Now in the inquiry. HUME does seem quite explicitly to rule out intuition as a source of the principle of extrapolation. 53 00:06:24,660 --> 00:06:34,380 He also discusses at some length whether a sensation can ground such an extrapolation sentence as it can't in the treaties. 54 00:06:34,400 --> 00:06:42,840 You don't get that. In the treaties, as soon as he's identified this move of extrapolation, what's commonly called his uniformity principle. 55 00:06:43,530 --> 00:06:50,940 He immediately goes on to say, Well, can this be founded on demonstration or probability, knowledge or probability? 56 00:06:52,000 --> 00:06:56,290 And he doesn't consider any other possible sources of foundation. 57 00:06:59,290 --> 00:07:07,870 Now, obviously, you could claim or think that that suggests a fundamental change of perspective between the treaties and the inquiry. 58 00:07:07,960 --> 00:07:12,100 I don't think that. I just think the inquiry argument is more complete. 59 00:07:12,760 --> 00:07:16,030 I think I think in the treaties, when HUME came to the point of saying, 60 00:07:16,570 --> 00:07:22,120 how can we justify this principle of extrapolation from observed to unobserved? 61 00:07:22,570 --> 00:07:26,950 I think he just took for granted that it couldn't be intuitively justified, that it wasn't self-evident. 62 00:07:27,580 --> 00:07:35,530 I took I think he just took for granted that sensation wasn't going to ground it and he just went straight for the two types of argument, 63 00:07:35,830 --> 00:07:43,390 demonstrative and probable. Whereas in the inquiry he dealt with those four different sources. 64 00:07:47,910 --> 00:07:57,840 Now what I'm going to do now is look at the logic of the argument, trying to sort of analyse it in terms of its bare logical bones. 65 00:08:00,150 --> 00:08:12,300 So my claim is that the argument can be represented using pretty much a single relation, the founded on relation with these various relator. 66 00:08:13,650 --> 00:08:20,640 So we've got probable inference, causal reasoning, experience, uniformity principle. 67 00:08:21,180 --> 00:08:24,870 That's the principle of extrapolation from past of future. 68 00:08:25,780 --> 00:08:30,220 We've got reason, demonstration, intuition and sensation. 69 00:08:33,710 --> 00:08:41,480 So here we have a formal representation of the argument, as I've just described it. 70 00:08:43,170 --> 00:08:52,110 So we have at the top there Hume's claim, the probable reasoning that is reasoning from the observed to the unobserved all depends on causation. 71 00:08:53,140 --> 00:09:02,480 And we've seen how he argues that. Below that, we've got the claim that all knowledge of causation arises from experience. 72 00:09:03,630 --> 00:09:11,010 And he draws the conclusion that all probable inference, all inference to anything unobserved has to depend on experience. 73 00:09:15,550 --> 00:09:23,960 Okay. We then have that any inference from experience has to be based on a principle of extrapolation. 74 00:09:24,350 --> 00:09:30,740 What I call the uniformity principle, the principle that the future will be uniform with the past. 75 00:09:33,020 --> 00:09:42,709 So again, since probable inference is is founded on experience, since we can only infer anything from experience by founding, 76 00:09:42,710 --> 00:09:49,220 that's on the uniformity principle, it follows that any probable inference has to depend on the uniformity principle. 77 00:09:53,120 --> 00:10:01,350 Okay. Now we get to the heart of the argument. We have, but any probable inference is founded on the uniformity principle. 78 00:10:02,410 --> 00:10:08,710 It follows then that you cannot further uniformity principle by probable reasoning 79 00:10:09,250 --> 00:10:16,100 if inference from observed to unobserved depends on the uniformity principle. 80 00:10:16,900 --> 00:10:23,530 It follows that you cannot use such inference to ground the belief in the uniformity principle. 81 00:10:25,500 --> 00:10:29,430 But the uniformity principle cannot be founded on demonstrative arguments either. 82 00:10:30,790 --> 00:10:34,900 Nor can it be founded on intuition, nor can it be founded on sensation. 83 00:10:35,980 --> 00:10:39,580 Therefore, HUME concludes, it cannot be founded on reason. 84 00:10:41,350 --> 00:10:46,990 And if. The all probable inference depends on the uniformity principle. 85 00:10:48,290 --> 00:10:55,970 And the uniformity principle cannot be founded on reason. It follows that probable inference cannot be founded on reason. 86 00:10:57,290 --> 00:11:03,590 Okay, so that's the structure of the argument as it appears in the inquiry. 87 00:11:05,030 --> 00:11:08,060 My suggestion is that in the treaties these two bits are missing. 88 00:11:09,600 --> 00:11:14,730 But other than that, the logic of the argument, I think is very similar. 89 00:11:15,790 --> 00:11:19,470 Now, this isn't very apparent when you look at the text of the treaties. 90 00:11:20,720 --> 00:11:25,790 The reason it's not apparent when you look at the text of the treaties is that HUME doesn't 91 00:11:25,790 --> 00:11:30,080 simplify things at the beginning of the argument in the same way as he does in the treaties. 92 00:11:30,920 --> 00:11:36,050 What happens is he starts off talking about the inference from the impression to the idea. 93 00:11:36,200 --> 00:11:39,950 We've seen how that fits into the pattern of the treaties that he's taught. 94 00:11:40,040 --> 00:11:43,520 He started talking about the impressions of the senses and memory. 95 00:11:43,940 --> 00:11:52,370 He's now moving on to talk about the inference from the impressions of the senses or memory to an idea of something as yet unobserved. 96 00:11:52,520 --> 00:11:59,870 The inference from the impression to the idea and at the beginning of one, three, six, that's his focus. 97 00:12:01,420 --> 00:12:07,750 Then he says, Well, any such inference is going to have to depend on this supposition of uniformity, 98 00:12:08,380 --> 00:12:15,300 but the course of nature continues, always uniformly the same. And the things we have not observed must resemble those we have observed. 99 00:12:16,300 --> 00:12:19,300 And then he asks, What is the basis for that? 100 00:12:20,810 --> 00:12:25,700 And it's only then that he even mentions probability. 101 00:12:26,630 --> 00:12:32,930 He says, well, let's look at demonstration probability, see whether either of those conform to this principle. 102 00:12:34,100 --> 00:12:38,840 Now because the beginning of the argument was all about the inference from the impression to the idea, 103 00:12:39,590 --> 00:12:42,050 and he's only just mentioned probable reasoning. 104 00:12:43,130 --> 00:12:48,560 He now has to run through the argument showing that probable reasoning depends on the uniformity principle. 105 00:12:49,310 --> 00:12:53,390 He's already said that the inference from the impression to the idea depends on the uniformity principle. 106 00:12:53,510 --> 00:12:56,960 But he has to repeat that argument when talking about probable reasoning. 107 00:12:57,900 --> 00:13:02,880 So we get an extra epicycles in the argument. It's that bit more complicated. 108 00:13:03,810 --> 00:13:12,960 And on the diagram, you'll see. At the bottom of stage three there what I have called stages i, j and K. 109 00:13:14,110 --> 00:13:17,830 Those add that extra complication. Whereas in the inquiry. 110 00:13:18,520 --> 00:13:23,800 HUME Right from the start, he's talking about probable reasoning inference from observed to unobserved. 111 00:13:25,160 --> 00:13:31,160 And therefore, he when he comes to talk about the basis of the uniformity principle, he can very, 112 00:13:31,160 --> 00:13:36,440 very smoothly move from saying the probable inference is founded on the uniformity principle. 113 00:13:36,590 --> 00:13:44,600 Therefore, trying to find the found the uniformity principle on probable reasoning would be evidently going in a circle and taking that for granted, 114 00:13:44,600 --> 00:13:53,070 which is the very point in question. So I'm suggesting that the argument in the treaties is in this respect more complicated. 115 00:13:53,980 --> 00:13:57,520 And. In this respect, it's less complete. 116 00:13:58,870 --> 00:14:02,199 But I think essentially it's very similar. 117 00:14:02,200 --> 00:14:05,740 The overall thrust of it is, I think, exactly the same. 118 00:14:11,210 --> 00:14:20,290 Now. At about the same time as HUME was working on the inquiry, he was writing the letter from a gentleman, which we've already come across. 119 00:14:21,650 --> 00:14:27,770 And I just want to point out how well the logic that I've just pointed out fits with that. 120 00:14:29,270 --> 00:14:36,860 We have HUME denying that the uniformity principle can be based on sensation or intuition or demonstration or probability. 121 00:14:37,760 --> 00:14:41,600 And he uses that to conclude that it can't be founded on reason. 122 00:14:43,520 --> 00:14:46,470 And then if we look at the letter from a gentleman, we get. 123 00:14:47,360 --> 00:14:53,420 It is common for philosophers to distinguish the kinds of evidence into intuitive, demonstrative, sensible and moral. 124 00:14:54,230 --> 00:15:03,890 Moral, of course. Meaning probable. So what we have here is HUME in 1745 saying there are four commonly accepted types of evidence. 125 00:15:05,010 --> 00:15:10,860 And we have. HUME in 1748 producing an argument which ought which says, in effect, 126 00:15:11,340 --> 00:15:19,860 the uniformity principle on which all probable inference is based cannot be founded on any of these four types of evidence. 127 00:15:22,820 --> 00:15:29,480 So that seems to imply that the argument concerning induction is sceptical. 128 00:15:30,490 --> 00:15:34,750 He starts by showing that all probable inference is founded on the uniformity principle. 129 00:15:35,410 --> 00:15:39,850 He undermines every kind of evidence you could have for the uniformity principle. 130 00:15:40,940 --> 00:15:45,650 He draws the conclusion that probable inference is not founded on reason. 131 00:15:47,120 --> 00:15:54,050 That looks like a sceptical result and it's the sceptical result for which HUME is most famous. 132 00:15:57,090 --> 00:16:02,730 We'll go on to discuss that in more detail next time and to look at the implications of it. 133 00:16:03,060 --> 00:16:07,700 Thank you. Okay. 134 00:16:07,700 --> 00:16:12,800 So we've looked at Hume's argument concerning induction from a high level perspective, 135 00:16:13,310 --> 00:16:19,970 focusing on the version in the inquiry, which is the most polished of the three versions and the fullest. 136 00:16:20,810 --> 00:16:28,010 But now I'm going to go back and look at Treatise one, three six, which is where the argument first occurs. 137 00:16:28,760 --> 00:16:34,190 And we're going to take a closer look at that and see how it's structured and where it leads. 138 00:16:36,340 --> 00:16:39,580 So let's first of all, recall how HUME got here. 139 00:16:40,510 --> 00:16:49,420 He's trying to understand our idea of necessary connection in order to understand that as a component of our idea of a cause. 140 00:16:50,710 --> 00:16:55,330 And that leads him to ask why we conclude that particular causes must necessarily 141 00:16:55,330 --> 00:16:59,260 have particular effects and why we form an inference from one to another. 142 00:16:59,710 --> 00:17:07,600 If you remember, he goes by this rather circuitous route looking at the causal maxim and then turns in this direction. 143 00:17:09,580 --> 00:17:16,510 Now the key part of the process, the inference, the he calls the inference from the impression to the idea, 144 00:17:16,720 --> 00:17:21,040 I'm going to use the term causal inference sometimes a shorthand for that. 145 00:17:25,990 --> 00:17:35,500 Okay. So HUME starts by arguing that causal inference cannot be a priori because we can always conceive things coming out differently. 146 00:17:35,650 --> 00:17:46,180 Whenever we have A followed by B and we infer B from A, we can quite easily conceive A not being followed by B, so therefore the can't be. 147 00:17:46,180 --> 00:17:55,540 HUME concludes an a priori connection between the two. He seems to be making the the assumption here a common assumption common even 148 00:17:55,540 --> 00:18:01,570 in our day that any a priori inference would have to yield complete certainty. 149 00:18:02,320 --> 00:18:08,590 So if we can't be certain that a will be followed by B, it follows that the connection can't be a priori. 150 00:18:09,760 --> 00:18:16,960 It follows then that when we infer an effect from the cause, that has to be based on experience. 151 00:18:19,590 --> 00:18:29,520 What kind of experience? Well, the relevant kind of experience, HUME says, is repeated patterns of one thing followed by another. 152 00:18:29,550 --> 00:18:33,780 I'm going to call those and be without any father ceremony. 153 00:18:34,140 --> 00:18:41,730 When we've observed this repetition, we call the one cause and the other effect and infer the existence of the one from that of the other. 154 00:18:42,840 --> 00:18:48,960 Thus, we have discovered a new relation betwixt cause and effect when we least expected it. 155 00:18:49,560 --> 00:18:56,430 This relation is their constant conjunction. So we have this idea that we've gone round these neighbouring fields. 156 00:18:56,430 --> 00:18:59,610 We've been looking at the causal maxim and now at causal inference. 157 00:19:00,000 --> 00:19:05,610 And lo and behold, a surprise. We found the key to a necessary connection. 158 00:19:08,000 --> 00:19:18,800 Now, if you remember that passage back in one 3211 when HUME said contiguity and priority are not enough. 159 00:19:18,830 --> 00:19:25,660 There is a necessary connection to be taken into account. He capitalises necessary connection. 160 00:19:26,770 --> 00:19:31,840 Now, in this passage I've just quoted, he capitalised this constant conjunction, 161 00:19:32,530 --> 00:19:37,720 and the text also makes very clear that this is harking back to that earlier passage. 162 00:19:38,440 --> 00:19:42,849 Contiguity and succession are not sufficient to make us pronounce any two objects to 163 00:19:42,850 --> 00:19:48,340 be cause and effect unless these two relations are preserved in several instances. 164 00:19:48,640 --> 00:19:51,730 That is, there's a constant conjunction between them. 165 00:19:52,630 --> 00:20:02,470 So it's very clear that humans here identifying constant conjunction as the key to that elusive component of necessary connection. 166 00:20:05,470 --> 00:20:08,530 But we still got a question of how that comes about. 167 00:20:08,560 --> 00:20:13,840 How is it that constant conjunction gives rise to this elusive idea? 168 00:20:15,050 --> 00:20:20,060 Well, perhaps it will appear in the end that the necessary connection depends on the 169 00:20:20,060 --> 00:20:24,230 inference instead of the inferences depending on the necessary connection. 170 00:20:25,040 --> 00:20:29,360 So he was just going to go ahead, talk more about the inference and how that works. 171 00:20:29,870 --> 00:20:39,200 And perhaps who knows? We'll find in the end that sorting out how the inference works will give the key to the origin of that idea. 172 00:20:43,670 --> 00:20:49,760 Okay. So we've got causal inference from cause A to effect B, 173 00:20:50,780 --> 00:20:56,330 we've established that it's based it must be based on past experience because it can't be a priori. 174 00:20:57,140 --> 00:21:00,950 And the relevant experience is memory of constant conjunctions. 175 00:21:02,330 --> 00:21:06,190 The next question is whether experience produces the idea. 176 00:21:06,200 --> 00:21:11,720 That's the idea of the effect that we're inferring by means of the understanding or imagination, 177 00:21:12,260 --> 00:21:18,980 whether we are determined by reason to make the transition or by a certain association and relation of perceptions. 178 00:21:20,000 --> 00:21:23,750 Now notice in this passage I've highlighted three of the words. 179 00:21:24,230 --> 00:21:32,330 Understanding, imagination, reason. It's absolutely clear that reason is here being used as a synonym for understanding. 180 00:21:33,170 --> 00:21:41,600 And notice this instead of having another synonym for imagination, he talks about a certain association and relation of perceptions. 181 00:21:42,170 --> 00:21:52,970 So it's clear that if the inference turns out to be due to association of ideas, Hume's going to attribute it to the imagination as opposed to reason. 182 00:21:54,290 --> 00:21:59,960 And here, the key question of this section is posed as which faculty is responsible. 183 00:22:00,650 --> 00:22:04,040 Now, in an earlier lecture, we looked at Hume's faculty theory. 184 00:22:04,050 --> 00:22:07,730 We're going to be coming back to that when we discuss human scepticism. 185 00:22:08,270 --> 00:22:12,740 But notice, crucially, that that plays the major role here. 186 00:22:14,450 --> 00:22:20,870 So what Hume's now going to do is argue that reason cannot be responsible for the inference, and that leaves the imagination. 187 00:22:24,100 --> 00:22:28,809 Okay. So if reason determine this to make causal inferences, 188 00:22:28,810 --> 00:22:38,140 HUME says it would proceed upon that principle that instances of which we have had no experience must resemble those of which we have had experience, 189 00:22:38,500 --> 00:22:42,550 and that the course of nature continues, always uniformly the same. 190 00:22:43,270 --> 00:22:48,130 Now that's commonly called Hume's uniformity principle, or UAP. 191 00:22:48,310 --> 00:22:52,900 He doesn't call it that. But many people since Stroud, I think, have done so. 192 00:22:55,040 --> 00:23:06,110 Now notice that it seems to be expressed as a conditional it saying if a reason is involved, then it must be based on this principle of uniformity. 193 00:23:08,680 --> 00:23:12,100 Notice also that the principle seems very implausible. 194 00:23:13,110 --> 00:23:17,060 It's it's too strong. It seems to say. 195 00:23:17,870 --> 00:23:21,409 But the course of nature continues always uniformly. 196 00:23:21,410 --> 00:23:27,530 The same instances of which we've had no experience must resemble those we have experienced. 197 00:23:27,980 --> 00:23:34,880 That's more than we want. We don't want to say that everything in the future is going to be just like it was in the past. 198 00:23:35,360 --> 00:23:40,880 Sure, we're going to expect the same general laws to apply, but we don't expect precise repetition. 199 00:23:44,690 --> 00:23:50,870 Now in both of those respects, we get a significant contrast in the version in the inquiry. 200 00:23:51,890 --> 00:23:57,590 So he says things like this all our experimental conclusions, in other words, 201 00:23:57,590 --> 00:24:05,810 conclusions from experience proceed upon the supposition that the future will be conformable to the past, all of them. 202 00:24:06,380 --> 00:24:09,500 Whenever we make such an inference, we're proceeding on that supposition. 203 00:24:09,950 --> 00:24:15,769 We're taking for granted the uniformity principle. Now, there's no hint of a conditional here. 204 00:24:15,770 --> 00:24:20,360 It's not. If reason determined us, then we would be applying the uniformity principle. 205 00:24:20,840 --> 00:24:26,130 We are apparently applying it in all reasonings from experience. 206 00:24:26,150 --> 00:24:30,470 There is a step taken by the mind, the inference from past a future. 207 00:24:32,650 --> 00:24:43,629 Notice also that in the inquiry the uniformity principle is much vaguer than the one in the treaties, and that means it is less open to refutation. 208 00:24:43,630 --> 00:24:50,350 It's actually better. It's talking about the future being conformable to the past, resembling the past. 209 00:24:51,130 --> 00:25:00,590 We take the past as a rule for the future. Now. 210 00:25:00,590 --> 00:25:05,480 Importantly, HUME is not suggesting, even in the inquiry, 211 00:25:05,690 --> 00:25:11,390 where it seems that he's saying that the uniformity principle is always applied in inductive inference. 212 00:25:12,020 --> 00:25:17,870 He's not saying that we think of the uniformity principle explicitly when we make inductive inferences. 213 00:25:18,710 --> 00:25:27,170 Rather, I think the best interpretation here is that when we make an inductive inference, we manifest confidence in the uniformity principle. 214 00:25:28,010 --> 00:25:32,180 We are basing our predictions for the future on the past. 215 00:25:32,510 --> 00:25:37,190 And in doing so, we do in fact take the past as a rule for the future. 216 00:25:40,230 --> 00:25:44,130 So the question arises, can this assumption, 217 00:25:44,160 --> 00:25:51,420 this assumption that the past can legitimately be taken as a rule for the future, cannot be based on reason? 218 00:25:52,080 --> 00:25:55,920 Or is there some other explanation for why we do it? 219 00:25:58,020 --> 00:26:06,120 Well in what follows. We need to be aware that there are two types of argument within the standard LOCKEAN framework. 220 00:26:06,600 --> 00:26:10,020 We have demonstrative reasoning and probable reasoning. 221 00:26:10,770 --> 00:26:20,429 Demonstrative reasoning what we now call deductive reasoning yields knowledge in the strict sense, in the sense in which HUME uses it in treaties. 222 00:26:20,430 --> 00:26:28,330 One. Three. And the demonstrative inference is made up of links that are intuitively certain that are self-evident. 223 00:26:28,750 --> 00:26:36,880 So imagine a mathematical proof consisting of a number of steps, and each step is one that you can just see to clearly be valid. 224 00:26:37,860 --> 00:26:41,440 The each step follows absolutely obviously from the one before. 225 00:26:41,890 --> 00:26:46,000 And a chain of such inference constitutes a demonstrative argument. 226 00:26:46,540 --> 00:26:53,440 And in the case of a demonstrative argument, you can be quite sure if the premise is true, the conclusion will have to be true as well. 227 00:26:55,300 --> 00:26:58,300 Now, probable inference isn't so certain. 228 00:26:59,200 --> 00:27:04,690 Some links in that inferential chain and probable inference are themselves only probable. 229 00:27:06,010 --> 00:27:12,219 When you read the arguments in the inquiry, be aware that HUME uses a number of different words for this kind of inference. 230 00:27:12,220 --> 00:27:16,660 He calls it moral reasoning or reasoning concerning matter of fact and existence. 231 00:27:18,640 --> 00:27:23,110 That's pretty close to at least many modern uses of the word induction. 232 00:27:24,400 --> 00:27:31,420 I'm not going to argue the case now, but I think it is pretty clear that when HUME talks about demonstration, 233 00:27:31,750 --> 00:27:35,799 it's more or less the same as deduction in the informal sense, 234 00:27:35,800 --> 00:27:43,209 not in not formal deduction, but in the informal sense where we take an argument to be deductively valid if it's 235 00:27:43,210 --> 00:27:47,470 not possible for the premises to be true and the conclusion false at the same time. 236 00:27:50,760 --> 00:27:59,290 Okay. So HUME has just stated the uniformity principle in the Treaties. 237 00:28:00,070 --> 00:28:07,810 And without any further ado, he says, Let us consider all the arguments upon which this principle may be supposed to be founded. 238 00:28:08,440 --> 00:28:14,980 These must be derived either from knowledge or probability, and he is clearly using knowledge here to mean demonstration, 239 00:28:15,670 --> 00:28:21,790 because he immediately goes on to make the point that no demonstrative argument can do the job. 240 00:28:21,940 --> 00:28:29,800 You cannot demonstrably prove that the future will conform to the past because we can easily conceive it non-conforming. 241 00:28:30,310 --> 00:28:34,480 That's enough to show that it can't be demonstrably proved. 242 00:28:36,110 --> 00:28:41,570 What about probable reasoning? Can we use probable reasoning to argue in favour of the uniformity principle? 243 00:28:43,190 --> 00:28:48,230 Well, no, we can't. Why not? Because probable reasoning must be causal. 244 00:28:49,340 --> 00:28:56,210 Humans made the point that all inference from experience depends on the relation of cause and effect. 245 00:28:57,070 --> 00:29:01,940 Uh, we only know about cause and effect by experience. 246 00:29:02,300 --> 00:29:05,330 All inference from experience requires the uniformity principle. 247 00:29:05,750 --> 00:29:10,850 So trying to use that kind of inference to justify the uniformity principle is going in a circle. 248 00:29:12,900 --> 00:29:18,300 So neither demonstration nor probable reasoning can lead to the uniformity principle. 249 00:29:18,780 --> 00:29:23,640 And HUME immediately concludes that therefore, the uniformity principle is not founded on reason. 250 00:29:28,280 --> 00:29:32,120 Now I just want to point out that there is quite an important gap in Hume's argument. 251 00:29:34,810 --> 00:29:39,700 He's arguing that the uniformity principle can't be founded on the demonstrative argument. 252 00:29:41,360 --> 00:29:47,630 Because a change in the course of nature is possible. We can conceive of things not going on as they have in the past. 253 00:29:48,140 --> 00:29:56,150 That's enough to show that we can't demonstrate from what's happened in the past that the same thing will happen in the future. 254 00:29:58,260 --> 00:30:04,980 The mere possibility that things could turn out wrong scuppers the chance of any demonstrative argument. 255 00:30:06,150 --> 00:30:07,370 What about probable inference? 256 00:30:07,380 --> 00:30:15,900 Well, we can't use a probable inference because any probable inference, HUME says, must depend on experience and any argument from experience. 257 00:30:16,110 --> 00:30:20,460 Ipso facto, requires the uniformity principle. Okay. 258 00:30:21,600 --> 00:30:26,730 But just suppose there was a mode of reasoning that was less than demonstrative, 259 00:30:26,760 --> 00:30:32,340 that was merely probable, but was a priori that did not depend on experience. 260 00:30:33,210 --> 00:30:40,710 So you might think of some kind of statistical or probabilistic inference if there were such a form of inference available. 261 00:30:40,920 --> 00:30:48,720 HUME would not have anticipated it. He is taking for granted that all probable inference depends on experience. 262 00:30:49,770 --> 00:30:58,380 So there is a potential gap there, which quite a number of scholars and John Mackey, for example, Simon Blackburn, Roy Harrod have tried. 263 00:30:59,310 --> 00:31:06,150 To work through. They have tried to exploit that gap to give an answer to humans apparently sceptical argument. 264 00:31:09,270 --> 00:31:19,710 So let's look at Hume's conclusion. Thus not only reason fails us in the discovery of discovery of the ultimate connection of causes and effects. 265 00:31:19,980 --> 00:31:23,730 Okay. We can't discover causes and effects a priori i. 266 00:31:25,020 --> 00:31:28,890 But even after experience has informed us of their constant conjunction, 267 00:31:29,400 --> 00:31:34,350 it is impossible for us to satisfy ourselves by our reason why we should extend that 268 00:31:34,350 --> 00:31:38,790 experience beyond those particular instances which have fallen under our observation, 269 00:31:39,450 --> 00:31:45,479 we suppose, but are never able to prove that there must be a resemblance between those objects of 270 00:31:45,480 --> 00:31:49,680 which we have had experience and those which lie beyond the reach of our discovery. 271 00:31:52,140 --> 00:31:55,590 Now I've put a question mark after the word sceptical. 272 00:31:56,580 --> 00:31:59,330 It looks like a sceptical conclusion. 273 00:31:59,340 --> 00:32:06,750 We saw before that the structure of the argument, especially in the inquiry, makes it look like a sceptical conclusion. 274 00:32:07,320 --> 00:32:12,360 He's saying all this inference relies on the uniformity principle and then he's knocking 275 00:32:12,360 --> 00:32:18,220 away all the various supports that the uniformity principle might have in the treaties. 276 00:32:18,570 --> 00:32:24,270 He's taking for granted that the only two possible supports are demonstrative and probable reasoning in the inquiry. 277 00:32:24,480 --> 00:32:27,840 He's also knocking away intuition and sensation. 278 00:32:29,580 --> 00:32:38,250 So it looks like a sceptical argument. But when we come back to consider the nature of Hume's scepticism later, we'll see that that's not so clear. 279 00:32:42,640 --> 00:32:52,960 At any rate, where we are in the treaties. Hume's main business is to sort out the nature of causal inference and the nature of the idea of cause. 280 00:32:55,070 --> 00:32:58,430 He's shown that reason can't explain inductive inference, 281 00:32:59,270 --> 00:33:05,510 and he concludes that instead it must arise from the associative principles of the imagination. 282 00:33:06,410 --> 00:33:15,170 Remember, he's identified three associative principles uh, contiguity, resemblance and causation. 283 00:33:16,250 --> 00:33:22,430 And he's saying that similar kinds of principles are at work when we make causal inferences. 284 00:33:25,150 --> 00:33:32,230 Now notice that the kind of association at work here isn't quite the same as ordinary causal association. 285 00:33:33,010 --> 00:33:38,470 I mean, suppose, for example, I look at a picture and it makes me think of the artist. 286 00:33:39,340 --> 00:33:42,820 That's a causal association, an association of ideas. 287 00:33:43,780 --> 00:33:49,240 But it's not this kind of causal inference. It's not seeing a cause and inferring an unseen effect. 288 00:33:49,840 --> 00:33:54,880 Okay, it's a different thing. HUME does sometimes confuse the two, 289 00:33:55,750 --> 00:34:04,000 but I think there's a fundamental difference between the kind of association of ideas that you get with resemblance and contiguity and causation. 290 00:34:05,340 --> 00:34:11,550 Between that and causal inference where you infer something unknown, you don't simply associate ideas. 291 00:34:14,340 --> 00:34:21,310 HUME calls this associative process custom. And he says he's not by calling it custom. 292 00:34:21,330 --> 00:34:25,770 All he means is that it is a habitual process. 293 00:34:25,980 --> 00:34:29,700 It's the kind of process we identify as customary. 294 00:34:30,180 --> 00:34:36,719 You see A followed by B, you get used to A following B that start B following A and then you see and 295 00:34:36,720 --> 00:34:41,010 I and you invariably that's just the kind of thing we call habit or custom. 296 00:34:42,960 --> 00:34:47,010 Now notice that his attitude to it is not as you might expect. 297 00:34:47,220 --> 00:34:51,780 Negative custom, then, is the great guide of human life. 298 00:34:52,230 --> 00:34:56,670 It is that principle alone which renders our experience useful to us. 299 00:34:57,690 --> 00:35:02,220 That's rather an echo of Joseph Butler, who said Probability is the guide of life. 300 00:35:02,790 --> 00:35:07,139 Here we have HUME saying Custom is the great guide of life. That's from the inquiry. 301 00:35:07,140 --> 00:35:13,110 But he makes a very similar comment in the abstract, which was probably written about ten months after the treatise was published. 302 00:35:16,530 --> 00:35:24,440 He knew the end of the section 136. He says that this is essentially the same kind of custom as that which explained general ideas. 303 00:35:24,450 --> 00:35:33,120 You remember his theory of general or abstract ideas. He appeals to the idea of a customary association between a word and a revival set of ideas. 304 00:35:34,080 --> 00:35:38,220 So here he is finding another key use for the same principle. 305 00:35:38,460 --> 00:35:43,710 The custom is an absolutely central thrust in his philosophy.