<p>Joshua Greene and Jonathan Cohen predict that neuroscience’s growing predictive power will motivate anti-retributivist penal reform by amplifying our conviction that we lack free will. According to one agent-causal libertarian position, ACT libertarianism, we often exercise free will by satisfying the conditions that ACT, an agent-causal account of free will, posits for its exercise. This paper argues that, even if compatibilism is false, ACT’s account of free will implies that, for us, neuroscience’s growing predictive power is epistemically irrelevant to whether we have free will in the sense that it can provide us only unquantifiable epistemic risk regarding whether we have it. Since retributivism’s definitive tenet is that satisfying a criminal wrongdoer’s negative desert justifies legal punishment independently of any good consequences that might result, neuroscience’s growing predictive power cannot provide us any reason to implement anti-retributivist penal reform.</p>