Determinism is the view that, given the laws of nature, all events are the necessary effects of previous events. When applying this to systems, one can say that a system is deterministic if there is only one way the system could evolve given its initial conditions and the laws governing it.
This view may be incoherent insofar as (1) the view often hides a set of prescriptions or statements that tell what we should maintain and (2) these prescriptions are impossible if determinism is indeed true. After all, according to determinism there is only one way things can unfold at any particular moment and prescriptions plausibly presuppose there are more than one.
Let me approach a justification for my first claim with this lengthy but helpful overview of C.S. Peirce’s anti-determinism views (cf. the entry on Peirce in the Stanford Encyclopedia of Philosophy):
“Against powerful currents of determinism that derived from the Enlightenment philosophy of the eighteenth century, Peirce urged that there was not the slightest scientific evidence for determinism and that in fact there was considerable scientific evidence against it. Always by the words “science” and “scientific” Peirce understood reference to actual practice by scientists in the laboratory and the field, and not reference to entries in scientific textbooks. In attacking determinism, therefore, Peirce appealed to the evidence of the actual phenomena in laboratories and fields. Here, what is obtained as the actual observations (e.g. measurements) does not fit neatly into some one point or simple function. If we take, for example, a thousand measurements of some physical quantity, even a simple one such as length or thickness, no matter how carefully we may do so, we will not obtain the same result a thousand times. Rather, what we get is a distribution (often, but not always and certainly not necessarily, something akin to a normal or Gaussian distribution) of hundreds of different results. Again, if we measure the value of some variable that we assume to depend on some given parameter, and if we let the parameter vary while we take successive measurements, the result in general will not be a smooth function (for example, a straight line or an ellipse); rather, it will typically be a “jagged” result, to which we can at best fit a smooth function by using some clever method (for example, fitting a regression line by the method of least-squares). Naively, we might imagine that the variation and relative inexactness of our measurements will become less pronounced and obtrusive the more refined and microscopic are our measurement tools and procedures. Peirce, the practicing scientist, knew better. What actually happens, if anything, is that our variations get relatively greater the finer is our instrumentation and the more delicate our procedures. (Obviously, Peirce would not have been the least surprised by the results obtained from measurements at the quantum level.) What the directly measured facts of scientific practice seem to tell us, then, is that, although the universe displays varying degrees of habit (that is to say, of partial, varying, approximate, and statistical regularity), the universe does not display deterministic law. It does not directly show anything like total, exact, non-statistical regularity. Moreover, the habits that nature does display always appear in varying degrees of entrenchment or “congealing.” At one end of the spectrum, we have the nearly law-like behavior of larger physical objects like boulders and planets; but at the other end of the spectrum, we see in human processes of imagination and thought an almost pure freedom and spontaneity; and in the quantum world of the very small we see the results of almost pure chance.”
I quote at length so you can see a plausible case for the thesis that determinism doesn’t logically follow from actual phenomena studied in the laboratory – we can’t simply infer it from empirical analysis. Determinism is a metaphysical view that can be preferred over other metaphysical views. Why might it be preferred?
Well, typically determinism is aligned with materialism or the view that reality is matter in motion governed by the laws of physics. This alignment seems promising since
a) by reducing all reality to matter in motion we have a simpler account than accounts that posit immaterial powers and
b) we have an account that is consistent with the knowledge established through contemporary science and
c) we have an account that promises to explain everything since everything is presumably material and
d) we have an account that allows for testability and physical, third-person verification – something which, since the Enlightenment, is desired as a means to knowledge and a just society.
But substance dualists who argue in defense of immaterial minds with free will might say that free decisions cannot be explained by determined matter in motion. Moreover, they might claim a material brain is simply not the kind of thing that can reason, entertain true or false propositions, experience qualities, and have a moral sense. But perhaps minds can have these abilities. As a result, they will say their perspective helps account for things determinism and materialism cannot and, as such, is justified despite being more complex. Naturally, they will be unimpressed by complaints that dualism fails to square with the results of science. For one, it may be the case that some of these results are immaterial or at least can only be understood with immaterial things like numbers. But, more importantly, any objection that says reality must fit with a materialistic and deterministic interpretation of scientific results begs the question. Finally, demands that minds be directly demonstrated with third person verification might beg the question as well: minds, if they are indeed immaterial subjects rather than physical objects, could only be known directly from a first person perspective and indirectly by others. To demand otherwise is to assume the truth of materialism. In light of these points they will prescribe adopting some form of dualism over materialism and determinism in order to account for more phenomena.
The point right now isn’t to see if some form of dualism is correct (I argue substance dualism may be correct here). The point is to see that the empirical evidence is underdetermined and our efforts to discover the truth about determinism, materialism, dualism, and so on are caught up in prescriptive efforts that offer us guidance in adopting one worldview over another. Which view offers more scope? Which is simpler? Which is more internally consistent and coheres with what we already know? Which is testable? Which promises to explain more? In answering these questions we are trying to infer the best explanation. And this process allows us to construct a justification that prescribes one view over the other.
But now we see the potential for incoherence in the determinist’s worldview: such prescriptions, on the one hand, imply that there is more than one way to go (for example, accept materialism over dualism) and, on the other hand, a determined world can only offer one outcome at any moment. Thus if determinists are indeed prescribing a worldview over another then they are maintaining an incoherent world view that both allows for, and denies, the possibility of prescribing.
For a related self-referential argument against determinism, go here.
For an argument that a determinist may presuppose freedom in the very assertion that determinism is true, go here.