Gil Harman (RIP 1938-2021): On Adam Smith, Kuhn-Loss and Historiography

One important type of ethical theory treats moral properties as analogous in certain respects to "secondary qualities" like colors. According to this sort of theory, whether something is right or wrong depends on how impartial spectators would react to it. In the 18th Century, the Scottish philosophers Francis Hutcheson, David Hume, and Adam Smith explored theories of this type.1 In the 20th Century this sort of theory has sometimes been discussed under the name "ideal observer theory."2 Recently, especially in England, there has been renewed interest in this sort of ethical theory and its comparison between moral properties and secondary qualities.3

One possible objection to an impartial spectator theory is that it seems to require an -overly aesthetic conception of morality to take the primary point of view in ethics to be that of a spectator rather than that of the agent.4 If the spectator is taken to be primary, then the agent's aim would seem to be to produce something that will or would please the spectator. But that is just wrong. Such an aim is too "outer directed" to count as a moral motive. Morality is more agent-centered than that. It is much more plausible to take the agent's point of view as primary. In the first instance morality is a matter of the moral reasons an agent has to act in one way or another, where these reasons derive from the relevant moral rules rather than from a desire to gain the approval of spectators.
A few years ago, I pressed this objection myself when I discussed the ideal observer theory in a textbook of ethics.5 But I was too hasty. In reading Adam Smith's Theory of the Moral Sentiments, I discovered that Smith explicitly considers this issue and provides a plausible reply to the objection.
Because 20th Century discussions have tended not to consider such "psychological" questions as why agents might be motivated to act in ways that impartial spectators would approve (or, for that matter, why impartial spectators would care about anything), I will in this paper ignore recent discussion and return to the three great versions of the theory that were developed in the 18th Century by Hutcheson, Hume, and Smith, indicating why I think Smith's version of the theory is superior to the others.--Gilbert Harman (1986) "Moral Agent and the Impartial Spectator" The Lindley Lecture, The University of Kansas, p. 1. [HT: David M. Levy]

For many years, I heard a story about the late Gilbert Harman, who by all accounts was a supportive mentor, that he put up a sign on his door that said, ‘History of Philosophy: Just Say No!’ And I heard the story from folk, who approvingly treated Harman as an authority within the profession, and those (mostly historians of philosophy) that were horrified by the message this sent about the value of their work and the development of curricula across the profession. The story always puzzled me because thanks to my outside reader on my PhD, the economist David M. Levy, I was familiar with Harman's Lindley Lecture, which is a major contribution to the study of Adam Smith (and his relationship to Hutcheson and Hume).

I quietly wondered whether Harman subscribed to the view -- that I am familiar with from (recall) reading, say,  Samuelson's obituary of Viner -- that only disciplinary royalty should work in history of the discipline, while the rest are worker-bees solving particular, localized problems (including puzzles in the presentation of historical works relevant to the undergrad education of the discipline). And I mused on the fact to what degree his colleagues and students even knew about the Lindley Lecture, which was, in fact, very hard to acquire in the early-internet days when I was a PhD student in Chicago in the 1990s. 

And so I always quietly assumed that Harman was on party 'team progress' (familiar from Kuhn): we need a division of labor within philosophy, and with text-book driven education sufficiently shared background commitments as a baseline that will make, just like the practice of the natural sciences, such progress, or the illusion of progress, possible. And that requires a kind of domestication of the past (and, as Michael Della Rocca has argued the rest of philosophy, too). 

At some point, I became aware of Tom Sorrell's correspondence with Harman on this very topic. But I never tried to figure out its contents. Helpfully, the Princeton website records (a good chunk) of the exchange here. And it suggests the sign was put up in the context of a controversy over the undergrad curriculum. In addition, it quotes Harman's own views on the matter:

I believe my views about the history of philosophy are mostly orthodox nowadays. The history of philosophy is not easy. It is very important to consider the historical context of a text and not just try to read it all by itself. One should be careful not to read one’s own views (or other recent views) into a historical text. It is unwise to treat historical texts as sacred documents that contain important wisdom. In particular, it is important to avoid what Walter Kaufmann calls ‘exegetical thinking’: reading one’s views into a sacred text so one can read them back out endowed with authority. For the most part the problems that historical writers were concerned with are different from the problems that current philosophers face. There are no perennial philosophical problems.

On the whole, these views about the history of philosophy are quite close to those of my late friend Margaret Wilson.

For reasons I do not fully understand, I have sometimes upset people by distinguishing between philosophy and the history of philosophy or by noting that philosophy is what the history of philosophy is the history of.

I also think as an empirical matter that students of philosophy need not be required to study the history of philosophy and that a study of the history of philosophy tends not to be useful to students of philosophy. (Note ‘tends’.) Similarly, it is not particularly helpful to students of physics, chemistry, or biology to study the history of physics, chemistry, or biology.

Of course, it may be helpful for students of physics to start with classical Newtonian physics before taking up relativity theory and quantum mechanics. But it tends not to be helpful for them to read Newton.

The playful sign that was once on my office door, ‘History of Philosophy: Just Say No!’ was concerned with whether our students should be required to do work in the history of philosophy.

That is not to say that I have anything against the study of the history of philosophy. I do not discourage students or others from studying the history of philosophy. I am myself quite interested in the history of moral philosophy for example and have occasionally taught graduate seminars in Kant. I have done a certain amount of work on Adam Smith’s relation to Hume and others.

These remarks suggest I was mostly right about Harman. I have spent a good chunk of my blogging here and previously (sometimes in debates with Mohan Matthen) at NewAPPS, explaining why I think the Kuhnian team progress picture is a bad idea for philosophy (and the historians of philosophy), so won't repeat it here. But as Harman himself stresses his adherence to it is only partial; he recognizes and accepts explicitly that sometimes the history of philosophy is useful to the philosopher. And he reminds his interlocuter that he himself teaches history, and has done it.

As an aside, the denial of perennial philosophy is a self-aware form of the Kuhnian picture of team progress. Within the sciences, the textbooks will treat the pre-paradigmatic states of the discipline as immature babblings about the same topic. Within philosophy, the status quo will treat much of the pre-paradigmatic states as very imperfect expressions on our topics, or material irrelevant to our concerns. After all, so much was packed into philosophy once that we now ignore (for various reasons).

Okay, and returning to main thread, what's left obscure by the exchange reported on the Princeton website is when it is useful according to Harman to do history. As it happens the introductory material in The Lindley Lecture articulates this: the past can be a source of information on moves that are missed in present professional discussions.

This is especially so, when a particular view or theory is re-discovered or renewed without -- one may add -- much attention to prior discussions. In such circumstances an argument or objection can be taken as decisive by a particular, localized professional community and the argumentative (ahh) dialectic is halted prematurely. And while one can discover this in the armchair, too, this is much harder if one is a partisan to the debate (or swayed by local authority). In those latter cases, a careful, sympathetic engagement with the past, without reading the present status quo back into the past, can help those on top of contemporary discussions discern that one has missed an argumentative turn.  Notice, that this suggests another, related use to history. When discussions on particular topics are renewed/revived, it might be helpful to do a careful survey of previous episodes.

Of course, the uses discussed in the previous paragraph -- let's call them 'The Useful Retrieval Task(s)' or 'TURT' -- require the availability of material/texts and specialist knowledge of where to find such episodes. That is to say, what we may call contextual historians of philosophy are a collective good to the profession that it pays to have a large enough pool to have around. Somewhat amusingly, the more antiquarian and bookish these historians of philosophy are the more useful they might possibly be on the rare occasion that retrieval of a missed distinction or argument can be useful. The same argument applies to historians of philosophy that specialize in (so-called) non-Western philosophy. (I once quoted a plea by David Chalmers to that effect, and discussed it critically in an engagement with Amy Olberding here.) Much of such philosophy will be about different matters (again no perennial philosophy), but that does not prevent multiple discoveries of the same problem at different times and locations far apart.

Now, at this point I wanted to discuss Harman's treatment of Hutcheson, Hume, and Smith as an illustration and examination of what I take to be the practice that he preaches. This is also worth doing because I think I can show some of the limitations of that approach. But since this digression has gone longer than it ought, I won't do that now, and hopefully can return to it some other time. 

So, let me close on another detail of Harman's Lindley Lecture. Near the end of his lecture, he writes that "Smith works this theory out with a mass of detail which I cannot try to summarize. I believe that the book in which he works this out, his Theory of the Moral Sentiments, is one of the great works of moral philosophy." (p. 13). And then he adds a remark (labelled postscript 2):

Finally, it is perplexing that Adam Smith's ethics should be so relatively unread as compared with Hume's ethics when there is so much of value in Smith. What I have talked about here only scratches the surface. Why should Smith's ethics be so neglected? Is it that Hume also had a metaphysics and an epistemology and that Smith did not? Or is it that Smith was a more important economist than Hume? And why should that matter? I do not know.19--Harman (p. 14)

Footnote 19 starts with "I am indebted to David Levy for getting me interested in this project." The older I get the more this footnote amuses me because it turns out that TURT can be outsourced to different disciplines. Of course, since professional economists have actively tried to eliminate professional historians of economics from professional economics (on grounds that look like the party of progress familiarized by Kuhn, and, as I have shown [here and here], actively promoted within professional economics before and after Kuhn), there is a real risk (tragedy of the commons) that no discipline will curate the possibility of TURT.

I am also amused because at my dissertation defense, David Levy asked me Harman's questions. At the time (almost twenty years ago), I already knew that Smith's ethics was constantly forgotten and rediscovered in philosophy. My favorite example of this can be found in C.D. Broad's (1950) review in Mind of A.N. Prior's Logic and the Basis of Ethics. Although, I don't think I realized then, as I do now, that this is itself an older, recurring trope. Levy himself thinks that Harman's use of the trope is distinctive because he is rare in thinking that Smith's ethics is in some key respects superior to Hume's (recall the quote at the top of this post).

In my (2017) book I show that Smith does have a metaphysics and epistemology (one that also crucially deviates from Hume), but they are mostly submerged in The Theory of Moral Sentiments. I reconstructed them primarily (note!) from other texts collected in Essays on Philosophical Subjects. To be sure, I did not write my book to address Harman's questions! And I am unsure if having showed this we can make (ahh) progress on Harman's last three questions in postscript 2.

But if we take a step back, we can see that postscript 2 asks the reader to consider the grounds of what Harman takes to be (what has come to be known as) "a Kuhn-loss." And here Harman clearly recognizes that he is on unsure grounds. He plausibly suggests that Smith's TMS got written out of the discipline (philosophy) in virtue of the fact that Smith became part of economics. And I have noted, these two disciplines became conceptualized as non-overlapping complements in the late nineteenth and twentieth century (see especially this article).

But I do think the answer to Harman's Kuhn-loss question is relatively simple. Sidgwick, who is (recall this post on the effect on Rawls) partially responsible for developing the party-of-progress conception in philosophy, wrote textbooks, including Sidgwick's Outlines of the History of Ethics, in which Smith was clearly presented as a lesser (proto-Utilitarian) Hume.

This happens in two ways, when in the Introduction of that influential book, Smith is first introduced, alongside Hume and Hartley, only Hume's views are briefly summarized and Hume's views are treated as not especially significant for ethical matters. They are treated as psychologically significant. (Sidgwick is here developing and anticipating the manner in which the empirical-normative contrast turns out to prefigure disciplinary salience). So, a natural effect of Sidgwick's introduction is that it looks like there is little in Smith for those interested in ethical affairs.

That Sidgwick actually does not intend this implicature about Smith only becomes clear obliquely. When discussing J.S. Mill, Sidwick notes that Mill's treatment of justice is very indebted to Adam Smith (p. 241 in the 1886 edition). 

But that point itself may not be noticed by those readers especially interested in Sidwick's views on Smith. For earlier, he treats Dugald Steward as a (superior) synthesis of the insights of Shaftesbury and Smith (note 2 on p. 221 in the 1886 edition).

Of course, if one reads the section devoted to Smith in Sidgwick's Outlines, one may well come accross two important features. First, that according to Sidgwick, Smith clearly improves on Hume on what Sidgwick calls the "quality of the moral sentiment" (205). But, this, too, is treated more as a contribution to psychology and the "analysis" of its phenomena (206), than to ethics. 

Crucially, for Sidgwick (p. 207), Hume and Smith jointly anticipate the view of the origin of the moral sentiments "current in the utilitarian school." (A point, as noted above, he clarifies on p. 241 in the context of Mill.) But he goes on to say that jointly their "methods of explanation compare unfavorably with that of Hartley," (pp. 207-208). So, at this point there is really no reason for a reader of Sidgwick's Outlines to come away thinking they should read The Theory of Moral Sentiments (TMS). And when he summarizes his view of Hartley, he (Sidgwick) observes that:

Hartley is obviously in earnest in his attempt to determine the rule of life, the systematic vigour which still gives an interest to his psychology, in spite of his defects of style and treatment, is not applied by him to the question of the criterion or standard of right conduct; on this point his exposition is blurred by a vague and shallow. optimism that prevents him from facing the difficulties of the problem. A somewhat similar inferiority appears in Adam Smith's work, when he passes from psychological analysis to ethical construction. He takes care to assure us that the general rules of morality impressed on us by the complicated play of sympathy which he analyzes are "justly to be regarded as the laws of the Deity;" but it can hardly be said that his theory affords any cogent arguments for this conclusion, or in any way establishes these rules as objectively valid. It would seem that the intellectual energy of this period of English [sic] ethical thought had a general tendency to take a psychological rather than a strictly ethical turn. In Hume's case, indeed, the absorption of ethics into psychology is sometimes so complete as to lead him to a confusing use of language.--Sidgwick, Outlines, pp. 211-212.

Again, a natural effect of Sidgwick's treatment is that in so far as Smith's account in TMS is interesting at all, it is primarily as a moral phenomenology or psychology of empirical affairs. But that even in that area, Hartley is more interesting. I think Sidgwick's conclusion is, in part, the result of projecting backwards a particular (anachronistic) version of the fact-value distinction onto Hume and Smith. One that become part of the professional DNA of both economics and philosophy such that a mutual, agreeable division of labor is possible (and in context Sidgwick and Marshall accepted this division).

So, I hope to have answered here Harman's postscript 2 questions (at least in provisional manner). But I also want to note that my answer illustrates one of the dangers of relying on authoritative textbook traditions (which is required by the party of progress). These can shape the self-conception of even advanced students, that even once important and famous books are mistakenly neglected. So, even the party of progress must leave room, as Liam Kofi Bright has suggested to me, for arbitrage opportunities and allow the development of seemingly epiphenomenal intellectual activities (such as history of philosophy and/or economics) that make it possible. 

Comments

[Tad Brennan]

"This happens in two ways, when in the Introduction of that influential book, Smith is first introduced, alongside Hume and Hartley, only Hume's views are briefly summarized and Hume's views are treated as not especially significant for ethical matters. They...."

I wonder whether the third instance of "Hume" was meant to read, "Smith"?

[Eric Schliesser]

Hi Tad, Thank you for reading this post with care. But I think the third instance of "Hume" is fine. The point being that Smith barely registers a mention and seems to be treated by Sidgwick as a very junior to Hume side-kick in that context.

[Aaron V Garrett]

Sidgwick treats Smith this way because his overall goal is to show the evolution of plausible methods of ethics, i.e. comprehensive and coherent normative moral theories, and as far as he can see Smith does not really contribute greatly to this. He makes a related claim about the relative importance of Gay and Hume. Hume is a far greater philosopher, but Gay is far more important to the rise of utilitarianism. Smith's normative ethics is still mysterious to me after twenty-plus years of thinking about it, so I empathize with Sidgwick who I fear has become part of my own academic impartial spectator.