Technical Artifacts, Engineering Practice, and Emergence 283
tunities to causally transform the physical environment, that is, opportunities that “go
beyond” those presented by the (combination of the) causal powers of the parts of the
technical artifacts. Therefore the existence of such systemic properties does not undermine
the control paradigm. The functional properties of ordinary technical artifacts all fall into
this category: they are systemic properties without causal powers of their own, which
makes them according to the second interpretation ontologically nonemergent properties
(in contrast to the conclusion we reach starting from the fi rst interpretation of ontological
emergence). The causal powers or capacities corresponding to the functions of ordinary
technical artifacts can all be reduced (or are assumed to be reducible) to the causal powers
or capacities of their parts. If that were not the case, the use of techniques like functional
and physical decomposition would not make much sense.^14
Is the possibility of emergent features with causal powers of their own to be taken seri-
ously within engineering practice, or at all? In discussions of the mind-body problem,
emergent causal powers play an important role because of mental-to-physical causation;
emergent mental states appear to have a causal impact on their own emergence base,
namely brain states. This kind of emergent causal power involves what is called “refl exive
downward causation”: the system as a whole causally infl uences the state of its own con-
stituents, which in turn determine the causal powers of the whole system (Kim 1999).
Whether the notion of “refl exive downward causation” is coherent is much disputed, since
it seems to involve the notion of “self-causation” or “self-determination.” Arguments that
take their cue from analogies with biological systems and advocate incorporating emergent
features with their own causal properties into technical systems often hinge precisely on
this possibility of refl exive downward causation. Desirable emergent properties such as
self-repair, self-optimization, self-learning, and adaptability all imply that systems with
these properties change their own emergence base one way or another. Again the question
may be raised whether this is a coherent conceptual possibility. Another, equally disput-
able, conceptual possibility would be the emergence of causal powers in technical systems
that allow changing the environment of those systems.^15 Such causal powers would be
neither refl exive nor necessarily downward.^16
If we assume, for the sake of argument, that emergent causal powers, whether refl exive
downward or not, are a conceptual possibility, then their actual occurrence in technical
systems would pose serious problems for traditional engineering practice. Assuming that
ontological nonreducibility implies epistemic nonreducibility,^17 it would not be possible
to predict the occurrence of such emergent causal powers in a particular technical system
on the basis of considerations that start from the causal powers of its parts. Such phenom-
ena could be predicted only empirically (inductively; see the next section). That by itself
may not be a serious drawback for engineering practice, in which direct experience with
regard to the functioning of technical artifacts (e.g., the actual testing of prototypes) is
more or less standard procedure. But the existence of emergent causal powers would
immediately raise the question whether it is possible to harness them and to design