Does Believing in Free Will May Make You Punish More Harshly?

Neuroscience, moral psychology, and inferential issues.

Philosophers have long debated about free will—are we in control of our actions or are they ultimately predetermined? Regardless of what the answer to this question is, recent work in moral psychology shows that our beliefs about agency affect our decision-making. Krueger et al. (2014) explored how belief in free will—which they termed “BFW”—affects our decisions about third-party punishment—which they termed “TPP” (1143). Third-party punishment is a way of punishing those who violate social norms; in TPP, rather than have the person wronged choose and execute the punishment of the wrongdoer, this responsibility is given to an impartial decision-maker such as a judge or court system (1143). Krueger et al. collected behavioral and fMRI data on how BFW affects choice of TPP for a variety of criminal actions. They also examined how the affect of a crime—that is, its emotional salience—impacted choice of punishment. Their paper demonstrates how reverse inference can be done well and shows that our discussions about free will have real-world consequences.

Using the Free Will and Determinism questionnaire (Paulhus & Carey, 2011), Krueger et al. divided up their subjects into libertarians—those who have BFW—and determinists—those who don’t have BFW. In the experiment, subjects were placed in an fMRI machine and presented with a series of written criminal scenarios, ranging from theft to rape and murder. They were then asked to choose, on a scale from 0 to 100, the appropriate level of punishment the criminal deserved for committing each crime. Subjects were also asked to rate their “subjective affective experience” of each scenario. This metric captured the emotional valence and arousal level elicited by the scenario. Using this information, they divided the criminal scenarios into a “low affective” and a “high affective” group (1144).

Krueger et al. found that for low affective crimes, libertarians punished more severely (10% so) and displayed significantly greater activation in the right tempo-parietal junction (rTPJ) than determinists. They posited that the libertarians display greater activation because the rTPJ is “probably involved in the…attribution of temporary intentions and beliefs of others” (1147). Thus, the libertarians punished more severely than the determinists for the low affective crimes because they felt the criminals were more responsible for their actions (1147). However, for high affective crimes libertarians and determinists punished similarly and showed similar levels of rTPJ activation. Both also showed increased activation in the right anterior insula (rAI), an area involved in “aversive interoceptive-emotional processing”—that is, internal awareness of negative emotions—as well as feelings of moral disgust (1147-1148). The rTPJ and the rAI are both part of the salience network of our brain which dictates what we attend to, and Krueger et al. argue that they compete in a winner-take-all fashion (1148). Thus, determinists and libertarians punished similarly for the high affective crimes because the scenarios elicited an extreme, negative emotional response which dominated processing (1148).

This paper has a similar flavor to Greene et al.’s (2001; 2005) work on the dual-process theory of moral judgements. As is the case with Greene et al., one might be a bit skeptical of the reverse inferences made by Krueger et al., especially after reading Poldrack (2006). Poldrack claims that reverse inferences—inferring that a task involves a cognitive function because the task causes increased activity in a brain region associated with that function—are epistemically weak because a single brain region can be involved in many cognitive functions. Thus, according to Poldrack, if we do want to make a reverse inference, we must account for all the cases in which that area of the brain is active yet the cognitive function of interest is not recruited.

Making a perfect Bayesian reverse inference as Poldrack would like requires knowing all the possible brain states in which our brain region of interest is active; however, this is impossible, as there could be an infinite number of cases in which it is active. Therefore, all we can do is try to make better, rather than worse, reverse inferences. Krueger et al. do so by providing multiple lines of converging evidence. For example, they discuss a previous study in which transcranial magnetic stimulation to the rTPJ “caused subjects to judge attempted harms as less morally forbidden and more morally permissible” (1147). This supports their inference about why determinists showed lower rTPJ activation and punished less severely as compared to the libertarians for low affective crimes. They also cite work which shows that the rAI and the rTPJ are “structurally and functionally connected to each other,” supporting their claim that these regions compete with one another to dominate processing (1147).

Krueger et al.’s reverse inferences are also strong because they focus on small areas—the rTPJ and rAI—with specialized functions— “aversive interoceptive-emotional processing” and “attribution of temporary intentions and beliefs of others”—as opposed to broader functions such as ‘emotion’ or ‘cognition’ (1147). Furthermore, Krueger et al. used psychological surveys to control for confounding variables such as differences in empathy or emotional awareness between libertarians and determinists, which strengthens their experimental design overall (1144). Thus, though the rTPJ and rAI are involved multiple cognitive functions, we have good reason to believe that Krueger et al.’s reverse inferences are justified.

Some might think that debating about free will is the arcane and unimportant work of philosophers in the ivory tower. However, as this paper shows, BFW impacts our decision-making in important and practical ways. For example, Krueger et al. briefly mention how BFW or lack thereof might bias jurors in a criminal case towards finding a defendant innocent or guilty, depending on the psychological affect of their offense (1148). Though contrived, one could imagine two people being tried for exactly the same low affective crime, yet one faces a jury of determinists and the other a jury of libertarians. Suppose that the former is found innocent and the other guilty. I don’t think we could call such a legal process just. Therefore, in order to eliminate bias, perhaps we should vet jurors by having them perform a series of psychological questionnaires, one of which tests for BFW. Even if we don’t go this far, we should at least be aware that our beliefs can influence our decision-making in unexpected and consequential ways.


Greene, Joshua. 2005. “Emotion and Cognition in Moral Judgement: Evidence from Neuroimaging.” In Neurobiology of Human Values, ed. Jean-Pierre Changeaux, Antonio R. Damasio, Wolf Singer, and Yves Christen, 57–66. Berlin: Springer-Verlag.

Greene, Joshua, R. Brian Sommerville, Leigh E. Nystrom, John M. Darley, and Jonathan D. Cohen. 2001. “An fMRI Investigation of Emotional Engagement in Moral Judgement.” Science 293(5537): 2105–2108.

Krueger, Frank, Morris Hoffman, Henrik Walter, and Jordan Grafman. 2014. “An fMRI Investigation of the Effects of Belief in Free-will on Third-party Punishment.” Social Cognitive and Affective Neuroscience 9(8): 1143–1149.

Paulhus, Delroy, and Jasmine Carey. 2011. “The FAD-plus: Measuring Lay Beliefs Regarding Free Will and Related Constructs.” Journal of Personality Assessment 93(1): 96–104.

Poldrack, Russell A. 2006. “Can Cognitive Processes Be Inferred from Neuroimaging Data?” Trends in Cognitive Science 10(2): 59–63.