286 Peter Kroes
cannot be explained in principle on the basis of knowledge of features of its emergence
base, and weakly nonexplainable when it cannot be explained relative to a given state of
knowledge. We do not enter into a discussion about whether the two criteria for epistemic
emergence are independent of each other. That would require an in-depth analysis of the
(logical) relations among the various models of prediction and explanation. The strong
form of epistemic emergence occurs when a feature of a system is in principle neither
predictable nor explainable. This is a “timeless” form of epistemic emergence. All other
forms are weak in the sense that they depend on a given knowledge base, which implies
that a feature may acquire or lose its status of being epistemically emergent in the course
of time.
With these distinctions in mind, what can we say about the epistemic emergence of
functions of technical artifacts? We take knowledge of the physical parts of which they
are made as our knowledge base. Since it is easy to come up with weak forms of epistemic
emergence by restricting the actual physical knowledge about the components of technical
artifacts in the knowledge base, we assume here that there are no limits on this knowledge
base; it contains all possible physical knowledge about the components. So we are inter-
ested in the strong form of epistemic emergence.
Take again the example of our mechanical clock. In what sense, if any, can its function,
measuring time, be claimed to be epistemically emergent on the knowledge of its physical
structure? In this case, too, we have to distinguish carefully between the function of the
clock and the physical behavior (capacity) of the clock corresponding to that function. The
prediction and explanation of the physical behavior of the hands of the clock, starting from
the physical properties and the arrangement of the various parts of the clock, do not seem
to pose any problems. There is no reason to assume that this physical behavior is an epis-
temically emergent property of this physical system.
Whether the function of the clock is an epistemic emergent property is another question.
Can its function be predicted on the basis of knowledge of its physical structure?^21 Of
course that would not be a problem for someone who is familiar with mechanical clocks;
on inductive grounds its function can be reliably predicted. But theoretical, not inductive,
prediction is the issue here. Suppose some archaeologists who also happen to be mechani-
cal engineers, but who are totally unfamiliar with mechanical clocks, dig up an object that
was used by an extinct civilization as a (mechanical) clock and try to “retrodict” its func-
tion. By clever reverse engineering they may come up with the claim that its function is
to produce a regular motion of the hands (probably they would make use of a functional
decomposition of the object that explains how all the parts with their subfunctions together
realize a regular motion of the hands). But their engineering and physical knowledge
would not enable them to retrodict that its function was to measure time.^22 The function
is also not explainable from the given knowledge base.^23 The explanandum is the function
of the technical artifact. What can be explained are the regular movements of the hands.
But that still does not explain that the function of the physical device is to measure time.