Demea now breaks into the conversation, and asks whether it is only God's nature and not God's existence that is being subjected to skeptical doubt? His friends assure him that this is the case. Well, he continues, as far as the latter is concerned, it is almost as impious to claim that we can actually understand God's nature as it is to claim that there is no God at all. God, he declares, is inherently beyond human comprehension, and must necessarily remain mysterious to us.
Philo agrees that God's existence is beyond doubt and he also agrees that God's nature cannot be known. He provides arguments for both of these claims. First, God must exist because every effect has some cause and so there must be some ultimate cause of the universe. We call this ultimate cause God and we "piously ascribe to him" every possible type of perfection. But there is no reason to think his perfection bears any resemblance to anything we know, so there is no reason to think that we have any idea what God is like. We say that he is wise, knowledgeable etc., but we only use these words for lack of any others. We have no idea what any of these attributes could mean when put in the divine context. Philo points out that this conclusion is based on the most common-sense reasoning: our ideas are produced by our experience, and we have no experience of divine attributes and operations. Therefore, we have no idea what God could be like.
Cleanthes disagrees with Demea and Philo. He thinks he can prove that, even though we do not have any direct experience of God's attributes and operations, there is enough evidence in nature to allow us to draw justified conclusions about what God is like. By looking at the natural world we see that it resembles nothing so much as a work of human artifice (which, for the sake of ease, we can just call a "machine"). Though we have never experienced God, we have experienced machines and we know a thing or two about them. Most relevantly, we know that wherever there is a machine, there is some intelligent designer behind it. Machines do not just come together by chance; they are created by skilled human beings. Given that the universe is obviously just an elaborate machine, with each part from smallest to largest perfectly adapted to the harmony of the whole, we can reasonably infer that, just like any other machine, the universe was created by an intelligent designer. That intelligent designer, i.e. God, must be similar to a human designer, only much more perfect, in proportion with the greater perfection of his art.
Demea is the first to react to Cleanthes' argument from design. He does not approve of the claim that God and man are at all similar. Second, he is unhappy that Cleanthes is trying to use an a posteriori proof rather than a priori for God's existence (since the argument by design proves both God's nature and his being). A posteriori proofs are only probable proofs, not definite proofs. That is, when we give an argument from experience we can only prove that our conclusion is more likely than not; we can never prove that it is definitely true. It is only when we give an a priori demonstration that we can prove something with certainty.
Philo does not mind that the argument is a posteriori; his only complaint is that it is a bad argument. He will spend the rest of the book showing just how bad the argument is. In this chapter all of Philo's objections aim to show either that the argument from design does not really have the proper form necessary for an inductive inference or that it is an improper use of an argument by analogy.
Philo brings up several ways in which the argument from design fails as an inductive inference. First, the analogy between the universe and a machine is weak, for he claims that the world does not really resemble a machine all that well. Second, the analogy between the universe and a machine does not necessarily work because it is not an analogy between two separately existing entities, but between the universe as a whole and certain parts of the universe (i.e. man and the objects he manipulates). Thus, drawing an analogy between a machine and the universe might be like trying to figure out how an entire man develops by looking at how a single hair in his head grows. Third, it seems to be false to claim that all order in the world is the result of intelligence. The process responsible for the intricate order and adaptability in the case of organic bodies seems to be animal and vegetable reproduction, not design. So why suppose that the order of the universe is like the order of man-made machines, and not like the order of organic bodies? Finally what makes a causal inference work is that we repeatedly experience examples of A-like events followed by B- like events. But here A, God, is a unique cause and B, the universe, is a unique effect. Thus Philo concludes that the argument from design is not so much an inductive inference, as a whimsical conjecture.
There are a lot of very important arguments in this section, and each one should be analyzed carefully: there is Philo's argument for God's existence, the argument from design, and each of Philo's objections to this argument.
Philo's argument for God's existence is particularly interesting, given that we know Hume had a reputation for atheism. Looking closely at Philo's argument, we see that the conclusion is actually very weak and also irreligious. The conclusion is only that there is some first cause of the universe and that we can call this initial cause "God". Just saying that there is some first cause of the universe, though, is not in itself a religious claim. This first cause might be some singular event that required no cause, and not an intelligent being. No theist would be very happy with Philo's conception of his deity. The theist's conception does not enter Philo's account until after the argument has already concluded. It is tacked on at the end: we "piously ascribe" to this first cause all the perfections we can think of. But why do we piously ascribe all of this? According to Philo's account, we do not have any rational reason for doing so. This might have been Hume's way of showing that, despite what his characters claim, there is no rational basis for believing in God's existence.
Next, Cleanthes' argument from design: the argument is meant to be an inductive inference, helped along by an argument by analogy. In order to understand all of Philo's objections, it is important to see how the design argument is supposed to work step by step.
In general, an inductive inference is supposed to look like this: You see that A is B numerous times. Let's take the example of fire. Every time we've seen fire (which we'll substitute for A), we notice that it is 'something which has the power to burn other things' (which we'll call B). Because of this constant conjunction of A and B in our experience, we infer that there is some sort of connection between A and B and therefore expect that all future A's will be B's. Thus, we assume because we have noticed many times that fire is something which has the power to burn, we assume that the next time we see fire, it will have the power to burn. In his Treatise, Hume shows that we are not logically justified in believing that the next time we see fire, it will have the power to burn. However, he argues that we gain evidence by seeing repeated conjunctions, and are being reasonable if we expect the next instance of fire to have the power to burn.
Inductive reasoning functions in the argument from design in that every time we have seen 'a machine' (which we can call A), that machine was 'made by a man's intelligence' (B). Therefore, we conclude that whenever we see a machine in the future, it will have been created by human intelligence. Let us call the argument that allows us to infer design as the cause of machines the Machine Inference.
The argument from design then adds a twist onto the Machine Inference. This twist, called an argument by analogy, allows us Machine Inference to talk about the universe even though the universe is not technically a machine. In general, an argument by analogy allows us to take some D and claim that it is similar enough to A that we can use it in inferences as if it actually were something similar to B. Since D is so like A, and every A is a B, every D must be a C, which is something similar to B. Let us take an example of such an argument: we know, for instance, that fire which is red (A) is something which has the power to burn my hand if I leave it in the fire for ten seconds (B). We think that fire that is blue (D) is analogous to fire that is red. Thus, we believe that fire which is blue must be have a property which is similar to the power to burn my hand if I leave it in the red fire for ten seconds. In this case, the argument by analogy works, for fire that is blue has the power to burn my hand if I leave it in the blue fire for three seconds (C). C in this case is similar to B. However, arguments by analogy are tenuous at best, for instead of choosing the property of burning, we could have chosen color, in which case the argument would fail.
Now we can see how the argument by design works as a whole. (1) In my experience, whenever I have encountered a machine, that machine was made by a human intelligence. Therefore, (2) all machines are made by human intelligence. (3) The universe is analogous to a machine. (4) Therefore, the universe must have been made by something which is similar to a human intelligence.
Now understanding the argument by design, we can better understand all of Philo's objections. First, he claims that the analogy is no good. He claims that the universe and a machine are not similar in the way that the red and blue fires are similar, and for this reason, an argument by analogy is not valid. Philo's second objection is that the analogy does not work because it is between a whole and a part of that whole. A machine is a part of the universe, and it makes no sense to assume that one part of the universe is analogous to the whole of the universe just because we have no experience of the other parts.
Philo's third objection is that not all order is the result of design. Thus, it is possible that the universe is not analogous to a machine even though it is ordered; it might be analogous to some other form of order and not to a man-made structure. For instance, some highly ordered systems that we know of are the result of reproduction instead of intelligent design. Just because there is order, therefore, it does not mean that we can conclude that such an ordered system is analogous to a man-made system, and it is therefore even less reasonable to assume that all order is the product of design.
Philo's last objection is that the origin of the universe is a unique effect. That it is unique would make inductive reasoning impotent to discuss it, for induction would require that we had witnessed several creations of several universes and knew that God had created each universe. However, Cleanthes does not appeal to such reasoning, and as such it is clear that Philo thinks that because the universe is a unique effect, it is probably not analogous to machines, which are not unique.