Which is worse—doing harm or allowing harm? The answer to this question depends, of course, on many different factors: the type of harm involved (physical, psychological, etc.), the comparative seriousness of the harm, the motivation involved, and so on. The answer also depends on one’s fundamental ethical assumptions. In fact, this question highlights one of the central ethical debates between utilitarian consequentialism and non-consequentialism (or, deontology). Utilitarian accounts typically assume an equivalence between doing harm and allowing harm, whereas non-consequentialists assert that doing harm is typically worse than allowing harm. Here, I will articulate the main contours of this debate, and then briefly look at some practical ethical applications of this distinction.
Ordinary moral thinking would seem to be more non-consequentialist—so much so that moral philosophers sometimes refer to deontological approaches as “commonsense morality.” The data seems to support the idea that most people’s moral intuitions align with the non-consequentialist view that doing harm is worse than allowing harm. Utilitarian-minded moral philosophers and moral psychologists, however, tend to question ordinary moral intuition on this and other matters. Nonetheless, there have been some important recent attempts by moral philosophers to provide support for ordinary moral intuition against those who try to discredit it.
So, why do utilitarian-consequentialists claim that actively doing harm is no worse than allowing harm? It has to do with their conception of the good. Utilitarians claim the greatest good comes from impartially maximizing human happiness or well-being. Thus, morally right actions are those that either maximize happiness, or minimize harm. Because all that matters for utilitarianism is that the overall good is maximized, there is nothing especially wrong with harm doing, as opposed to allowing harm. If harm doing brings about the greater good, then so be it— this view that harm doing is justified by the greater good is the target of what is variously referred to as the “justice of personal rights” or “end-justifies-the-means” objection to utilitarianism. Furthermore, if allowing harm has the same bad consequences as doing harm, it is equally as bad. Because of this, utilitarianism sees no intrinsic significance to the doing/allowing distinction. From a non-consequentialist (or deontological) perspective, good consequences are not the only relevant moral factors. It also matters how the good consequences are brought about; that is, whether it involves harming innocent people, whether it involves intentional actions, or whether it violates other important moral principles. This gives rise to what I call the deontological asymmetry, namely, the view that duties to avoid harm (so-called negative duties) have a greater moral weight than duties to help (so-called positive duties). For our purposes here, our options will have comparable seriousness; the asymmetry might not hold if the harm is minimal and the good to be brought about is massive.
This deontological asymmetry is clearly illustrated by well-known thought-experiments in moral philosophy. In one such case, known as “Transplant,” a doctor is responsible to help save the lives of five people, each of whom has an urgent need for a different organ to survive. Clearly, it would not be morally permissible for such a doctor to kill one innocent person (whose organs miraculously matched the five) in order to bring about the good consequence of saving the five patients. Our intuitive responses to this scenario imply that the negative duty to avoid harming the one outweighs the positive duty to help the five people. On the other hand, in a contrasting case called “Drug,” if five people in some remote area each need one-fifth of a medication to survive, but another person needs the entire amount to survive, many would argue that we should give it to the five (other things being equal). If the conflict is between two positive duties to help, perhaps most people would say we should save the larger number. However, when the conflict is between a negative duty not to harm and a positive duty to help (and the harm that would be prevented is comparable), many would say it is impermissible to kill the one to save the five. Again, this seems to show that negative duties outweigh positive duties. This debate often focuses on the now famous (perhaps notorious) so-called Trolley Problem. (I will save a discussion of this well-known (perhaps notorious) thought experiment for another time, however.)
Another relevant difference between doing and allowing harm can be the cost involved. Fulfilling the duty not to harm others rarely involves a significant cost to ourselves, but our duty to prevent harm may well involve significant sacrifice. Thus, suppose someone standing on the end of a pier observes another person drowning and he or she could easily throw a life-preserver to that person. Even from a non-consequentialist view, it would be morally wrong not to do so, though it would certainly have been worse if he or she had intentionally tied-up that person with ropes and thrown him in the ocean in the first place. However, if it were during a storm, or even if the waves were particularly large, the person would most likely not be obligated (morally or legally) to jump in to save the drowning person due to the probability that he or she may not even survive the rescue attempt. Thus, often our duty not to harm is more stringent than our duty to save (i.e., not allow harm) because there is a high risk involved in fulfilling the latter. In this type of case, helping would be considered above and beyond duty, or “supererogatory.” As the pier example also illustrates, another reason why cases of doing harm are usually considered worse than allowing harm is that doing harm often involves unethical motivation, whereas allowing harm may simply be motivated by self-preservation or fear, and not by any ill intent.
Practical Applications of the Distinction
While there is much more we could say about the ethical/philosophical debate concerning the do/allow (as well as the intent/foresee) distinction, we will simply end here with several questions that highlight the practical, real-life relevance of this distinction. Think for example of the contemporary concern that advances in AI (artificial intelligence) technology will cause widespread job displacement. The question is whether that is something that companies who use AI to replace workers actually cause, or merely allow, and whether they might still be responsible in some way for what they allow. What if an Amazon or Uber uses AI—whether in the form of AI agents or robots, drones using AI technology, or self-driving cars—in such a way that it directly displaces other workers? Perhaps they are more directly doing harm (though one might argue they are justified in doing so). Perhaps they are simply allowing harm. What about Facebook or other social media companies? Where do we draw the line between what companies such as Facebook do and what they allow? Here we would need to further explore the question of what exactly it means to allow something. Whether or not they are legally responsible for social media addictions, an increase in teenage anxiety, the detrimental effects on relationships, and so on, we might still ask whether they are still in some way morally responsible for the foreseeable consequences of their technology. These are difficult questions. Notice also that these questions are broader than the killing/letting-die version of the do/allow harm distinction, on which we have mostly been focused here.
We also see a practical version of the do/allow distinction in the question of individual responsibility/obligation to give to those in need. On the side of those who view allowing harm almost as bad as doing harm, Peter Singer presents the extreme utilitarian view that not helping those in “absolute poverty” is almost as bad as killing them. Others, such as Philippa Foot, see it differently. As she puts it, perhaps with some understatement, “We are not inclined to think it would be no worse to murder to get money for some comfort such as a nice winter coat than it is to keep the money back before sending a donation to Oxfam or Care” (Foot, 1994). However, that does not mean one is not blameworthy who never helps. Furthermore, even non-consequentialists would agree that those who have more also have a greater moral responsibility to help. As Kant puts it, “To be beneficent, that is, to promote according to one’s means the happiness of others in need, without hoping for something in return, is everyone’s duty” (Kant, 1996, emphasis added). Thus, for someone like Bill Gates, perhaps giving is more like a duty of easy rescue.
Foot, Philippa. Virtues and Vices and Other Essays in Moral Philosophy. Oxford: Oxford
University Press, 1978.
Foot, Philippa. “Killing and Letting Die,” in Killing and Letting Die. Ed. Bonnie Steinbock and
Alastair Norcross. (New York: Fordham University Press, 1994).
Greene, J. Moral Tribes. (New York: Penguin Press, 2013).
Kagan, Shelly. Normative Ethics. (Oxford: Westview Press, 2006).
Nagel, Thomas. “You Can’t Learn about Morality from Brain Scans.” In The New Republic, Nov.
2013. Link to Article
Seligman, E.P. & Railton, P. & Baumeister, R. & Sripada, C. 2016. Homo Prospectus. Oxford:
Oxford University Press.
Singer, Peter. “Rich and Poor,” in Practical Ethics, 3rd Edition. (Cambridge: Cambridge
University Press, 2011)
Thomson, Judith. “The Trolley Problem.” The Yale Law Journal Vol. 94, No. 6 (May, 1985), pp.