Skip to content

Two Tales of Marshmallows and their Implications for Free Will

Patricia Churchland, a prominent Neurophilosopher, just published a book on neuroscience and its ethical implications which led to a rather nasty exchange in the New York Review of Books with fellow philosopher Colin McGinn.  His pointed, to put it mildly, criticism of her work was based on philosophical considerations about the implications neuroscience has, or, as he argues, lacks, for the philosophy of mind. This criticism evoked two sentiments in me. First, I felt a strong sense of hopelessness for a world in which not even two philosophers can engage in a sober, respectful argument about something they disagree on; not even under the tutelage of the editors of the New York Review of Books, one of the so-called sanctuaries of intellectualism. Good luck Palestine and Israel! Thereafter, I remembered the unease I at times felt as a psychologist when hearing or reading about Churchland’s work.

In December 2013, for instance, she talked on the Philosophy Bites podcast with David Edmonds and Nigel Warburton about self-control and its implications for free will.  The starting point is the idea that if a person has little self-control then in what sense can we consider this person to have a free will. To build her argument, Churchland focuses on Walter Mischel’s Marshmallow Experiment, one of the most written-about experiments in psychology’s history. Briefly, in this experiment, young children around 4 years old are put in a room in front of a plate with one marshmallow and told that if they wait a long time, they will receive another marshmallow.  But if they felt that they could not wait longer, they had to ring a bell, and then could eat the one marshmallow immediately. Some children were good at delaying the gratification, and waited for the second reward, others were not. When Mischel and colleagues contacted the children as adolescents and adults, they found that the longer they had waited for the second marshmallow in the experiment years before, the more success (i.e., academic, social, and financial) they had later in life.

Churchland argues that this shows that people substantially differ in self-control, and cites research with rats – some rats can wait, others not – to further suggest that this difference is biologically hard-wired; self-control might be modifiable, but probably only a little bit. The person born with little self-control can do not much about it. While she says little about the philosophical implications of this idea, she suggests that philosophy has to accept that there are biological differences in self-control, and philosophers should give up the idea of a pure free will that can direct behavior without any interference, and start to consider the implications. And these ideas are very much in line with one tale science tells us about the marshmallow experiments: individuals have stable differences in their capacity for self-control, and we should investigate the biological underpinnings of these differences.

However, there is at least one other tale science has to tell about the marshmallow experiments.  Several years after the first experiment, Walter Mischel again studied children in the same one-marshmallow now, two later – scenario. However, this time, he instructed the children to think about different aspects of the marshmallow in front of them. One group was instructed to think about the aspects that make eating the marshmallow so pleasant (e.g. taste, smell), while another group was instructed to think about the more abstract aspects, such as imagining that the marshmallow is a cloud. When the children were thinking about the enticing aspects, they went for the single immediate reward, yet when they imagined the marshmallow as a cloud, they waited. A simple change in the instructions changed the capacity for self-control. Hence, one might argue that the differences between the children in the original experiments stem from knowing or lacking strategies to control themselves. This started a second tradition in research on self-control, which looks at self-control strategies that enable people to put their choices and intentions into action. At its most extreme, this research tradition argues that people do not differ in the capacity to self-control, but differ in their strategies and beliefs. If two people use the same strategy, or have the same belief about a situation, they will have the same amount of self-control.

While the latter research tradition is mainly situated in psychology, the former is more at home in neuroscience. And there are good reasons for it. Neuroscience, for instance, cannot detect if a person exerts poor self-control due to biological limitations or due to a belief that a person holds. Self-regulatory strategies or beliefs cannot be observed directly, but only by their consequences. And so far, all attempts to find the biological foundations of self-control failed, most recently the idea that self-control is related to glucose. Right now, we do not know if there is a biological foundation for self-control, let alone a genetic one, or if people differ substantially in their capacity to exert self-control, and thereby differ in their capacity for free will.

This brings me back to the unease I felt during the podcast. While I agree with Patricia Churchland that people show differences in self-control, it remains unclear what the reasons are. Cherry-picking some results in neuroscience and psychology, and neglecting a host of others in order to argue for one position, is not a great conversation starter. And as a trailblazer such as Churchland, one has to be extra careful.


Share on

3 Comment on this post

  1. Another interesting follow-up study to the marshmallow study is “Rational snacking: Young children’s decision-making on the marshmallow task is moderated by beliefs about environmental reliability” by Kidd, Palmeri and Aslin. In it they suggest that a confounding reason is that kids are much more likely to wait for the second marshmallow if they trust the experimenter to actually provide it and not take the first one away, which means that kids from more stable backgrounds are more likely to wait for a second marshmallow. Kids from more stable backgrounds are also more likely to do well later in life. Thus the marshmallow experiment might not have much to say about intrinsic levels of self control at all.

  2. I really like that study; that is the reason why I sneaked in the “beliefs” in the section on other interpretation. Maybe of interest is also that Walter Mischel hinted at something similar from his observations when doing the Marshmallow test in Africa. None of the kids waited, and when asked why, the children expressed disbelief that the experimenter would actually come back; they didn’t trust the experimenter. The Kidd et al study picked up this idea and nicely demonstrated the causality. In a similar vein, there is work by Joseph Kable (Penn), in which he and his colleagues show not only that certain expectations / beliefs about the relation of time and reward can explain the Marshmallow test perfectly, but also demonstrate that a short intervention can change such beliefs. From their and Kidd et al’s perspective, whether people wait or not seems to be a rationale decision based on past experience.

Comments are closed.