Understanding the process of designing artifacts is the theme in phi- losophy of technology that most directly touches on the interests of en- gineering practice. This is hardly true for another issue of central con- cern to analytic philosophy of technology, which is the status and the character of artifacts. This is perhaps not unlike the situation in the phi- losophy of science, where working scientists seem also to be much less interested in investigating the status and character of models and theo- ries than philosophers are.

Artifacts are man-made objects: they have an author. The artifacts that are of relevance to technology are, in particular, made to serve a purpose. This excludes, within the set of all man-made objects, on the one hand byproducts and waste products and on the other hand works of art. Byproducts and waste products result from an intentional act to make something but just not precisely, although the author at work may be well aware of their creation. Works of art result from an intention directed at their creation but it is contested whether artists include in their intentions concerning their work an intention that the work serves some purpose. A further discussion of this aspect belongs to the philos- ophy of art. An interesting general account has been presented by Di- pert.

Technical artifacts, then, are made to serve some purpose, generally to be used for something or to act as a component in a larger artifact,

which in its turn is either something to be used or again a component. Whether end product or component, an artifact is ‗for something‘, and what it is for is called the artifact‘s function. Several researchers have emphasized that an adequate description of artifacts must refer both to their status as tangible physical objects and to the intentions of the peo- ple engaged with them. Kroes and Meijers have dubbed this view ‗the dual nature of technical artifacts‘; its most mature formulation can be found in Kroes. They suggest that the two aspects are ‗tied up‘, so to speak, in the notion of artifact function.

This gives rise to several problems. One, which will be passed over quickly because little philosophical work seems to have been done con- cerning it, is that structure and function mutually constrain each other, but the constraining is only partial. It is unclear whether a general ac- count of this relation is possible and what problems need to be solved to arrive there. There may be interesting connections with the issue of mul- tiple realizability in the philosophy of mind and with accounts of reduc- tion in science, but these have not yet been widely explored; an excep- tion is.

It is equally problematic whether a unified account of the notion of function as such is possible, but this issue has received considerably more philosophical attention. The notion of function is of paramount importance for characterizing artifacts, but the notion is used much more widely. The notion of an artifact‘s function seems to refer neces- sarily to human intentions. Function is also a key concept in biology, however, where no intentionality plays a role, and it is a key concept in cognitive science and the philosophy of mind, where it is crucial in grounding intentionality in non-intentional, structural and physical properties. Up till now there is no accepted general account of function that covers both the intentionality-based notion of artifact function and the non-intentional notion of biological function—not to speak of other areas where the concept plays a role, such as the social sciences. The most comprehensive theory, that has the ambition to account for the bio- logical notion, cognitive notion and the intentional notion, is Ruth Mil- likan‘s; for criticisms and replies, see Preston, Millikan, Vermaas and Houkes and Houkes and Vermaas. The collection of essays edited by Ariew, Cummins and Perlman presents a recent introduction to the gen- eral topic of defining the notion of function in general, although the em- phasis is, as is generally the case in the literature on function, on biolog- ical functions.

Against the view that the notion of functions refers necessarily to in- tentionality at least in the case of artifacts, it could be argued that even there, when discussing the functions of the components of a larger de- vice and their interrelations, the intentional ‗side‘ of these functions is of secondary importance only. This, however, would be to ignore the pos- sibility of the malfunctioning of such components. This notion seems to be definable only in terms of a mismatch between actual behavior and intended behavior. The notion of malfunction also sharpens an ambigui- ty in the general reference to intentions when characterizing technical artifacts. These artifacts usually engage many people, and the intentions of these people may not all pull in the same direction.

A major distinction can be drawn between the intentions of the actual user of an artifact for a particular purpose and the intentions of the arti- fact‘s designer. Since an artifact may be used for a purpose different from the one for which its designer intended it to be used, and since people may also use natural objects for some purpose or other, one is invited to allow that artifacts can have multiple functions, or to enforce a hierarchy among all relevant intentions in determining the function of an artifact, or to introduce a classification of functions in terms of the sorts of determining intentions. In the latter case, which is a sort of mid- dle way between the two other options, one commonly distinguishes between the proper function of an artifact as the one intended by its de- signer and the accidental function of the artifact as the one given to it by some user on private considerations. Accidental use can become so common, however, that the original function drops out of memory.

Closely related to this issue to what extent use and design determine the function of an artifact is the problem of characterizing artifact kinds. It may seem that we use functions to classify artifacts: an object is a knife because it has the function of cutting, or more precisely, of ena- bling us to cut. It is hardly recognized, however, that the link between function and kind-membership is not that straightforward. The basic kinds in technology are, for example, ‗knife‘, ‗airplane‘ and ‗piston‘. The members of these kinds have been designed in order to be used to cut something with, to transport something through the air and to gener- ate mechanical movement through thermodynamic expansion. However, one cannot create a particular kind of artifact just by designing some- thing with the intention that it be used for some particular purpose: a member of the kind so created must actually be useful for that purpose. Despite innumerable design attempts and claims, the perpetual motion

machine is not a kind of artifact. A kind like ‗knife‘ is defined, there- fore, not only by the intention of the designer of each of its members that it be useful for cutting but also by an operational principle known to these designers, and on which they based their design.

This is, in a different setting, also defended by Thomasson, who in her characterization of what she in general calls an artifactual kind says that such a kind is defined by the designer‘s intention to make some- thing of that kind, by a substantive idea that the designer has of how this can be achieved, and by his or her largely successful achievement of it. Qua sorts of kinds in which artifacts can be grouped, a distinction must therefore be made between a kind like ‗knife‘ and a corresponding but different kind ‗cutter‘. A ‗knife‘ indicates a particular way a ‗cutter‘ can be made. One can also cut, however, with a thread or line, a welding torch, a water jet, and undoubtedly by other sorts of means that have not yet been thought of. A ‗cutter‘ is an example of what could be looked upon as a truly functional kind. As such, it is subject to the conflict be- tween use and design: one could mean by ‗cutter‘ anything than can be used for cutting or anything that has been designed to be used for cut- ting, by the application of whatever operational principle, presently known or unknown.

This distinction between artifact kinds and functional kinds is rele- vant for the status of such kinds in comparison to other notions of kinds. Philosophy of science has emphasized that the concept of natural kind, such as exemplified by ‗water‘ or ‗atom‘, lies at the basis of science. On the other hand it is generally taken for granted that there are no regulari- ties that all knives or airplanes or pistons answer to. This, however, is loosely based on considerations of multiple realizability that apply only to functional kinds, not to artifact kinds. Artifact kinds share an opera- tional principle that gives them some commonality in physical features, and this commonality becomes stronger once a particular artifact kind is subdivided into narrower kinds. Since these kinds are specified in terms of physical and geometrical parameters, they are much closer to the nat- ural kinds of science, in that they support law-like regularities; see for a defense of this position Soavi. A recent collection of essays discussing the metaphysics of artifacts and artifact kinds is.

There is at least one additional technology-related topic that ought to be mentioned because it has created a good deal of analytic philosophi- cal literature, namely Artificial Intelligence and related areas. A full dis- cussion of this vast field is beyond the scope of this entry, however. In-

formation is to be found in this encyclopedia‘s entries on Turing ma- chines, the Church-Turing thesis, computability and complexity, the Turing test, the Chinese room argument, the computational theory of mind, functionalism, multiple realizability, and the philosophy of com- puter science.