DRAFT - Work in Progress

The Metaphysics of Looks

Matthew McGrath

University of Missouri

It is commonplace to discuss the looks of things. Consider:

1) Eeyore has a sad look.

2) From the look of the tree, it is healthy.

3) The sun has a reddish look this evening.

4) This year’s model has a different look than last year’s.

When asserted of the right objects in the right contexts, these statements seem to state easily verified facts. The facts stated, moreover, might be ones we care about, for their information value in some cases and for purely aesthetic value in others. This paper asks what looksare, and what it is to have a look.It seeks to make progress on the metaphysics of looks.

There is a tendency among philosophers to speak, loosely, of looks and other appearances as subjective, as somehow“in the mind” rather than “in the world,” or better as mind-dependent rather than independent. There are varieties of mind-dependence, though. On a simple modal understanding, a thing’s having a look is mind-dependent justhad a thing not been perceived, it would not have a look. But it is not hard to see that this claimclashes with what seem to be humdrum facts about looks. Eeyore continues to have his sad look even when no one sees him. The sunwould not cease to have a reddish look at sunset on a certain evening if should turn out that no subject perceived it that evening. Should the entirety of perceivers vanish, things would still have looks. But philosophers who think of conceive of the having of looks as mind-dependent probably have in mind a subtler sort of dependence – of explainability in terms ofthe mental, or at least in terms offacts or phenomena that constitutively involve mentality. The having of looks might be mind-dependent in this sense even if not modally mind-dependent. One promising avenue for this middle ground between full-blown objectivism and a simple modal subjectivism is to conceive of looks as dispositions to affect perceivers in certain ways. The having of looks is thereby explained in terms of the mental (in terms of perceivers being affected in certain ways), without commitment to thinking that if a thing should go unperceived it would cease to have a look. One ofthis paper’s main concerns will be to determine whether this sort of dispositionalismabout looks is correct.

One doubt needs addressing from the outset, however. Don’t statements like 1-4 commit the mistake of reifying strange entities into existence? If so, there is no need for a metaphysics of looksanymore than there is a need for a metaphysics of sakes?

Looks are closely related to things looking certain ways, just as smiles are closely related to ways of smiling, accents to ways of talking, and habits to ways of acting, and so on.Consider my coffee cup. It looks a certain way W. This is all it takes for it to have a certain look L. It has the look L because it looks way W. Compare smiles. You have a certain smile (on an occasion, let’s suppose). All it takes for you to have that smile is for you to smile in a certain way. Once we have ways of smiling, it seems there is no reason to balk at smiles. Similarly, once we have ways of looking, there is no reason to balk at looks. I don’t deny there are important questions about how to understand “all it takes” here. We might take smiles to be ways of smiling. Alternatively, we might take smiles to be distinct from ways of smiling but for their possession to be groundedby one smiling a certain way. The same goes for looks. I will not attempt to resolve the question of whether the identity or grounding view is to be preferred. What I hope is plausible is this: to the extent that one is inclined to accept that things look certain ways – or that people smile certain ways – one ought to be inclined to accept that things have looks – that people have smiles. I will not try to refute nominalist theories that deny that things can look, smile, walk, talk, etc. in certain ways. But there is nothing especially mysterious about things looking certain ways; and thus, I want to suggest, given that things have looks because they look certain ways, there is nothing mysterious about things having looks.[1]

The plan for the paper is as follows. In section 1, we will examine in some detail different sorts of looks and different ways a thing might be said to have a look. In section 2, we examine two prima facie attractive objectivist proposals concerning looks: (i) Michael Martin’s parsimonious account on which looks are complexes of ordinary visible properties such as shape, color and size, and (ii), an account inspired by Frank Jackson’s writings on which looks include not only such Martin-style properties but also other situational features, including illumination, distance, angle of view, etc. I will argue neither of these accounts is acceptable. These conclusions might tempt one to subjectivistproposals, or proposals under which looks are mind-dependent in the explanatory sense described above. In section 3, the heart of the paper, I argue against subjectivism, especially against its most plausible form, dispositionalism. Finally, in sections 4, I lay my cards on the table, sketching an alternativeobjectivist view of looks.

1. Kinds of looks

Looks are plausibly categorized as properties. Two things can have the same look. For instance, two coke cans might have the same look. Looks thus seem at least to be repeatables, had by distinct things over time. Moreover, looks do not seem to be repeatables in the way abstract continuants like pieces of music or dances are. There is no copying or tracing condition that must be satisfied for things to have the same look. Two rocks in different galaxies may have the same look. Moreover, there can be looks that some things had in the past but which nothing has now, as well as looks that things might have had but never in fact do have. Certain now extinct species had looks which nothing now has (assuming no replicas of them exist), and had evolution taken a different path, there would have been creatures with looks which nothing in fact now or ever will have. Looks would seem to be properties that ordinary objects like rocks can possess.[2]

All looks, then, are properties. But we can distinguish, within the class of looks, different sorts of properties.

There is a class of looks that a thing has only relative to a viewpoint. At 9pm EST on a day in July the sun may well have a reddish look with respect to viewpoints in Maine while having a bright white look with respect to viewpoints in Seattle. It’s not that the sun has both a reddish look simpliciter and a bright white look simpliciter. It has each look but relative to different viewpoints. Similarly, a coin has one look relative to a viewpoint on the normal to the coin’s surface and a different look with respect to a viewpoint on a line which is at an angle to the surface.[3]

Viewpoint-relative looks are also time-relative. From the same viewpoint, my wall has one look in the morning, a slightly bluish one, and different look in late afternoon, a yellowish one. Again, it’s not that my wall has either look simpliciter (even relative to a viewpoint). It has one look from viewpoint V at t1 and another look from the same viewpoint at t2. Thus, the class of looks under discussion are both time- and viewpoint-relative.

Should we also say these looks are illumination relative? No. Fixing a time and a viewpoint is enough to fix a look of the sort we are discussing. They are not relative to illumination, only influenced by or “dependent” on it. The same goes for other factors such as medium and surrounding scene. If a distorting lensis placed between the thing and the viewpoint, its look from the viewpoint changes.If a light gray patch is placed against an all-white background, it has a lighter look. Following Schellenberg (2008), I will call factors like thesethat affect looks without affecting a thing’s intrinsic character situational parameters. The class of looks we have been discussing are viewpoint and time relative, and they are situation-dependent but not situation-relative.I will callthese looksviewpoint-relative looks.

We often talk of having looksin ways that abstract from the thing’s situation at a time in favor of the looks it would have indifferent situations. Consider my wall again. Suppose I’m at a friend’s house discussing my wall. It’s well past sundown and so I knowthe wall is shrouded in darkness at the time. Yet I might say to my friend, “my wall has a bright, clean look.”I don’t say “it has no look at all” or “a dark look” although I could truly say such things with some stage-setting (“of course, now it’s in the dark, so…”).On other occasions, I might disregard the current medium in favor of some other medium; or I might disregard the current surrounding scene in favor of some other scene. Thus, I might say, “this piece of paper of course looks white, except not now” when viewing something through a red filter, or I might say, “The patch looks dark gray, except against such a dark background as this.”

There are parallels here with our talk in other domains. In a conversation about a friend, Imight sayshe “has a great smile” even though I know shelives in the UK and is now asleep. Ifyou say of someonethat he “has a strong Italian accent,” Ican’t prove you wrong if I point out that he happens to be playacting at the time across town and speaking without that accent. When we say these things, we are using ‘has’ to indicate what is the case in normal or conversationally relevant sorts of situation. Just as we shouldn’t postulate special kinds of smiles and accents that remain present when a person isn’t smiling or speaking with that accent to explain how these statements could be true, so we shouldn’t postulate a special kind of looks that remain present in the dark. Instead, we are speaking of the same sorts of smiles/accents/looksand saying a person is disposed to have them occurrently. Once we have an account of viewpoint-relative looks, then, we have the materials we need for explaining the truth of claims about “having” them, whether the having is dispositional or occurrent. To keep things simple, unless otherwise noted I will use ‘have’ in an occurrent sense, on which to say that “x has reddish look” is false if x does not there and then does not look reddish.

So far, then, Ihave discussed only one kind of look, viewpoint-relative looks, and I have considered different ways a thing might be said to “have” such looks. However, arguablythere are looks beyond the viewpoint-relative variety. One way to argue this is as follows. To vary an example of Michael Martin’s, consider Michelangelo’s David. If a museum visitor only looked at it from one viewpoint and then turned to leave the gallery, one might say to him, “seeing it only from one viewpoint, you can’t fully appreciate what it looks like.” The thought expressed is that the statue has a look that one cannot appreciate without taking up multiple viewpoints with respect to it. Such a look would not be a viewpoint-relative look. One might immediately think that such looks are something like collections of viewpoint-relative looks. But this is not the only option, as we will see in later sections. For the moment, I will merely note the plausibility of believing of such looks – what Brian O’Shaughnessy (1990, 138) calls these“looks in the round,”by analogy with theater in the round.

Just as there are more or less determinate ways of looking, so there are more or less determinate looks things can have (whether the looks in question are viewpoint-relative or looks in the round). There is a certain determinable Mason and Hamlin grand piano look that different Mason and Hamlin grand pianos share, but it is less specific than the overall look of a particular normal Mason and Hamlin grand. These pianos, moreover, share an even more determinable look – the grand piano look – with other grand pianos. Then there is the still more determinable piano look shared by uprights as well as grands. And so on. A particular piano might have all these looks. We can define a thing’s maximal look as the look its having of which implies its having each of its other looks and its having of which is not implied by its possession of any proper subset of its other looks.

Any acceptable account ought to explain what viewpoint-relative looks are and how they relate to looks in the round. And, when supplemented with a reasonable semantics, it ought to have the resources to explain how various humdrum statements about looks are true. In the next two sections, I will mostly concentrate on viewpoint-relative looks. In section 2, I will argue against two otherwise appealing objectivist accounts. In section 3, I will argue that subjectivist accounts fail poorly as well. This will pave the way for an alternative sort of objectivist account, explored in section 4. At the end of section, I briefly discuss the question of how to understand looks in the round.

2. Two objectivist accounts.

It would be an advantage in a theory of looks if we could see how looks fit into our current ontology. Here is Martin:

“Various recent discussions of appearances in general, and looks in particular, have been moved by concerns with conflicting appearances to posit a range of properties in the world as the looks of things over and above the other properties that we are committed to supposing objects have, such as their shapes and colours. According to these accounts we need to recognize such things as perspectival shapes (‘oval from this angle’, for example) and apparent colours to play the role of the looks of things through which we come to experience their colours and shapes. In contrast, the parsimonious view of looks that I will sketch proposes that we identify the looks of objects with their basic visible properties, including their colours and shapes. Accommodating within our ontology the looks of objects does not require that we posit additional features of these objects over and above those properties we are otherwise committed to supposing them to have through what we can know of them through perception.”

It this section, we examine two non-reductionistviews that meet Martin’s strictures: Martin’s own parsimonious account and a second account, inspired by remarks of Frank Jackson, that supplements Martin-style looks by adding situational features.

2.1. Martin’s view

Looks, for Martin, are complexes of visible properties such as shapes, sizes, textures, and colors.[4]They are thus complexes of familiar properties we already accept. However, if this is what looks are, no looks are viewpoint-relative. For these visible properties are not viewpoint-relative (or even situation-dependent). In fact, as I read him, Martin denies there are viewpoint-relative looks at all, only “overall” looks, i.e., looks in the round.Still, he must explain how statementsostensibly about viewpoint-relative looks could be true. Consider these: “my wall has a different look in the evening than it did in the morning” and “the coin looks different from this viewpoint than it does from that one.” These are true. What could account for their truth?

Martin’s main resource for giving these explanations is what he calls aminimalistsemantics. Whereas Frank Jackson’s(1977) semantics of statements of the form ‘x looks F’ appeals to sense data,and Chisholm’ssemanticsappeals to ways sensing, Martin claims his is minimalist insofar as it is ontologically neutral. It appeals only to looks and similarities between them. Generally, the content of a statement of the form ‘x looks F’, according to Martin, is a proposition to the effect that x has a look which is similar to the look of a certain class of Fs. Martin applies this account to all statements of this form, including ones ostensibly about viewpoint-relative looks. We should be clear that, being ontologically neutral, Martin’s semantics does not presuppose his parsimonious view of looks. His claim is that the minimalist semantics, together with the parsimonious view, accommodates the truth of true looks-statements.

Martin considers the familiar example of the straight stick half-submerged in water. “The stick looks bent” is a good example of a statement ostensibly about a viewpoint-relative look. Ontheminimalist semantics, this is true if and only if:

s (Has (the stick, s)  (Look(s)  SIM(C(Bent, Look, k), (s))). (Martin, 197)

where ‘SIM’ expresses similarity, ‘C’ denotes the “getting-the-characteristic” function, and ‘k’ is a contextual restriction, here giving us the set of sticks. A getting-the-characteristic function takes a property (e.g., bentness), a respect of similarity (e.g., in look) and a restriction and returns a look characteristic of things in the restricted set with that property. In most contexts, C will return a paradigm look associated with the property within the relevant class k. Thus, C returns the paradigm look of bent sticks. What is expressed by the statement, then, is the proposition that there is similarity between the look of the stick andthe paradigm look of bent sticks. Martin observes that usuallysuch a statement conveys, even if it doesn’t express, something beyond the fact of similarity; it will convey the fact that the stick’s look is similar to a particular look L(the one in fact characteristic of bent sticks).