Aesthetic Testimony and the Test of Time

It is a relatively simple matter to specify, at least in broad terms, the conditions under which we can come to know that a particular artwork has various descriptive properties; that a painting is a certain size, a sonata in a specific key, a novel by a particular author, and so forth. By contrast, there is considerable disagreement as to how to properly judge the evaluative aesthetic properties of an artwork. In this paper I consider two influential claims concerning such judgements (hereafter ‘aesthetic judgements’) and argue that there is a hitherto underexplored tension between them. The first of these (TT) maintains that the surest test of the aesthetic value of an artwork is how well its reputation weathers ‘changes of climate, government, religion, and language’ (Hume: 1757 / 1875: 255); that is, whether it passes ‘the test of time’. 1 The second (NT) is the view, often referred to as ‘pessimism concerning aesthetic testimony’, according to which testimony cannot serve as a legitimate source for aesthetic judgements. That ‘a thing has pleased others could never serve as the basis for an aesthetical judgment’ (Kant 1790 / 2005: 94).


3
Those who endorse TT typically construe the relevant test as asking whether a particular aesthetic judgement survives across a range of times. For the judgement that Homer is a great poet -and, mutatis mutandis, that Michelangelo's David is beautiful or that Mozart's symphonies are superior to Salieri's -to pass the test of time it must be endorsed not only by critics during Homer's lifetime, or during our own, but by a range of suitable critics throughout different historical periods. And it is an interpretation of TT along these lines which I will adopt throughout this paper. In particular, I think we should construe TT as an epistemic claim, one according to which we can often come to know that a work possess a particular aesthetic property on the basis that the judgement that it possess that property passes the test of time. I am also construing the test here as primarily concerned with whether an aesthetic judgement endures across time. As such, I intend talk of a work's passing the test merely as a shorthand for the claim that a particular judgement concerning that work -which, unless stated otherwise, I will take to be the judgement that the work is excellent -passes the test. This interpretation is, however, a controversial one and others have formulated the test so as to involve a more direct appeal to something like the survival of the work itself. I will have much more to say about why I believe we should accept an interpretation of the kind I have proposed in §3, but for now I will content myself with mentioning some further points of clarification with respect to my preferred understanding of TT.
First, I intend TT only to serve as a source of information concerning those works which pass the test of time and to be silent with respect to those which fail to do so. Works can fail to pass TT for a number of reasons which have no bearing on their aesthetic character. Most obviously because, as Savile (1977: 203) suggests, they 'are never even brought to the test; they are simply not available, being lost or destroyed, damaged or obscured'. And even those works which are brought to the test might systematically fail to pass it for reasons which are irrelevant to their artistic status. For example, (as I will discuss in §5) many works which possess great aesthetic value are systematically judged to lack such value owing to factors such as the sex, race, or class of the artists who created them. As such, I do not believe that we can draw any interesting conclusions about the aesthetic value (or lack of same) of a work merely from its failure to pass TT.
Second, it is worth noting that the test, as I have outlined it, is rather underspecified in a number of respects. I have not, for example, said anything about how to deal with a class of difficult cases highlighted by Hume himself (1757Hume himself ( / 1875, where the apparent consensus surrounding a work results, at least in part, from its being interpreted very differently by different critics. 5 One might wonder, then, to what extent some work, w's, passing TT is compatible with the critical consensus surrounding w having resulted from its being interpreted in radically different ways. 6 My strategy will largely be to sidestep this issue by focusing on cases where it seems clear that the critical consensus I am discussing does not result from such radically different interpretations. This does not, of course, require that there are no differences amongst critics in interpreting these works (it is not clear whether any work would pass that test), but merely that such differences are typically not central to their interpretation and evaluation of the works in question. Nor does it require that there is absolute unanimity with respect to such interpretations. As such, the existence of, for example, trenchant Freudian critics who insist on a psychoanalytical interpretation and evaluation of all of the works I discuss below would not undermine my claims. What is required is that there exists a wide range of suitable critics across time who have interpreted the works in question in substantially the same way and who have, on that basis, converged on particular aesthetic judgements concerning them.
Nor is this the only important question which my account has left unanswered. I have not, for example, said anything about precisely what makes someone a suitable critic, or given any indication as to how long the critical consensus concerning a work must endure in order for it to pass TT. Of course, anyone intending to provide a complete account of the nature and value of the test of time would need to addresses these questions (and many more besides). However, since my intent here is nowhere near this ambitious -I wish merely to defend the plausibility of a particular general approach to TT rather than the details of any specific account -it will serve my purposes better to avoid such controversies altogether and to focus instead on some clear core cases of judgements which successfully 5 See, e.g., Feagin (1982). 6 I thank an anonymous referee from the journal for pushing me to consider this point in greater detail. pass TT. I will, for example, only discuss cases where the consensus concerning a work has endured for well over a century, thus meeting a range of different proposed standards for the length of time required to pass time's test. 7 Similarly, I will aim to allay worries about which critics we take to be suitable for inclusion by focusing on well-known canonical works which have been discussed and evaluated by a wide range of critics in different ages, rather than on judgements concerning more esoteric works which may only ever have been discussed in rather narrow circles.

Pessimism
In order to understand NT (and, indeed, TT discussed above) we must ask what is meant by the notoriously promiscuous phrase 'aesthetic judgement'. I intend 'judgement' here to be taken in the sense in which it is often used in contemporary debates in metaethics, such that a judgement is the mental correlate of a declarative sentence and is expressed 'by sincere assertoric use of [such] a sentence' (Sinclair 2006: 253). 8 Understanding 'judgement' in this way will doubtless raise immediate concerns for some readers, and I will endeavour to address these in the next section. For now, though, I will proceed on the assumption that NT is to be interpreted as the claim that aesthetic judgements of this kind cannot be legitimately formed on the basis of testimony.
A version of NT which maintains that aesthetic judgements formed on the basis of testimony cannot achieve the status of knowledge has been widely defended.. Philip Pettit (1983: 25), for example, argues that '[a]esthetic characterisations are essentially perceptual' meaning that perception is the only route to 'full knowledge […] of the truths which they express' and Daniel Whiting (forthcoming: 17) that 'testimony cannot deliver aesthetic knowledge'. 9 I do not, however, mean to limit NT to serving such a narrowly epistemic role 7 See Savile (1982: 9-10). 8 See also Lillehammer (2002: 1-2). 6 since any such restriction would exclude two interesting varieties of pessimism from consideration. First, the view, most notably defended by Rob Hopkins (2006 according to which aesthetic beliefs are constrained by certain 'non-epistemic' norms, norms which would render such beliefs illegitimate even in cases where 'the belief the recipient is offered would count as knowledge' (ibid. 147). 10 Second, the view defended by those, such as Cain Todd (2004: 290), who maintain that advocates of NT should not 'hold that aesthetic judgements are beliefs […] at all' but, rather, that they should account for such judgements in an expressivist or quasi-realist manner. 11 In what follows I will, for ease of exposition only, assume that aesthetic judgements are beliefs in a straightforward sense, and that they are governed exclusively by epistemic norms. As such, I will present both NT and TT as narrowly epistemic claims. Crucially, though, neither of these assumptions is relevant to my overall aim in this paper, and the conclusions I reach in later sections could equally well be supported (mutatis mutandis) by arguments which appeal to non-epistemic norms of belief or expressivist accounts of aesthetic judgement. 12

Where the Conflict Lies
Given what I have said above, the reader will, I suspect, already have some idea as to why I take NT and TT to be in conflict. NT maintains that we cannot achieve aesthetic knowledge on the basis of testimony, whereas TT maintains that we can frequently come to know that a work possesses some aesthetic property, p, by learning that the judgement that it possesses p passes the test of time. Yet, an aesthetic judgement's passing the test of time is a matter of that judgement's being endorsed by a range of different critics across time. That is, of the 10 See also Gorodeisky (2010). 11 See also Scruton (1976). 12 Similarly, my account is intended (for reasons discussed in, e.g., Robson forthcoming) to be only superficially dependent on the assumption that aesthetic judgements are constitutively governed by norms rather than, say, aims or functions. claim that the work possess p being suitably supported by testimony of the relevant kind (testimony, perhaps at second or third hand, from these critics).
One might reasonably worry, though, that this account of the apparent conflict is problematically simplistic in some key respects. 13 First, it might reasonably be pointed out that while TT, as I have construed it, concerns judgement (a private mental state), testimony requires someone to actually assert -or otherwise express -the judgement in question (a public social act). As such, it would, at least in principle, be possible to rely on TT without violating NT if one could form aesthetic judgements entirely on the basis of the unexpressed private judgements of the relevant critics. In practice, though, such a feat would, of course, be impossible (we do not have access to the private mental states of contemporary critics and, still less, to those of critics long since dead). As such, simplifying things by focusing only on the expressed judgements of the critics in question, rather than their judgements simpliciter, will be unproblematic.
A deeper worry concerns the means by which such judgements are expressed. While critics will often express their judgement that, e.g., a particular painting is excellent by straightforwardly testifying as to its excellence, this is not the only means by which such a judgement can be conveyed. A critic may, for example, signal this judgement by making approving sounds or facial expressions in the presence of the painting or by paying large sums of money to acquire it. Further, a critic will often go beyond merely expressing her judgement with respect to a particular work, she will also typically be 'committed to backing up her evaluations with reasons' (Carroll 2009: 7-8). Yet, despite such complications, I take it that simplifying things as I have done above is unproblematic for two reasons. First, it is (as a matter of descriptive fact) extremely rare for us to have records of these other expressions of approval which spread as widely or endure as long as those of a critic's testimony. As such, TT will still rely primarily -though by no means exclusively -on testimony rather than these other indicators of critical judgement. Similarly, while many critics -from other 13 I thank Daniel Whiting and an anonymous referee from the journal for encouraging me to address these concerns in detail. 8 times as well as our own -have provided some indication as to their reasons for believing, say, that The Marriage of Figaro is excellent, it is commonplace for someone who is aware of this critical consensus to be entirely ignorant as to the reasons particular critics offer in support of it.
Second, those who are critical of our reliance on testimony in particular domains are typically equally reluctant to permit judgements based on other kinds of deference to the views of others. 14 Indeed, pessimists in aesthetics tend to be even more restrictive concerning what they take to be the proper sources of aesthetic judgement. While the letter of their pessimism only commits them to excluding testimony, most pessimists reject not only deference more generally but also any source of judgement (such as inferences from so called 'principles of taste', enumerative induction etc.) other than first-hand acquaintance with the object judged. 15 Such sweeping prohibitions are typically justified by endorsing some broad principle of aesthetic judgement such as Richard Wollheim's Acquaintance Principle, according to which 'judgments of aesthetic value [...] must be based on first-hand experience of their objects ' (1980: 233). And even pessimists such as Hopkins who are keen to explicitly differentiate claims about the legitimacy of testimony from those concerning the legitimacy of these other sources (2000: 212-3, 2006: 86-90) still end up appealing to wider principles (Hopkins 2011: 149-53) which rule out all such sources as legitimate for aesthetic judgment. 16 Nor is this merely a descriptive issue concerning the behaviour of typical pessimists.
Those who accept NT will, of course, be under some pressure to explain why testimony -14 See McGrath (2009) 15 See, e.g., Tormey (1973: 39), and Wollheim (1980: 233). These claims are often qualified to some extent to allow for, e.g., experience of appropriate surrogates but such complications are irrelevant for my purposes.
which consistently proves so fruitful in other domains -is inadmissible as a source of aesthetic judgement. And the most plausible explanations which pessimists have offered typically appeal to some putative general feature of aesthetic judgements, such as their being 'essentially perceptual' (Pettit 1983: 24), or their needing to be 'based upon a feeling of pleasure' (Todd 2004: 278). 17 It is, however, difficult to conceive of any such explanation which would plausibly rule out testimony as a legitimate source of aesthetic judgement while admitting the other expressions of critical approval discussed above.
A related worry is that TT not only admits a number of sources other than testimony, but that, strictly speaking, it omits appeal to testimony altogether. The concern here is not that the test won't involve appeal to the assertions of past critics, but that the assertions in question won't count as genuine testimony (or, at least, not as testimony to those of us applying TT in the present day). A number of recent accounts of testimonial warrant have stressed the need to incorporate 'the role seemingly played by a distinctive kind of interpersonal relationship between testifier and testifiee in our everyday practices of justifying testimonial-based beliefs' (Wanderer 2013: 92). Advocates of such 'interpersonal views' are typically keen to stress the epistemic significance of certain social aspects of testimonial exchanges, such as trust between speaker and hearer (Faulkner 2011), or the fact that the testifier, in offering their assurance to their hearer that P is the case, thereby takes responsibility for the truth of P (Moran 2005). Without taking such factors into account, they argue, we are merely treating our interlocutors as truth-gauges of various degrees of reliability and are ignoring what makes testimony so epistemically interesting. We might worry, then, that someone sympathetic to this family of accounts would claim that TT doesn't, appearances to the contrary, typically involve any appeal to testimony. After all, we rarely have any kind of interpersonal relationship with the contemporary critics whose views we encounter, let alone with those from past ages. 17 Kant's (1790Kant's ( / 2005: 94-7) own account of the nature of aesthetic judgement excludes not only testimony but any 'empirical ground or proof' as well as any 'a priori proof'.
There are a number of things which could be said in response to this worry. I would not, for example, be particularly concerned to learn that there was a genuine tension between the claims I have made above and interpersonal views of testimony, since, to put things bluntly, I believe (for reasons offered in, e.g., Lackey (2008: 221-50)) that such views are mistaken. Arguing for this would, however, take me far beyond the scope of this paper.
Fortunately, more concessive responses are available. One is to merely highlight that I am intending 'testimony' in this paper be used in a very broad sense so as to include 'tellings in general (i.e with no restrictions either on subject matter, or on the speaker's epistemic relation to it)' (Fricker 1995: 396-7); a sense which will, as Sosa (1991: 219) points out, include 'posthumous publications' which the speaker 'might direct to the world at large and to no one in particular'. Nor is this usage merely stipulative. Rather, it is commonplace within the aesthetic testimony debate as it has been conducted thus far. Kant's (1790Kant's ( / 2005 original defence of NT appealed to the impotence of 'a hundred voices all praising' a work in serving as grounds for an aesthetic judgement without feeling the need to specify anything about the relationship, if any, between speakers and hearer (and discussion of such factors is conspicuously absent from subsequent discussions). Further, a number of discussions of NT (such as Hopkins (2011: 154), Laetz (2008: 355), and Levinson (2005: 213)) directly discuss cases of testimony from past critics. As such, it remains clear that TT is in conflict with NT as it has been understood by both sides of the aesthetic testimony debate.
Finally, it is important to stress that -while their focus typically remains on cases of direct assertion between individuals engaged in conversation -advocates of the interpersonal often maintain that their view of testimony can be extended to cover a range of other cases.
For example, in his recent book defending a version of the interpersonal view, Paul Faulkner allows that we 'can acquire testimonial knowledge […] from illicitly reading someone's diary ' (2013: 197) as well as from historical records (Ibid. 1). As such, it is not clear that the appeal to the judgements of past critics in TT doesn't involve testimony of a kind which would be amenable even to advocates of interpersonal views.
It seems, then, that there is a clear conflict between NT and TT as I have understood them.
However, some might worry that the putative conflict I am addressing does not reflect a genuine tension between NT and TT themselves but that it is merely the result of some of the assumptions I have made in spelling out my preferred interpretations of these claims. In the next section I will argue that this objection is mistaken and that there is no way to successfully dissolve the conflict I have highlighted.

Appeals to Equivocation
One strategy of this kind maintains that the tension arises only because of an equivocation over the meaning of 'judgement'. In one sense, of course, this objection is clearly mistaken since I have stipulated above that I intend 'judgement' to be understood in the same manner in both NT and TT. The real worry, though, is that those who have endorsed NT have typically meant something very different by 'judgement' than those who have accepted TT.
If this were the case, then it would seem that the conflict is merely an artefact of my prior stipulations and that we have no reason to believe that the claims, as standardly understood, are in tension. Further, such an equivocation would hardly be surprising given the manifold different ways in which 'judgement' has been used within aesthetics (and even more so within philosophy more broadly).
The most plausible version of this story appeals to an interpretation of NT according to which it should be taken as meaning that we cannot appreciate the aesthetic value of an object merely on the basis of testimony. 18 Understood this way, NT certainly seems plausible; further, it removes any apparent conflict with TT construed as an epistemic claim. It is perfectly consistent to accept, on the basis of the testimony of generations of past critics, that the language in Shakespeare's sonnets is beautiful while maintaining that we have not ourselves appreciated their beauty. Further, it does seem eminently plausible that some of the controversy over NT has been a merely verbal dispute between those who deny that testimony is a legitimate source of aesthetic appreciation and those who accept it is a 12 legitimate source of aesthetic belief. 19 Yet it is clear that this strategy has its limitations. Even those, such as Aaron Meskin (2004: 76), who argue against pessimism, frequently concede that 'there are things that testimony may never provide-aesthetic experiences and artistic appreciation' and, more importantly for my purposes, a number of pessimists have been explicit that their claim is not merely one concerning appreciation. The quotes from Hopkins, Pettit, Todd and Whiting discussed above, for example, make it clear that all of these authors have adopted something like the understanding of NT which I have been proposing. As such, it is not plausible to maintain that the tension between NT and TT can be dissolved in the manner proposed.
A second possible source of equivocation concerns the 'aesthetic' part of 'aesthetic judgement'. I have thus far been treating both NT and TT as if they were general claims concerning a wide variety of evaluative judgements which we might make concerning artists and artworks. Yet, some may deny this. Those who discuss NT typically focus on judgements relating to aesthetic concepts in Frank Sibley's (1959) sense; that is to concepts such as gracefulness, gaudiness garishness and so forth. Whereas many of those who discuss TT have focused on what Sibley (1965: 136) terms 'verdicts'; that is, on 'purely evaluative judgments' as to 'whether things are aesthetically good or bad, excellent or mediocre, superior to others or inferior, and so on'. Given this, we might, again, maintain that there is no genuine conflict between the two claims. 20 As with the concerns relating to 'judgement' above, there is certainly something to this worry, and it may well be that certain restricted versions of our two key claims are not genuinely in conflict. However, this strategy clearly offers no defence against the claim that NT and TT are in tension when applied to judgements of the same kind. And I will argue, in §4 and §5 below, that there is good reason to accept TT across a whole range of different 19 See Lopes (2014: 169-76). 20 I thank David Davies for bringing this point to my attention.
13 judgements (including those relating to both aesthetic concepts and aesthetic verdicts in Sibley's sense).

Savile
As a final strategy for dissolving the apparent conflict, consider, again, Johnson's (1765) remark that when it comes to artworks 'no other test can be applied than length of duration and continuance of esteem'. Johnson might be taken as appealing here to two importantly different standards: whether a work itself endures, and whether the work is held in continued esteem. Clearly, though, a work may endure in the sense of merely surviving, and even in the sense of being of continued interest to those in the artworld, without receiving continued esteem. Perhaps, then, what really matters for TT is not that a work is consistently judged to be a certain way by critics, but merely that the work itself endures in some relevant sense. Indeed, the most prominent and fully developed contemporary account of the test of time -presented by Savile (1977Savile ( , 1982) -makes precisely this claim. 21 Savile (1977: 195) focuses on the test of time as a method for determining whether a work 'is a truly great work of its genre'. He does not, of course, think we can establish this merely by appeal to a work's duration in a straightforward sense since this is clearly neither necessary nor sufficient for that work passing the test of time . A mediocre work which survives unnoticed in the corner of an attic for centuries does not pass the test and a work can pass the test without itself surviving if, e.g., an accurate reproduction or photograph endures in its place. Instead, Savile appeals to the ability of a work to 'survive in our attention ' (Ibid. 197), which, I take it, requires something like the work's being the continued focus of interest, thought and discussion.
Even with this clarification in place, though, Savile's account is not yet complete since a work could survive in our attention for all the wrong reasons. First, a work may be of 21 It is worth noting that even if we find such an interpretation of TT plausible, it will still be dialectically limited. Since, while it may be compatible with accepting the letter of NT, it is clearly in tension with some of the wider principles used to motivate NT which I discussed in above.
interest for reasons which are completely irrelevant to its status as an artwork. It may, for example, attract attention for purely historical reasons, or because of its great monetary value. Second, a work may be of interest for aesthetic reasons but in a way which does nothing to establish its possession of significant aesthetic value. Savile (1982: 8-9) himself gives the example of someone interested in comparing Dante's work to earlier Italian works written in the vernacular. Dante's own works have clearly passed the test of time but for later critics to appreciate the extent to which these surpass earlier works of the same kind they must also pay continued attention to these lesser works. And this attention, focused as it will be on matters such as the style in which the different works are written, will clearly be aesthetic attention. In response to the first kind of worry, Savile (Ibid. 7) argues that the attention given to a work must be given 'for reasons that bear on its critical estimation as the work it is', and in response to the second that attention must be given to the work autonomously, considering the work for its own sake rather than as a means to better understand and appreciate some other work. This leaves us, then, with Savile's final formulation of TT according to which a work passes the test of time if over a sufficiently long period it survives in our attention under an appropriate interpretation in a sufficiently embedded way. This condition will only be satisfied if the attention that the work is given is of a kind that generates experience relevant to its critical appreciation and attracts the attention that is given to it in its own right.  Yet, even this final formulation is problematic.
Savile maintains that his restriction to autonomous critical attention will successfully exclude cases where aesthetic attention is given to a work for the wrong reasons, but this is not so. These lines, and McGonagall's works more generally, have certainly survived in our attention -new editions of McGonagall's works are still being published more than a century after his death -and the reason for their survival seems to be intimately tied up with concerns relevant to critical appreciation. Yet, the reason in this case does not relate to any great aesthetic achievement, nor to comparisons with any other work, but rather to their spectacular failure as poetic works. And surviving in this way is clearly not evidence for a work's being excellent (indeed, it is evidence in the opposite direction). How, then, can we avoid the conclusion that McGonagall has even a defeasible claim to being numbered among the greats?
The only reliable method for avoiding claims of this kind is one which focuses not merely on the circumstances under which a work receives attention, or on the broad kind of attention which the works receives, but on the evaluative valence of that attention. Dante's works can securely be judged as excellent because they have been regarded as excellent by generations of qualified critics, and McGonagall's poems can be judged just as securely to be doggerel (albeit immensely entertainingly doggerel) for parallel reasons. Even Savile himself seems to concede this point at times. He suggests, for example, that a work which passes the test of time, and so is legitimately judged to be excellent, is one which not only receives continued attention but which is 'widely appreciated over time by those […] whom we recognize as artistically sensitive and deeply concerned with correct judgement and perception in the arts ' (1977: 202). And once we have granted this it seems a small step to extend this to establishing the judgement that a work is execrable (or that it is dainty, drab, or dumpy) based on the collective testimony of those same individuals. It is for this reason, then, that I believe that the test should, strictly speaking, be concerned with whether particular judgements concerning a work endure (rather than with the survival of the work itself). It is a relatively straightforward matter to understand various claims about a work's passing the test as merely a convenient shorthand for claims concerning judgements about that work. By contrast, there seems to be no plausible way of spelling out TT in a manner which is fundamentally concerned with the survival of the works themselves.

Cases of Common Knowledge
I have argued that there is a genuine tension between TT and NT. Merely highlighting this does not, however, tell us which (if either) we should accept. There are, of course, a number of arguments which have been put forward in defence of NT but offering anything like an adequate refutation of these would be far too extensive a task for a single paper. 22 I will not, therefore, rehearse these arguments here (nor the various extant arguments against NT presented by, e.g., Meskin (2004), Laetz (2008) and Robson (2014). Instead, I will focus on presenting some considerations in favour of accepting TT -considerations which will, if the position I have outlined above is correct, also serve as reasons for rejecting NT.
A first motivation for TT concerns what we might call 'aesthetic common knowledge'. There are certain aesthetic claims -concerning the beauty of Shakespeare's sonnets, the excellence of Caravaggio's paintings, and the superiority of Mozart's music to Salieri's -which are generally known amongst educated members of our society. In some cases this knowledge will arise (at least in part) from relevant first-hand experience of the objects in question. I will argue, however, that it need not do so, and that in many cases such knowledge arises merely from learning that the relevant judgements have passed the test of time. In order to demonstrate this, though, I will need to first make some general remarks concerning the debate surrounding aesthetic testimony which often focusses on 'toy' cases such as the following: Exhibition: I have recently been to see an exhibition of works by a new artist with whom we are both previously unfamiliar. I later inform you that the works in the exhibition are uniformly excellent. 23 22 Though see Robson (2015). 23 See e.g., Hopkins (2011: 138), Whiting (forthcoming: 1) and Robson (2015: 756).
Asking whether, on the basis of my say so alone, you can come to know that the works in question are excellent. Those who accept NT will claim that you cannot, and that the details of the case are irrelevant to determining this. By contrast, those of us who reject NT will maintain that the details matter a great deal. It is possible in principle for you to acquire knowledge that the works are excellent in this manner but whether you do so in this particular case will depend on a number of factors: my capabilities as an aesthetic judge, my track record for honesty, and so forth. 24 Yet, whatever we end up saying about such cases, it doesn't strike me that the answer to the knowledge question is obvious. I think that the pessimist is wrong that you cannot, as a matter of principle, achieve knowledge in such cases, but I can certainly feel the force of the pessimistic intuition concerning them. 25 Focusing on cases of this kind, though, can tend to stack the deck in favour of the pessimist, as we can see by discussing some rather different examples.
Consider, by contrast, the case of someone who has never encountered Shakespeare's plays or Beethoven's music for themselves. Would it really be plausible to suppose that such an individual doesn't know anything about the aesthetic qualities of these works? Unlike the first case there seems to be only one plausible answer to this question: such an individual would -presuming they are aware of the enduring reputation of these works -be in possession of at least some knowledge regarding the aesthetic character of these works.
Indeed Jerrold Levinson (2005: 213) goes so far as to classify the judgement that 'the Adagio of Beethoven's Third Symphony' possesses certain aesthetic properties 'on the basis of centuries of testimony' as a paradigmatic example of a legitimate second-hand judgement. 26 24 It is controversial precisely what is required for testimonial knowledge in even the most mundane circumstances. See, e.g., Burge (1993), Faulkner (2011), Fricker (1994, and Lackey (2008). 25 See Robson (2015). 26 The particular properties which Levinson highlights are not straightforwardly evaluative ones but Levinson's arguments are clearly intended to also apply with respect to judgements concerning evaluative aesthetic properties. I take it that anti-pessimistic intuitions concerning such cases are powerful and widely shared, but there will, doubtless, be some who remain unconvinced. Even setting aside general worries concerning the evidential value of intuitions, there will be those who either do not share these intuitions or who remain confident that the pessimist can explain them away. Further, even those who do take these intuitions to provide evidence in favour of TT may still judge that this is outweighed by the stronger intuitive evidence which cases such as Exhibition provide in favour of NT. In order to avoid such am impasse, then, I will move (in §5 below) beyond mere appeal to intuition and offer some more theoretically robust reasons for accepting TT. Before doing so, though, I will consider some more substantive responses which a defender of NT may offer to my appeal to aesthetic common knowledge.

Rejecting the Appeal to Common Knowledge
First, it might be suggested that the success of the appeal I have presented here is dependent on my earlier decision to spell out NT in narrowly epistemic / cognitivist terms. This dependence is, however, merely superficial. Those pessimists who appeal to non-epistemic norms can, of course, freely concede that the judgements I have discussed above count as knowledge without compromising their position. Such a move will, however, give them no dialectical advantage here since they will still be committed (qua pessimists) to denying that the judgements in question are legitimate. Similar claims also apply with respect to the expressivist defender of NT. Those expressivists who wish to maintain, following e.g. Blackburn (1980), that aesthetic judgements can literally attain the status of knowledge will, of course, encounter precisely the objection I have outlined above. Even those who don't make such a claim, though, will still need to accept that certain aesthetic judgements are legitimate in a way in which others are not. 27 If they place the judgements I have discussed above in the former category, then this will mean abandoning NT; if they place them in the latter, this will be highly counterintuitive for reasons paralleling those outlined above.
A second concern is whether such putative cases of aesthetic common knowledge provide any reason for rejecting NT not already provided by more prosaic cases. Some might worry that whatever reasons we have for judging that the test of time can serve as a legitimate source of aesthetic judgment will be derivative of our reasons for believing that the testimony of particular critics can do so. As such, the resolution of these debates will ultimately depend on the much more familiar debate concerning whether we can legitimately form aesthetic judgements on the basis of testimony from individual critics. 28 After all, the hypothetical objector claims, if we deny that the testimony of any individual critic carries any weight when it comes to the formation of aesthetic judgement, then it is difficult to see how the combined judgements of many such critics could do so. On the other hand, if we do accept that the judgements of individual critics carry such weight then we have already rejected NT and any appeal to TT will be redundant. Either way, an appeal to the test of time does nothing to advance the debate concerning NT.
There is some truth to both horns of this dilemma, but each of them, ultimately, misrepresents the current state of the dialectic. Taking the second horn first, I am more than happy to admit that the case of individual critics should be enough to settle matters in favour of the pessimist's opponent (the optimist concerning aesthetic testimony). Still, as a matter of descriptive fact, this view is not widely shared and many people's intuitions concerning cases of testimony from individual critics appear to favour pessimism. By contrast, appeals to the test of time tend (as with appeals to testimony concerning natural beauty or 'lost works') to elicit optimistic intuitions. 29 Yet, I do not take the difference here to be merely a descriptive one and while testimony from individual critics does provides good reason for accepting optimism, appealing to TT provides us with an even stronger case. 28 I thank an anonymous referee from the journal for pushing me to consider this worry. 29 See, e.g., Robson (2012: 5) and Laetz (2008: 255).

20
Consider first that, contrary to what the first horn of the dilemma presupposes, those who accept NT need not deny that the testimony of individual critics carries some weight.
Pessimists regularly concede that the testimony of individual critics can motivate seeking to experience a work for myself (Gorodeisky 2010: 59-60), give me grounds for reconsidering my own view (Hopkins 2001: 168-9), and provide some limited degree of warrant for forming the relevant belief (Hopkins 2000: 219). The crucial aspect of their claim is merely that such testimony cannot, by itself, provide me with legitimate grounds for forming an aesthetic judgement. Yet, even if someone were to judge, falsely in my view, that the testimony of each individual critic (considered in isolation) carried no weight this wouldn't entail that the combined testimony of centuries of such critics was similarly impotent. There may, for example, be some potential defeater which would undermine any particular critic's testimony, but which would be unable to similarly undermine a cross-temporal critical consensus. Of course, merely presenting this possibility in such an abstract fashion is little more than a promissory note which is unlikely to placate anyone. I will, however, argue (in §5.2 below) that there are a number of potential defeaters for the testimony of any individual critic which do not apply to judgements based on TT.

Two Sources of Unreliability
My second reason for maintaining that TT plays a pivotal role within aesthetic epistemology concerns the fragile and problematic nature of many of our first-hand aesthetic judgements.
In particular, I will focus on two kinds of worry concerning the reliability of our aesthetic judgements. The first of these involves some now standard considerations highlighted in Hume's own discussion of TT; prejudice, passing fashion and the like. The second concerns recent empirical evidence which shows that our first-hand judgements are also unreliable in some more surprising ways. I will argue that second-hand aesthetic judgements which have passed the test of time will tend to be less susceptible to both kinds of distorting factor than first-hand aesthetic judgements.
The idea that aesthetic judgements are uniquely, or at least unusually, susceptible to certain kinds of distorting factor is hardly a new one. Hume (1757Hume ( / 1875 mentions the propensity of '[a]uthority or prejudice' to give 'a temporary vogue to a bad poet or orator' and opines the rareness of the true judge who is 'free from all prejudice' allowing 'nothing to enter into his consideration but the very object which is submitted to his examination' (Ibid 263). Nor are these worries unique to Hume. Indeed, the claim that our aesthetic judgements can be problematically susceptible to such factors is hardly controversial To see how the test of time can help to correct such distorting factors consider Sibley's (1968: 50-51) claim that the possibility of error with a case that has elicited long-lasting convergence decreases as possible explanations of error become more obviously absurd; e.g. we could not sensibly reject a centuries-spanning consensus about Oedipus as being the result of personal bias, enthusiasm for a novel style, or passing fashions or fads. I do not mean that, in other cases, there is always some reason for doubt; only that the long-attested cases may virtually exclude the theoretical sceptic's doubt as absurd.
Why is it that the sceptic's doubts in such a case border on absurdity? Personal biases come in various kinds and it is, perhaps, overly optimistic to think that these are entirely eliminated when it comes to Oedipus. What is clear, though, is that many of the most straightforward kinds of bias, which may have influenced spectators in ancient Athens, have no application to those of us in later centuries. None of us regard Sophocles as either a friend or enemy, nor do we have any personal stake in his success or failure. Similarly, any enticing novelty which his plays once enjoyed has long since subsided and passing fads are, well, passing (and a critical consensus which has endured for over two millennia hardly qualifies). 30 This is not, of course, to suggest that all first-hand judgements are unreliable. I am perfectly happy to concede (as Sibley) that there are cases where we are able to eliminate 22 such possibilities even when the judgements in question have not passed the test of time.
What I do want to maintain, though, is that while these doubts are standardly eliminated with respect to judgements which pass this test they are much more pervasive with respect to first-hand aesthetic judgement. Nor is this kind of defence of TT limited to the sources of error which Sibley lists.

Empirical Evidence of Unreliability
Recent empirical work in aesthetics appears to show that our aesthetic judgements are often unduly influenced by irrelevant factors such as ordering effects, belief polarization, and confabulation. Consider, for example, the following studies which are representative of a much wider body of research. 31 A study by McLaughlin et al. (1983) found that '[a]esthetic preferences for asymmetric pictures' were partially 'determined by the direction of cerebral asymmetry'. That is, that those who were right-handed tended to find 'paintings whose areas of visual interest are primarily in the right portion of the painting were "more aesthetically pleasing" while left-handed subjects preferred visual interest to the left' (Irvin 2014: 39). Another (by Ginsburgh & Ours (2003)) revealed that 'a critical determinant of success' in musical contests 'is the order in which musicians perform' (289) even in cases where this order is randomly assigned. Finally, a pair of studies carried out by Ayumi Yamada (2009) demonstrated that subjects' preferences between representational and abstract paintings varied depending on which aspects of their evaluations they were asked to verbalise. Yamada found that 'when participants attempt to describe their reasons for liking each painting […] they will be more likely to prefer the representational painting over the abstract one' but 'when asked to describe their reasons for disliking' each they would be 'more likely to dislike the representational painting' (1141).
Again, we can see how appealing to TT can help here. While our first-hand aesthetic judgements are often unduly influenced by such irrelevant situational factors it seems highly implausible to think that a centuries-spanning consensus about a work will be. 31 See, e.g., Irvin (2014), Kieran (2011) and Lopes (2014). Different critics in different eras will encounter works in an incredibly diverse range of circumstances and, as such, the longer a critical consensus lasts the less likely it becomes that such irrelevant aspects of the particular situation in which a work is viewed play any significant role in its formation. As such, we once again have good reason to hold that aesthetic judgements which pass the test of time will typically be more secure than standard instances of first-hand aesthetic judgement.
Not only that but we can also see how, mutatis mutandis, they will typically be more secure than second-hand judgements formed on the basis of testimony from individual critics. Despite their expertise many critics will still be influenced by prejudice, passing fashion and the like. Similarly, there is good reason to believe that even expert critics will still be susceptible to the kinds of situational factors which recent empirical work has highlighted. 32 For this reason, then, we can see why the first horn of the dilemma I consider in §4.2 is misguided. With respect to particular individual critics there will often be a broad range of potential defeaters for accepting their testimony concerning the aesthetic properties of some work: perhaps they are influenced by some passing fad, or subject to ordering effects, or… Yet, such defeaters do not typically arise with respect to the cross-temporal consensus of such critics represented by TT. Given this, it would be perfectly consistent for someone to maintain that the testimony of individual critics carries no (or negligible) epistemic weight but that a centuries-spanning critical consensus is enough to provide us with aesthetic knowledge. Of course, I don't personally take our epistemic position with respect to the testimony of individual critics to be anything like this bleak. Potential defeaters need not, after all, be actual defeaters. Further, even when such defeaters are present we will sometimes have good reason to judge that they are themselves defeated in particular cases. 32 Indeed the research by Ginsburgh & Ours (2003) discussed above focused on theinfluence of ordering effects on the expert judges for the prestigious Queen Elisabeth music competition. There is, however, (as, e.g., Irvin 2014 discusses) some reason to believe that certain distorting factors will, while still present, be less pronounced with respect to experts. I have argued, then, that we have good reason to hold that aesthetic judgements formed on the basis of a cross-temporal consensus of critical testimony are typically more secure than either first-hand judgements or those formed on the basis of testimony from a single critic.
This claim would, however, only hold if we assume that the critics involved possess a sufficient degree of independence from each other; 33 something which would not be so if, for example, later critics treated earlier critics as what Goldman (2001: 101) terms 'gurus' such that '[w]hatever the guru believes is slavishly believed by his followers. They fix their opinions wholly and exclusively on the basis of their leader's views.' If this were the case, and later critics merely unreflectively aped the judgements of earlier critics, then their agreement would provide no additional warrant for accepting these judgements.
Fortunately, though, such an extreme positon is implausible, and the tendency of criticsboth synchronically and diachronically -to disagree with each other is well documented. 34 It would, however, be similarly implausible to maintain that critics are never influenced by the judgements of their predecessors, since there is -as I discuss elsewhere (Robson 2014(Robson : 2523(Robson -2524 -clear empirical evidence which shows that they frequently are. Yet, this concession does not undermine TT for two reasons. First, given the truth of optimism, some degree of deference to the aesthetic judgements of others (particularly expert critics) is precisely what we would expect from an epistemically responsible agent.
Second, and more fundamentally, the degree of deference which does exist is clearly compatible with a significant degree of independence between critics. In particular, it is important to stress just how rare it is for a work to achieve anything like the kind of crosstemporal critical consensus required by TT. Many works praised to the heavens by contemporary critics meet with corresponding levels of opprobrium in later generations.
Others eventually receive critical acclaim after long periods of neglect or disdain. Indeed, it is precisely this inability to predict which works from our own era will go on to receive 33 I thank Wlodek Rabinowicz for pushing me to address this point. 34 See, e.g., Ross (forthcoming). continued esteem which is often used to motivate appeals to the test of time. 35 Given this, it seems safe to suggest that the judgements of later critics are not systematically influenced in problematic ways by those of critics in earlier generations. This does not, however, show that their judgements aren't problematically influenced in other ways. For example, it has recently been claimed that, for a variety of reasons, works are included in, or excluded from, the artistic canon on the basis of features which have no relevance to their aesthetic standing.

A Worry Concerning Canon Formation
Worries of the second kind are relatively easily dealt with. It is doubtless the case, for example, that (as discussed in, e.g., Eaton (2008: 878-9)) a number of works have been excluded from the canon -and that various positive judgements concerning them have failed to pass TT -owing in large part to aesthetically irrelevant factors such as the gender, race or class of the artist. However, this raises no problems for my defence of TT since I do not take the failure of a work to pass the test to be in any way indicative of that work's lacking aesthetic value. Of course, the overarching concern here is not merely with works being excluded from the canon, since parallel worries arise with respect to those works which are admitted. Again, though, such worries aren't particularly bad news for the defender of TT. We can grant both that artists from certain privileged social groups -or artworks which, as Smith (1988: 51) puts it, 'reflect and reinforce establishment ideologies'are more likely to achieve canonical status and that those works that do attain it are very likely to possess significant aesthetic merit. After all, the vast majority of artworks produced by those in positions of great privilege who seek to reinforce establishment ideologies still fail to attain canonicity. This is not, of course, to deny that there is anything wrong with current methods of canon formation. The exclusion of various social groups from the artistic canon is clearly an issue which should be of great concern (from both a moral and aesthetic 35 Ironically, Hume's own (1757 / 1875: 235) example of the obvious superiority of Addison's work to Bunyan's provides an apt example here since the majority of modern critics are of the opposite opinion. point of view). It is not, however, one which undermines the particular epistemic claim (TT) which I am arguing for in this paper.
A more worrisome objection arises from James Cutting's (2003) celebrated experiments which demonstrated that his students' preference for particular impressionist paintings increased the more they were exposed (often without conscious awareness) to those paintings. That is that mere repeated exposure to a painting was enough to increase their positive evaluation of it. This so called 'mere exposure effect' has subsequently been replicated across a range of art forms and with respect to a number of different measures, leading many psychologists to endorse Cutting's (2003: 335) claim that artistic canons are promoted and maintained, in part, by a diffuse but continual broadcast of their images to the public by museums, authors, and publishers […making…] mere exposure a prime vehicle for canon maintenance. Tacitly and incrementally over time, this broadcast teaches the public to like the images, to prefer them, eventually to recognize them as part of the canon, and to want to see them again.
In turn, it seems likely that this implicit education also reinforces the choices made by professionals in what they present to that public.
If this is correct, though, and mere exposure plays such a significant role in determining which works are admitted into the canon, doesn't this undermine TT? If a particular work is only, or primarily, approved of by centuries of critics owing to their repeated exposure to the work then this is surely no grounds for judging that the work is genuinely excellent.
To begin on a concessive note, it is important to stress that I am in no way claiming that the test of time is infallible. As such, even if there are some cases where a judgement concerning a work passes the test for reasons irrelevant to its aesthetic value this will not necessarily undermine my defence of TT. Of course, such a response will only succeed if these cases are sufficiently rare, which leads us to the less concessive parts of my response.
First, note that our tendency to be exposed to particular artworks is not neutral with respect to the value of such works. Consider, for example, that the decision of a gallery or museum to display a work (and likewise to sell prints and postcards of that work, to use images of the work in their publicity material, and so forth) typically -though not universally, Massachusetts' Museum of Bad Art being one notable exception -involves a tacit endorsement of the quality of that work. As such, moving (albeit unconsciously) from increased exposure to a work to the judgement that the work in question actually possesses greater aesthetic value may be a useful -though far from infallible -heuristic. Further, there is reason to think that the mere exposure effect is rather more discriminating than has typically been claimed. An intriguing recent study by Meskin et al (2013) appears to show that while mere exposure increased liking with respect to high quality works (such as the impressionist paintings Cutting used in his studies) it actually decreased liking with respect to works of poorer quality (such those of Thomas Kinkade). Leading them to conclude that with respect to canon formation mere exposure cannot be the full story; frequent and repeated presentation (or representation) of artworks does not look as if it will ensure that they are in the canon, since mere exposure to bad paintings such as Kinkade's decreases liking for them.
For these reasons, then, it looks as if we should not be unduly concerned about the influence of mere exposure in canon formation.
For those unconvinced by what I have said above, the final aspect of my response to the canon formation worry takes the form of a tu quoque. If we assume, as I think we should not, that mere exposure plays an illegitimate and highly pervasive role in canon formation then this will certainly generate some serious sceptical worries for defenders of TT. Such worries are not, however, unique to aesthetic judgements formed by applying the test of time (or on the basis of testimony more broadly). Rather they are, as Cutting's experiments show, equally applicable with respect to our first-hand aesthetic judgements. Given this they cannot be used by the defender of NT to show that the test of time is a less effective source of aesthetic judgment than first-hand acquaintance with the relevant works. With respect to mere exposure, the two sources would be on a par whereas, as I have demonstrated above, there are a number of other respects in which the test of time clearly serves as a superior source of aesthetic judgement. I have shown, then, not only that there is a genuine tension between NT and TT but, also, that we have good reason to accept the latter and, therefore, reason to reject the former.
Given this, it becomes increasingly important to consider how precisely we should formulate the test of time and how best to apply it. This is, however, a task for another time. 36