How to Interpret Plato

Interpreting Plato is difficult work. It is made even more difficult with the strategy championed by Gregory Vlastos and commonly employed by Analytic philosophers. This strategy has three main commitments: (1) Plato’s views are distinguished from those of Socrates, (2) both views are constructed on the basis of the arguments presented in Plato’s dialogues, and (3) those arguments are interpreted with an eye to constructing a cohesive view for both philosophers. All three of these strategic commitments are problematic and hold us back from rightly understanding Plato’s corpus.[i] And all three commitments are derived from the same basic oversight: Plato wrote dialogues to be read.[ii]

Plato begot the tradition of Socratic dialogues as a medium for public debate. Artistic works such as tragedy and comedy were an important element of Athenian democracy. Poets were divinely inspired: the moral dilemmas presented in their works were given by the gods, and their lessons were taken seriously by the demos as a result. But by the end of the fifth century, Athens was declining. Devastating civil wars ravaged the city amid the Spartan victory in the Peloponnesian War and the subsequent dissolution of the Delian League, which provided Athens its wealth and its power.  The demos was hungry for change, and Plato, among others, was willing to bring it. He positioned philosophy as an alternative to tragedy and comedy. This is well known. But he also positioned it as continuous with these more dominant genres. It is this continuity that often goes ignored and is really at issue when we consider (1), (2), and (3). Let us turn to these commitments themselves, beginning with (1).

Plato is often denigrated as something of a scribe, someone who merely recorded the conversations of Socrates for posterity. This is what other Socratic dialogue writers purport to do. Xenophon, for instance, explicitly claims to present the conversations of Socrates as they really took place, according to what he and his sources could remember. But Xenophon criticises Plato for putting words in Socrates’ mouth that he did not say.[iii] This should be historically unsurprising since there was at least a decade between Socrates’ death and the publication of Plato’s first dialogue. Surely Plato would have forgotten the precise wording of Socrates’ conversations. But so too would have Xenophon. What seems more likely is that Xenophon’s critique is not merely pedantic but principled. He knows that Socrates didn’t say what Plato wrote because Socrates couldn’t have said those things. Socrates’ opinion differed from what Plato presented in a way that would have been well known to Socrates’ followers. So what, then, is going on? The answer is pretty simple given the history of Athenian poetry more generally. Socrates was a caricature in Plato’s dialogues just as he was in Aristophanes’ Clouds. Everyone knew who he was and some general features of his personality and his life. But more importantly, Socrates was well known to have been sentenced to death in 399, and by the time of Plato’s writing, this was a matter of severe public guilt. Plato was no doubt inspired by his teacher, but he also capitalised on the facts surrounding his life and his death for his own ends. And hence the picture that Plato gives of Socrates may not be a faithful one. It need only be close enough for his audience to recognise the character as Socrates the wandering questioner. So how then can we on the basis of Plato’s dialogues distinguish between Plato’s view and Socrates’? Quite obviously, we cannot.

The method that Vlastos and others appeal to to make that distinction is to evaluate the arguments presented in the dialogue and all of their various entailments. This method generates two significant questions, one of which we shall leave until we discuss (3). Nowhere does Plato come out and say “I, Plato, will argue for such and such in the following way.” It is left up to the audience to interpret what Plato means by presenting any given dialogue in the way that he has, and in many cases, it is precisely what is not said and not argued that is important for understanding a given dialogue. Perhaps the best example of this is the Meno. The dialogue begins with Meno asking Socrates “Can you tell me, Socrates, can virtue be taught? Or is it not teachable but the result of practice, or is it neither of these, but men possess it by nature or in some other way?” Here we have four possible sources of virtue: teaching, practice, nature, or something else. In the succeeding discussion, Socrates and Meno consider three possibilities: teaching, nature, and the gods. They end up with a hesitant commitment to the gods as the source of virtue. Yet throughout there are allusions to tasks commonly learned by practice, and its absence among consideration is conspicuous, especially given the orthodoxy of the period. This of course does not mean that Plato believed that practice is the source of virtue. It means only that Plato is clear on the inadequacy of the discussion and is pointing it out to his audience. Focusing only on the arguments of the Meno, however, would miss this possibility entirely. It would lead one to believe that Plato truly did believe that knowledge is recollection and that virtue is given by the gods. That may be the case, but it is far from certain, and Vlastos’s method doesn’t clear up the matter.

One way that Vlastos tries to clear up various ambiguities in Plato’s arguments is by committing himself to demonstrating that a given argument is consistent with other arguments elsewhere. He constructs Socrates’ and Plato’s positions as coherent wholes that manifest themselves in the arguments of Plato’s dialogues. Each dialogue then gives a piece of the overall whole that Plato’s audience must then hold together to understand what Plato seeks to show them. The difficulty of doing so should be sufficient to demonstrate the ridiculousness of this commitment. Holding all of Plato’s works in one’s head is difficult enough for specialists, let alone for Plato’s Athenian audience reading a dialogue before all of them had even been written. There of course may be some relationship between some dialogues. There was some speculation by Hellenistic commentators, for instance, that Plato presented his dialogues as tetrologies like the tragedians did.[iv] But aside from that, no one of Plato’s audience would be expected to understand how the dialogue he is reading fits into Plato’s broader view. This might be expected of Plato’s students, but the dialogues were published beyond the academy, and purposely so. As Steven Robinson points out, if Plato was simply writing for his students, he would have no reason to do so.[v] These students are philosophers themselves, and they know better that philosophy is no threat to the polis. What matters, on the other hand, is the opinion of the demos, and it is that which Plato seeks to change. And he can’t do that if the demos must first read and comment on everything that Plato has written and will write. (3) completely misses the point of Plato’s writing. Maybe we can find something common amongst Plato’s corpus as a whole, but this would be no help at all in understanding the point of any single dialogue. The really important interpretative work demands that we understand the individual dialogues, not Plato personally. And to this end, there is no reason why there can’t be ten Platos or ten Socrateses in ten different dialogues.

So how do we interpret Plato? Ultimately, I don’t know. But I do know that we can’t do it like we tend to do now. We have to treat Plato’s dialogues as dialogues. They are not essays in analytic philosophy: we cannot mine them for their arguments and be sated by that alone. We have to consider the traditions that influenced Plato. We have to understand his audience. We have to weigh what has been said against what has not. We have to admit the possibility of irony and satire and other dramatic tropes. All this is what makes Plato so difficult, but also so engaging and persistently rewarding.


[i] This is important for more than merely historical interest. I shall save a full demonstration of this for another time. Briefly, however, when we ask about how to rightly do philosophy, what philosophical methods are appropriate or otherwise, we must defer to what philosophy itself is for. This kind of functional demand can only be decided by looking to the foundations of philosophy and to the problems that it was established to solve. Plato happens to be the centrepiece of those foundations.

[ii] There is some dispute about whether Plato’s dialogues were performed. There is good reason to believe that they were, at least in some respect, but they need not have been in order to support the general appeal to the public upon which my account relies.

[iii] This criticism is specifically about the Lysis, an early dialogue of peripheral import.

[iv] Diogenes Laertius considers this rather plausible but attributes the hypothesis to earlier commentators.

[v] Steven Robinson, “Plato in the Crito” in Jonathon Lavery, Louis Groarke, and William Sweet [eds.], Ideas Under Fire: Historical Studies of Philosophy and Science in Adversity, Fairleigh Dickinson University Press (Madison, WI), 2013: 37-65

What Makes Teleology Immanent?

There are two broad ways to conceive of the metaphysics of teleology. Teleology may be dispositional or it may be axiological. What distinguishes these two kinds of teleology is the source of the teleological relation. If teleology is dispositional, some item i has some end e in virtue of some metaphysical or explanatory relationship between i and e. This is the kind of teleology that philosophers of science such as Hempel, Braithwaite, Wiener, and Wright had in mind.[i] If teleology is axiological, on the other hand, i has e in virtue of some value that e instantiates. Axiological teleology is far more common in the literature, stretching through history from the Ancients to contemporary metaethics, epistemology, et cetera.

For e to be immanent in i, e must be in some way contained in or determined solely by i. That is, i must be autotelic. This means different things for dispositional teleology than it does for axiological teleology, however. Let’s begin with the former. If i has some immanent disposition to realise e, two conditions must obtain: (1) there must be something about i that reliably picks out e, and (2) e explains i. If only (1) obtains, there is no teleology. It is at best efficient or formal causation. For any instance of immanent teleology, e makes i reliably pick out e. While this may sound crazy, there is no shortage of accounts that attempt to make this intelligible. The most common is Wright’s account of teleology as causation by consequence, or what he calls consequence etiology. Wright’s (explanatory) schema is as follows:

i is an element of some system s for the sake of e if and only if both s is disposed to satisfy e in virtue of i given some normal condition c and s was disposed to satisfy e in virtue of i given c in the past.

Both (1) and (2) above have representatives in Wright’s schema. (2) here is self-consciously framed in terms of evolutionary adaptation. This need not be strictly biological, but biological evolution is the core of the account. Insofar as this is the case, however, there is a clear bait and switch. e here is not the end of i (ei) but the end of s (es). ei is not properly explanatory in any adaptation; i may be disposed to realise ei, but ei alone gives no reason for why i should be an element of s. For instance, it might be nice for spiders to have wings so they can fly. But they have no need for flight given the niche they occupy. The ability to fly is not sufficient for explanation. Rather, ei must contribute to es and hence the latter bears the explanatory and metaphysical significance. And this means that i is not autotelic. Rather, e is transcendent of i.[ii] The same pattern occurs for every known account of immanent dispositional teleology and seems to follow immediately from (2).

Axiological immanent teleology faces the same fate, though in a much more interesting way. If e is immanent in i then three conditions must be met: (3) e is valuable with respect to i, (4) i has some duty to obtain the value that e provides because e is valuable for i, and (5) e explains i. (5) here is of course familiar to us as the problematic condition in dispositional teleology, and one might think that it would be problematic here too. But this is not necessarily the case. It is common in the history of philosophy to suggest that existence is more valuable than non-existence. If this is the case, then the value that e provides might be sufficient to explain i, even if the value is value for i alone. I suspect that the ultimate fate of this strategy is no better than that of dispositional teleology, but there is at least a tradition here to give prima facie support. Rather, the fundamental problem arises here from (4). For (4) to obtain, i must in some way give itself a duty to obtain the value that e provides. But objects cannot have duties to themselves.[iii] In a manner similar to Wittgenstein’s private language argument and his comments concerning the standard metre, there is no distinction between the definition of a duty and the performance of it. Objects do not hold themselves accountable or even evaluate themselves in the performance of some duty. They merely act. To insist otherwise is to add some extra element to i that serves as the locus for evaluation. In this case, though, the value no longer belongs to i per se but to this evaluating part. The value of e loses its explanatory power in this way. So either axiological immanent teleology misses out on (4) (and subsequently also (3)) or it misses out on (5). In either case, axiological teleology cannot be immanent.

When we ask what makes teleology immanent, two options present themselves. But when neither can possibly obtain, we are forced to admit that immanent teleology does not obtain. This leaves us with only two options. Either there is no teleology at all or teleology is transcendent. I think that we should lean towards the latter.


[i] In practice, both Wiener and Wright ended up with accounts of axiological teleology despite their attempts to restrict themselves to bare dispositions. In the long run, this probably counts as evidence against a purely dispositional account of teleology. It seems that any moderately successful account of teleology has to import axiological considerations.

[ii] Some philosophers have taken this to be sufficient to be a refutation of inflationary accounts of teleology per se and propose a deflationary account in their place: the teleology we see is an appearance generated by an objects causal role in a system and nothing more. In essence, this is the foundation of the dispute between causal role and selected effect accounts of functions in the philosophy of biology. Both types of accounts are confused about the relevant metaphysics.

[iii] One might suggest the counterexample that people at least sometimes have duties to themselves. This is only possible because people are self-distinct. It is possible that one’s definition of a duty is distinct from the performance of it. That is, people are not metaphysical simples. But this entails transcendence, not immanence.

Moral Fault and Moral Responsibility: Why We Should Always Blame

In this post, I will argue that blame is an expression of a judgement about moral fault. A moral fault is a property F of some object O in virtue of which O impedes the attainment of some morally optimific state of affairs. The orthodox position stands against this. It limits blameworthiness to those objects that are morally responsible for wrong actions. O is morally responsible for some act A if and only if O is responsive to moral reasons pertaining to A. This orthodox position does have a few arguments in its favour given that moral fault accounts of blame tend to have a number of unintuitive consequences. Moral responsibility, however, has a far worse consequence than any implied by moral fault, namely robust blame skepticism such as that defended by Gideon Rosen. My argument will focus specifically on this.

Rosen argues that blame requires either original responsibility or derivative responsibility. There are two conditions that must be met in order for O to be originally responsible for A: there must be some moral reason R that is a decisive reason for why one ought not to do A, and O must be aware of R. Sometimes, O might not be aware of R because O has failed to “discharge one of [its] procedural epistemic obligations.”[i] O may be originally responsible for this failure. And if that is the case then O is derivatively responsible for A.

So far so good. Rosen’s account meets all the normal expectations of an account of moral responsibility. It has the right sorts of excusing conditions and the right sorts of requirements. But blame is about more than just moral responsibility. Blame requires a judgement of moral responsibility, and a particular expression of that judgement. This expression is unpleasant and possibly harmful, and therefore the prospective blamer must be certain that O is in fact morally responsible for A before she can act. This certainty, however, will always elude us. The temporal distance and the opacity of other minds bars us from any knowledge about whether O was in fact aware of R when it did A. Hence we can never hold someone originally responsible for A. But we likewise cannot hold someone derivatively responsible for A either, since every instance of derivative responsibility requires some instance of original responsibility and the problem reasserts itself. Blame skepticism here is unavoidable.

Blame skepticism is a detestable conclusion. Blame serves an important role in moral education and in deterring wrongdoing.[ii] We ought therefore to preserve some account of blame. This is where moral fault comes in. Judgements of moral fault have a much less demanding epistemic burden, and hence are not vulnerable to blame skepticism in the same way. But this is so in part because moral fault accounts of blame are incredibly harsh. Objects are accountable for any failure to realise some morally optimific state of affairs, whatever the reason. They are blameworthy for the their own bad intentions and the ill effects of moral luck all the same. But this shouldn’t be too problematic. It is uncontroversial that ordinary objects of blame are imperfect. Humans and other creatures have their limitations, and we should not be denied from pointing these out from time to time. We need only remember that as blamers, we too would be blameworthy for failing to realise some morally optimific state of affairs, and we may fail to do so because we blamed another. Luckily, then, blame need not be fire and brimstone. It may also be a prod or a joke, a stern glare or a quick lesson.


[i] Rosen, “Skepticism about Moral Responsibility,” Philosophical Perspectives, 18(Ethics), 2004: 309

[ii] For more on the function of blame, see Christopher Bennett, “The Expressive Function of Blame” in D. Justin Coates and Neal A. Tognazzini [Eds.], Blame: Its Nature and Norms, Oxford University Press (Oxford, UK), 2013: 66-83; and Christopher E. Franklin, “Valuing Blame” in D. Justin Coates and Neal A. Tognazzini [Eds.], Blame: Its Nature and Norms, Oxford University Press (Oxford, UK), 2013: 207-223.