Chapter 4: Third paper
4.5 Two projects for definitions
The classical theory of concepts thrusts definitions as its central theoretical construct and understands definitions as necessary and sufficient conditions. An important reason that explains this way of understanding concepts has to do, we believe, with the usual conflation between individual cognition and the scientific
enterprise (i.e., psychology as philosophy of science writ small; see, e.g., Fodor (2000, p. 52)). Biased by this conflation, psychologists ask definitions to do two things at once: (1) account for individual understanding and use of categories—a project that falls within the domain of the psychology of concepts, but also (2) account for how scientific communities develop ways of carving nature that serve their goals of understanding and prediction—a project that falls (more) within the domains of the sociology and the philosophy of science.
In individual cognition, definitions will serve to explicitly correct prototypes, reject some exemplars in favor of others, or change the probabilities underlying a Bayesian network, as we outlined in the previous section. Moreover, as we mentioned previously, definitions can also serve to introduce a new category, opening a space (Carey (2009) calls this a “placeholder”) that prototypes, exemplars and theories can then fill. Being told about the difference between two species of birds, an amateur bird watcher will explicitly bring to mind the distinguishing features of each species while bird watching. The same goes for elms and beeches. Exemplars of each species will thus be distinctly remembered and, with time, prototypes and even theories of the two species will develop.
In the scientific enterprise, definitions play a role in the social identification, use, and often challenge of those explicit characterizations of categories that are most useful to further science’s goals of representation (adequate carving) and prediction. In astronomy, a planet is, since August 24th, 2006:
A celestial body that: (a) is in orbit around the Sun, (b) has sufficient mass for its self-gravity to overcome rigid body forces so that it assumes a hydrostatic equilibrium (nearly round) shape, and (c) has cleared the neighbourhood around its orbit. (IAU, resolution B5, online: <http://www.iau.org/static/resolutions/Resolution_GA26-5-6.pdf>)
Any object that does not satisfy all of these criteria is not a planet, and any that does is44. Such definitions are closer to necessary and sufficient because such conditions are in general well-suited for these purposes. The same goes for most scientific, legal and mathematical concepts, but also for many common categories that have explicit definitions (e.g., “your brother’s stuff” is “everything on your brother’s side of the room, except the computer”).
Scientific definitions are different from those illustrated by the dolphin example above because the conventions established by scientists rest on more than simple individual understanding. Science is a social project that aims at producing knowledge with the highest form of justification and, to do so, controls its definitions publicly. Although scientific definition will, of course, influence how we individually categorize and make distinctions, both are nevertheless distinct; we would not expect a parent to use the scientific definition of planet above in everyday life, e.g., when correcting her child. The goal of an account of scientific definitions is not the explanation of individual cognition; concept experiments will not reveal anything relevant to what a scientific definition is or how it is built and regulated; philosophy of science might.
Indeed, we already said that definitions need not be necessary and sufficient conditions; that the latter are only one form of Type 2 concept. Conflating the psychological and scientific projects regarding concepts—as mentioned above when we discussed Fodor’s idea that psychology is philosophy of science writ small—is most probably in large part responsible for their identification. By conceiving definitions as Type 2 processes, we link them to explicit, voluntary processing, and
to a large extent to language. Type 2 concepts, as we take them, are explicit ways of making categories, and this often involves language, i.e., making distinctions using language. However, what we do not want to argue is that all and only processes involving Type 2 concepts are linked to language. On the one hand, some Type 1 concepts are linked to language, especially when some explicit distinctions become overlearned or, more prosaically, when we speak daily—we do not need to pick out every single word we say. On the other hand, Type 2 concepts, as we said, and will continue to argue in the rest of this section, are involved in automatic (albeit probably slow) Type 1 processes, shaping the content of Type 1 concepts, especially in learning.
As we agreed above, definitions may not be central in the fast pace of day to day cognition. For the reasons expressed above, however, we would certainly not characterize them as “marginal” as Machery (2010, 2011) does. Definitions, he argues, are only used intentionally in particular conditions. Moreover, he believes there has to be cognitive control exercised upon tasks in this condition to inhibit the more automatic processes (e.g., using prototypes, exemplars and theories). For Machery, we rarely use definitions, even in cases where there are clear and formal definitions available:
I suspect people rarely use these definitions in reasoning, to categorize, and so on. Instead, people seem to use prototypes, exemplars, or causal theories of grandmothers or bachelors. Evidence is consistent with these suspicions. (Machery, 2011, p. 207)
This is surely true, but not an objection to our proposed view. Definitions sometimes are used in cognition and they sometimes influence the acquisition of Type 1 concepts—especially when prototypes and exemplars can mislead, as is the
case for many concepts in science and other formal contexts (judiciary, business, technological)—and we do not claim more than this.
In the context of what we have presented, to eliminate Type 2 concepts from concept science would amount to excluding many important processes that are causally relevant to the processes studied in this domain. The argument that Type 2 concepts should be excluded from the psychology of concepts because they are not “used by default” seems ad hoc once the natural kind assumption and single- process views of the mind are rejected. We think there is a case to be made for the inclusion of other kinds of concepts, such as those we started to describe here. If concepts have both a Type 1 and a Type 2 aspect and if, as we argued, they interact causally, then both must be studied.