3. Three Paths Out of the Naive Frame
3.3 The Conceptual Design Method
What I am calling the Conceptual Design Method approaches concepts from within the broad teleosemantic model. It begins by thinking about shared communicable concepts as things that can exist because of a set of shared purposes and norms by which we regulate each other’s use of the concept. The concept is a representational tool and what it manages to count as representing in our community is a function of the
representational purposes it was introduced to serve modified by the communities
evolving understanding of that shared purpose. For example, our understanding of the purpose that the concept water plays in picking out the natural kind has shifted over time as our understanding of what kind of thing water is has changed, this shift is possible because members of the community can recognize the concept as playing essentially the same core representational role throughout the change in our theory about the world. In fact, it is core to certain concepts like names and natural kind terms that we treat the concepts nearly like ostension concepts. Water is “that thing” whatever it is. In so doing we try to ensure that our concept is about a real thing in the world even when our
understanding of that thing is deeply flawed.6 This core purpose is reflected in the norms of revision we maintain for the concept. Through discoveries we revise our associations and individual conception of the concept, but we treat other members of our community and even past people as talking about the same thing.
Many concepts, however, are not regulated by such clear ostensive purposes. The concept good is an example. It would be a mistake to try to understand the concept by looking at the objects first picked out when someone called them “good”. Rather, an understanding of the concept comes from seeing that the concept isn’t an attempt to ostend some underlying feature of the world, but rather an attempt to mark out the thing as something that could give us reasons to pursue it. Good food is food it makes more sense to pursue than bad food, and of course this depends as much on the interest of the pursuer as the food itself. Often when we discover that something we thought was good is
6Of course, there is no guarantee even when we use terms in this ostensive way that there will be something real there we are pointing out, either because we are literally pointing at something that doesn’t exist or there are surrounding flaws in our ostension. For example, a concept like water picks out things at a certain generality, and it was certainly epistemically possible that the was nothing in common between all the things we originally grouped together as water. But, we will tend to revise if the is even any natural kind in the ballpark, so natural kind terms tend to be fairly sticky. This is perhaps why it is tempting to think about them in terms of reference magnets. (Lewis 1983)
different than we understood, we revise our evaluation that it was good, rather than our core understanding of goodness itself (though of course not always). As with the more ostensive concepts, the underlying purpose that the concept is being put to use for shapes the norms by which we choose to make revisions to our individual conception of the concept, and our understanding of when others are using the same concept.
The core difference between the Conceptual Design Method and merely looking at concepts from within this broad teleosemantic view is that the design method doesn’t solely focus on existing concepts. Concepts are representational tools and thus, like any tool, we can produce new tools as we come to have new purposes, or we recognize potential beneficial goals we had ignored all along. The concept of a double blind controlled study is a recent invention, but serves a highly necessary epistemic need, and the broadly teleosemantic view suggests that we can introduce new concepts to our community as long as we can adequately communicate the core purpose to which the concept is supposed to answer in such a way that even if the audience has an incomplete understanding of the concept they can tell what kind of input would count as corrective and thus engage in a process of refining their conception of the concept to ensure it better serves that role. They recognize when others in the community are using the concept with the same purpose and are thus attempting to communicate using that shared concept. To take our double-blind example, a student might not know at first whether a certain kind of information is blocked in a blind study, but if they know the purpose that double-blinding is supposed to accomplish, its epistemic purpose, they can evaluate reasons for why or why not studies that don’t block that information should be called double-blind.
The professor can also understand a student as using the shared concept, even when they
make a mistake in their use, because they can see the underlying intention to use a concept that serves the same purpose. Essentially the only limit to our conceptual design is whether the members of the community are sensitive enough to pick up on the features of the representational purpose the concept is meant to serve, can engage in corrective activities to refine their conception of the resulting concept, and recognize when others are attempting to use a concept with that purpose. This is largely why, for example, mathematicians can get away with so many stipulative definitions. If members of the community can recognize the purpose the concept has through the given definition, it’s not a problem if the concept didn’t exist prior, nor that much of their deep understanding of the concept will only exist after they have used it in proofs.
So, if we are designing epistemic concepts, what should our approach be? The basic approach advocated here is that in the case of epistemic concepts we should design concepts that would play important roles in the enforcement of epistemic norms and regulation of cognitive activity in service of our epistemic ends. We have some set of ends which it makes sense to label as epistemic ends be they truth, understanding, or knowledge. Given the set of facts about the abilities of ourselves and the members of our community there will be sets of norms that better and worse regulate our cognitive lives relative to those ends. The set or sets of norms that best regulate or cognitive lives relative to those ends are the norms we ought (at least epistemically) to let govern our lives. The actual governing of our activities by those norms will require some set of concepts for member of the community to represent the content of the norms and thus follow and enforce them. The epistemic concepts thus get their content from our intention to represent features of this normative space. For example, members of an epistemic
community will likely need a concept to talk about when people have succeeded or failed to live up to one or another of the norms. Justification might be an example of such a concept, and many epistemologists have certainly thought about justification in terms of following some set of epistemic rules. Of course, following the norms might require being able to recognize concepts which are only indirectly connected to the fulfillment of the norm. For example, you might need to have a concept of evidence to do what is necessary to develop justified beliefs. The indirect concepts that still count as epistemic like evidence will be so in part because the reason why things count as evidence or not evidence is precisely because of the central role evidence plays in letting you represent and follow the epistemic norms. Whereas, a concept like purple doesn’t have its content determined because of its functional role in the enforcement and following of epistemic norms in general7.
In sum epistemic concepts on this view are tools for enabling the expression, following, and enforcement of epistemic norms. The conceptual design method
encourages us to approach epistemic concepts by first thinking about what purpose in the larger normative structure this concept is supposed to serve and how best to tailor that purpose in service of the accomplishment of our epistemic ends. The explication of the content of a concept is thus constrained by a number of factors including (i) what our overall epistemic ends are, (ii) what our abilities are as a group and individually to use this concept in service of those ends, (iii) the interaction of this epistemic concept with other concepts given its place in the overall structure of our epistemic normative theory.
This is a complex methodological picture and thus it will be best to explain it in more
7Though of course a concept like purple might accidentally figure in a particular epistemic rule.
detail as we go through some examples, which will be the subject of the next chapter.