Harry Frankfurt argues for compatibilism by redefining free will not as alternative possibilities, but as the condition in which a person’s effective will is the will they reflectively want to have. He first distinguishes between different levels of desire. A first-order desire is a desire to perform an action, such as wanting to take a drug or wanting to refrain from taking it. A second-order desire is a desire about one’s desires, such as wanting to have the desire to resist the drug. More importantly, a second-order volition is a desire that a particular first-order desire become one’s effective will. For Frankfurt, the “will” is specifically the first-order desire that actually moves the agent to act. He thus defines a person as a being capable of reflective self-evaluation, because persons do not merely have desires but can care about which desires govern them. By contrast, a “wanton” has first-order desires but lacks second-order volitions. A wanton may act intentionally, but they do not care which desire becomes their will, so their action does not express reflective identification. Frankfurt illustrates this through his examples of addicts. The unwilling addict has conflicting first-order desires: he wants to take the drug, but he also wants to refrain from taking it. More importantly, he has a second-order volition that the desire to refrain should become his will. However, when the desire to take the drug actually moves him to act, his effective will conflicts with the will he wants to have. Therefore, he lacks freedom of the will. The willing addict, by contrast, also desires to take the drug, but he wants this desire to be effective. His first-order desire and second-order volition coincide, so he acts from the will with which he identifies. Frankfurt then distinguishes freedom of action from freedom of the will. Freedom of action means being free to do what one wants, whereas freedom of the will means being free to will what one wants to will. This distinction highlights that causal determination does not automatically prevent an agent’s effective desire from matching their second-order volition. Even if a person’s desires and actions are causally determined, their will can still be free if the desire that moves them is the desire they reflectively endorse. Therefore, free will depends on the internal hierarchy and harmony of desires, not on the absence of determinism.
Frankfurt’s compatibilism faces three serious weaknesses. First, Gary Watson’s “real self” objection challenges Frankfurt’s assumption. Frankfurt does not explain why a second-order volition is automatically more authentic than a first-order desire. For example, a student may have a first-order desire to rest, but also a second-order desire to suppress that desire and keep studying because they have internalized extreme parental pressure. Watson shows that this second-order volition may not express the student’s real self, but instead reflect anxiety, social conditioning, or fear of failure. Therefore, the problem is not simply whether the agent identifies with a desire, but whether that identification truly represents the agent’s values or rational self.
Second, Frankfurt faces an infinite regress problem. If a first-order desire is free only when endorsed by a second-order volition, then we may ask whether the second-order volition itself is free. If it needs endorsement by a third-order volition, and that by a fourth-order volition, Frankfurt’s account risks an endless hierarchy, and never fully explains what makes any desire genuinely free in the first place.
Third, Frankfurt’s theory struggles with manipulation and causal history. An agent may sincerely identify with their will, yet that identification may have been produced by earlier manipulation, addiction, brainwashing, or trauma. Frankfurt focuses on the agent’s present psychological harmony, but Derk Pereboom’s Four-Case Argument shows that internal harmony alone may not be enough for moral responsibility. Thus, Frankfurt explains the reflective endorsement of agency, but he does not fully explain why the endorsed will genuinely belongs to the agent.
Peter van Inwagen argues for incompatibilism by claiming that if determinism is true, then no agent is able to act otherwise, because every action is entailed by the remote past together with the laws of physics. He begins by defining causal determinism: for every instant of time, there is a proposition expressing the complete state of the physical world at that instant. If determinism is true, then any later state of the world is entailed by the conjunction of an earlier state of the world and the laws of physics. Van Inwagen then explains, to have free will is to have the power or ability to act otherwise than one in fact acts. To make this more precise, he introduces the phrase “render a proposition false.” If an agent could have done otherwise, then the agent could have brought about conditions under which a proposition describing the actual world would have been false. His central example is a judge, J, who refrains from raising his hand at time T, even though raising his hand would have prevented an execution. Let P₀ be the proposition expressing the state of the world before J was born, L be the conjunction of all laws of physics, and P be the proposition expressing the state of the world at T. If determinism is true, then P₀ and L entail P. However, if J had raised his hand at T, then P would have been false, because the actual state of the world at T would have been different. Therefore, if J could have raised his hand, he could have rendered P false. But since P is entailed by P₀ and L, rendering P false would require rendering false the conjunction of P₀ and L. Van Inwagen then argues that J cannot render P₀ false, because P₀ concerns the state of the world before his birth. Nor can J render L false, because laws of nature are universal and unchanging. Thus, since J cannot change the past or falsify the laws, he cannot render P false. Therefore, if determinism is true, J could not have raised his hand, and free will is incompatible with determinism.
Van Inwagen’s incompatibilism is stronger because it directly addresses the metaphysical threat determinism poses to freedom. First, his Consequence Argument is logically powerful because it is built through formal reasoning. Unlike Frankfurt, who redefines freedom through internal identification, Van Inwagen constructs a precise structure of syllogism using philosophical vocabulary such as propositions, entailment, laws of physics, and the state of the world. Its foundation also rests on the widely accepted Principle of Alternative Possibilities. Frankfurt must first revise the meaning of free will by shifting it from alternative possibilities to second-order volitions, whereas Van Inwagen begins from the more traditional and intuitive understanding of freedom.
Second, his argument targets control at the deepest metaphysical level. Some theories explain freedom in terms of acting without coercion, acting according to one’s desires, or identifying with one’s will. However, Van Inwagen asks a more fundamental question: are the action, the desire, and even the identification ultimately up to the agent? This makes his argument stronger than ordinary incompatibilist intuitions, because he does not need to posit dramatic cases of compulsion or manipulation.
Third, Van Inwagen exposes an inherent weakness in compatibilist replies because compatibilists cannot answer him simply by showing that the agent acts voluntarily. Frankfurt avoids the Consequence Argument by rejecting the importance of alternative possibilities, while David Lewis responds by distinguishing between “breaking” a law and acting such that a different law would have held. These replies show that compatibilists must work around the argument, either redefine freedom, weaken the meaning of “can,” or reinterpret laws of nature. Thus, Van Inwagen’s argument directly pressures the central compatibilist claim that determination and freedom can coexist.