 |
| Response-dependence Theories |
The term response-dependent was introduced by Mark Johnston (1989) for concepts, such as red, that support an a priori biconditional on roughly the following lines: "X is red if, and only if, X is such that it would look red under normal conditions."
Any concept of the intended kind will apply to something just in case the object has a property in virtue of which it would elicit a relevant response, on a par with the appearance of redness, under relevantly favorable conditions; it will be akin to the concept of a secondary quality, traditionally conceived.
The response to be elicited will involve a cognitive impression, so that the object looks, seems, or presents itself in a certain manner. And the conditions under which that response is guaranteed will have to be capable of independent specification; they cannot be defined just as whatever conditions will provide the guarantee.
 |
| cognitive impression |
Response-dependent concepts in this sense are meant to contrast with response-independent concepts whose application to an object depends solely on the nature of that thing in itself, not on the cognitive impression that the object makes on human beings.
As Crispin Wright (1992) has emphasized in ongoing reflections around the theme, there must be a sense in which the object is of the conceptualized kind because it elicits that response, and not (or not just) the other way around; there must be a sense in which an object is red because it normally looks red.
The interest of the notion of response-dependence lies in the prospect of illuminating the character of a variety of concepts: for example, concepts of an evaluative, affective, or aesthetic kind; concepts associated with practices such as praise and blame or intervening causally in the world; concepts that are anthropocentric in any such manner; or perhaps all concepts that are mastered ostensively, without reliance on prior definition.
 |
| illuminating the character |
There are two very different theories of responsede-pendence in the literature. The biconditional associated with response-dependence, so all sides assume, does not hold because people's relevant cognitive impressions never miss or misrepresent anything.
So what makes certain concepts response-dependent, assuming that some concepts are indeed of this kind? What underpins the truth of the biconditional that governs them? The two theories diverge on that question.
One theory, explored by Johnston himself, would say that certain concepts are response-dependent because the properties they designate are dispositions in things to evoke the relevant responses.
 |
| higher-order property |
Under this account we use a term like red to apply to those things that are such as to look red in suitable conditions; we think of the property of redness as the higher-order property of things that have a lower-order property, maybe this, maybe that, which makes them look red in suitable conditions.
According to this theory, the concepts are responsedependent because the properties are defined by reference to responses; the a priori biconditionals hold, because they reflect the character of the properties conceptualized.
This theory has the disadvantage that, as Johnston (1993, 1998) himself has argued, few of our concepts are response-dispositional in this sense. With concepts such as red, we want to say that something looks red because it is red, where this is a causal explanation.
 |
| causal explanation |
But it is not clear that that claim remains available if redness is construed as a disposition; looking red will be a manifestation of the disposition, not a contingent effect. The issue has been a focus of controversy (Menzines and Petit 1993, Miller 2001).
The alternative theory of response-dependence would avoid this difficulty (Jackson and Petit 2002, Petit 2002). While allowing that there may be response-dispositional concepts, it says that other concepts may be governed by an a priori biconditional, too (or, being partly response-dependent, by at least an "only if " conditional).
That will not be because they are paired with anthropocentric dispositions, but because the explanation of why they are paired with their particular, response-independent referents is that those properties have certain anthropocentric effects. On this account red may refer, not to the disposition to look red, but to a perfectly physical property, such as a certain profile of surface spectral reflectances.
 |
| anthropocentric dispositions |
The reason why it will refer to that property is that it is the one that elicits the appearances on which speakers rely in learning to use the term. And so a connection will remain in place between the presence of the property and the looks-red response.
Responsedependence will become salient, not at the level of semantics where we pair off terms with items in the world but, to invoke a distinction made by Robert Stalnaker (2004), at the level of meta-semantics where we try to explain the pairings that obtain between words and world.
Suppose that people generally rely on appearances in using the term red. Suppose that they intend to refer to a common, objective property in using the term; they are not content to go their idiosyncratic ways. And suppose that because of that intention they seek to coordinate their usage, discounting some of the appearances of redness on which they divide.
 |
| objective significance |
If this enterprise of coordination is to have objective significance, then there must be an objective reason for speakers to discount some appearances and not others; equivalently, there must be an objective reason to treat certain factors as perturbing or limiting influences on appearances.
Why should speakers indict some influences as perturbing or limiting, then, but not others? According to this approach, it will be right to indict factors such that by privileging situations of usage where they are absent—by treating those conditions as favorable—speakers can optimally satisfy their joint intention to pick out the same objective property in things (Pettit 2002); the associated practice will do best in helping them to triangulate on a common, presumptively objective feature.
This approach will make it a priori, for any property such as redness that is available to be named in our language, that X counts as red if, and only if, it is such that it would look red under favorable conditions, with favorable defined by reference to the practice of discounting.
 |
| joint intention |
The basis of the response-dependence will now lie, not in the nature of the property, but in the requirements that must be fulfilled for the property to deserve the name of red; that is why the biconditional is restricted to properties available to be named.
This theory of response-dependence allows us to say that while things may be conceptualized as red because of the associated appearances, still their redness is causally responsible for such appearances. It can mark out certain concepts as special, on a par with the concept of redness, particularly if the concepts are ineradicably responsedependent.
And yet it can allow us at the same time to be realists about redness and similar properties, even holding that a term like red refers rigidly to an actual-world property (Haukioja 2001). If one wants to espouse response-dependence without too deep a revision of common sense, this is the way to go. If one aspires to be revisionary, the other theory of response-dependence is the better option.
 |
| common sense |
