The question of whether sympathy or free will is more important for forgiveness is a bit like asking “Which is more important for human survival, oxygen or nuclear arms control?” Both questions demand comparisons between entities that cannot be placed on a common scale, so the questions might not have answers at all. Even so, the “Sympathy or Free Will” question offers a way into another important question that certainly does have an answer: Why are some harms easy to forgive and others hard to forgive? To answer this question, sympathy and free will turn out to be very relevant indeed.

Before going any further, let me define forgiveness. I think about forgiveness from an evolutionary point of view, so it has been important to define forgiveness functionally—that is, in terms of what the human capacity for forgiveness is for. Animal researchers have observed that many non-human animals that live in stable groups tend to engage in friendly behaviors (usually involving physical contact) after they have had a conflict or an aggressive interaction. These friendly post-conflict behaviors (known collectively as “reconciliation”) seem to serve the function of helping individuals to patch up important relationships. Based on these observations and other considerations, I have come to think that the thing we call forgiveness has a similar function, which is to inhibit retaliation and to motivate behaviors that will coax a transgressor back into a mutually beneficial relationship—on the condition that the harm-doer mend his or her ways and raise his or her regard for the forgiver’s welfare going forward.[i]

If the human capacity for forgiveness really did evolve for the function of coaxing transgressors back into constructive relations, then the behavioral systems responsible for forgiveness should be good at executing that function. (This is a standard adaptationist article of faith, and an empirically defensible one at that.[ii]) So far, so good, but what should we expect a well-designed forgiveness system to do? Well, the most important thing it should be good at is keeping itself alive (that is, at motivating choices that would have been fitness-enhancing during the era in the deep ancestral past when the system was evolving). But how would it do that? By motivating forgiveness for the right sorts of people (that is, people who are likely in the future to do things that would help the forgiver), and withholding it from the wrong sorts of people (that is, people who are likely in the future to do things that would harm the forgiver).

Forgiveness Is Easy When Facts Suggest That Forgiveness Is The Smart Move

In other words, forgiveness systems should be avid connoisseurs of data, obtained from the real world of social interaction, that enable them to determine whether a candidate for forgiveness is the kind of person that the forgiver would benefit from continuing to associate with. When a forgiver perceives that a transgressor continues to have potential value as a relationship partner (as a friend, say, or a neighbor, or a loved one, or a business partner), then he or she just might be a person who is “worth forgiving.”

It ought to be easy, relatively speaking, to forgive someone like that.

Sympathetic Forgiveness Feels Easy

In the brain, forgiveness systems are computing who is worth forgiving and who is not worth forgiving based on transgressor’s perceived relationship value and his or her perceived likelihood of harming the putative forgiver again in the future. In the felt experiences that our conscious selves can access, however, we know nothing of the brain’s computations. Instead, we talk about “feeling sorry” for offenders, or about feeling that “the offender has suffered enough” for his crimes and should be taken back as a friend, or about feeling “moved” by a heartfelt apology. In other words, sympathy matters. When you feel sympathy for a transgressor, forgiveness doesn’t seem to require much effort. Sympathy is a clue that you’re on the easy road to forgiveness.

By using the word sympathy, I’m actually thinking of empathic concern, which Dan Batson defines as “an other-oriented emotional response elicited by and congruent with the perceived welfare of someone else” that includes “feelings of sympathy, compassion, soft-heartedness, tenderness, and the like—feelings that are inherently other-oriented.”[iii] Research has shown that sympathy of this sort is one of the very best predictors of forgiveness.[iv] This is true in the context of day-to-day transgressions as well as in the context of the kinds of harms that send people in search of professional help. Sympathy’s efficacy in promoting forgiveness even explains why restorative justice interventions—in which willing offenders and victims meet up in a safe, monitored environment (usually after sentencing) to discuss a crime and its impact on the victim, and to give the offender a chance to apologize and atone—are so effective at reducing the desire for revenge and promoting forgiveness.[v] Sympathetic forgiving is an easy kind of forgiving. No choosing is really needed. Like vision, or hearing, sympathetic forgiveness “just happens” because the brain’s forgiveness systems make it happen.

When Forgiveness is Hard, (Something Like) Free Will Becomes Crucial

So there is a kind of forgiveness that is “easy.” Sympathy often is a clue that we’re on that easy road. However, sometimes we want to forgive (perhaps out of ethical or religious convictions, or because we’re tired of taking antacids for heartburn, or half a Benadryl to fall asleep) even when a transgressor has not apologized, tried to make amends, or demonstrated remorse—that is, in a sympathy-free social environment. Here, something like free will becomes important.

But hang on—is free will even real? Well, I see no reason to believe in any sort of (to use Daniel Dennett’s phrase) “God-like power to exempt oneself from the causal fabric of the physical world.”[vi] However, philosophers such as Dennett, and Richard Holton have persuaded me that there is a kind of human freedom that is neither illusory, nor trivial, nor incompatible with determinism. This freedom consists to no small degree in the ability to make causally efficacious choices. Sometimes (and perhaps much of the time) the choices we make are actual causes of our behavior. (I will not attempt to substantiate this claim here; take it or leave it, or read Dennett and Holton.) Reducing “free will” to “efficacious choosing” might look like a major demotion for such a venerated concept as free will, but with respect to understanding what helps people forgive, it is just the sort of freedom that ought to concern us.

On the hard road to forgiveness, when the offender has done little or nothing to merit it, we have to choose to forgive. Sometimes, that choosing will be efficacious, and sometimes it won’t be. The same is true of choosing to lose weight. On one hand, choosing to lose a moderate amount of weight is hard: Only 4 out of every 100 overweight or obese people who choose to slim down succeed in maintaining a 20% weight loss for a full year. On the other hand, more than 1 in 3 succeed in maintaining a 5% weight loss for a full year. In an important sense, then, we are less free (4% success rate) to choose to lose a lot of weight (20%) than to lose a little (37% success rate at keeping off 5%).[vii] Freedom to choose is not an “all or nothing” sort of thing: It is graded. [viii]

The same is true with forgiveness in a sympathy-free environment. Consider someone from school or work, perhaps, who said something that hurt your feelings a few years ago. Even if he has shown no signs of wanting to be forgiven, you still might find that your choice to forgive him causes new ways of thinking about him (perhaps you make a resolution such as “I’m going to try to view this person the way a loving mother might, even though he behaved boorishly toward me in the past”) or behaving toward him (perhaps you choose to defend him against someone else’s unfair criticism when he is out of the room) that, jointly, are causally efficacious in reducing your bitterness and increasing your willingness to re-establish a positive relationships. Your choice to forgive that person will quite possibly succeed. (Try it.) On the other hand, choices to forgive much more grievous harms, in the absence of the information that the mind’s evolved forgiveness circuitry is looking for, are more likely to fail. You are freer to choose to forgive the newspaper delivery person who keeps throwing your paper into the neighbor’s yard than to forgive the drunk driver that killed your daughter. This is a hard truth, but it’s good to know what you can (probably) change and what you (probably) can’t.


Forgiveness can feel easy, especially when the offender has admitted fault, apologized, offered to compensate you, promised to refrain from harming you in the same way again, and continued to make good on that promise day in and day out. Transgressors who go through these paces are relatively easy to forgive: The research on restorative justice shows us that the psychological effects of even very serious harms can be undone when these conditions are in place. This is because these realities give the human mind’s evolved forgiveness circuitry exactly what it’s looking for: Evidence that a harmdoer is both valuable and safe. In such cases, forgiveness can seem effortless; natural. In such cases, forgiveness might not require any choosing at all. Sympathy seems to do all the work.

Then, there’s the hard kind of forgiveness, which is what’s required when the offender has not apologized, offered to make amends, or changed her ways. In such cases, success in forgiving all turns on the efficacy of our ability to choose it. You won’t succeed in losing any weight at all if you don’t resolve to do so, and you won’t succeed in forgiving the unrepentant offender if you don’t at least try. Still, we ought to be realistic—and forgiving of ourselves—about how much forgiveness we can accomplish in a world in which nothing has changed.

(Please scroll down for questions).

[i] McCullough (2008).

[ii] Davies, Krebs, and West (2012).

[iii] Batson, Ahmad, and Lishner (2009).

[iv] Fehr, Gelfand, and Nag (2010).

[v] Sherman et al. (2005), Strang et al. (2006).

[vi] Dennett (2003), p. 13.

[vii] Kraschnewski et al. (2010).

[viii] Holton (2009).



Batson, C. D., Ahmad, N., & Lishner, D. A. (2009). Empathy and altruism. In C. R. Snyder & S. J. Lopez (Eds.), Oxford Handbook of Positive Psychology (2nd ed., pp. 417-426). New York: Oxford University Press.


Davies, N. B., Krebs, J. R., & West, S. A. (2012). An Introduction to Behavioural Ecology (4th ed.). Sussex, UK: Wiley-Blackwell.


Dennett, D. C. (2003). Freedom Evolves. New York: Viking.


Fehr, R., Gelfand, M. J., & Nag, M. (2010). The Road to Forgiveness: A Meta-analytic synthesis of its Situational and Dispositional Correlates. Psychological Bulletin, 136, 894-914.


Holton, R. (2009). Willing, Wanting, Waiting. New York: Oxford.


Kraschnewski, J. L., Boan, J., Esposito, J., Sherwood, N. E., Lehman, E. B., Kephart, D. K., & Sciamanna, C. N. (2010).


Long-term weight loss maintenance in the United States. International Journal of Obesity, 34(11), 1644-1654.


McCullough, M. E. (2008). Beyond revenge: The Evolution of the Forgiveness Instinct. San Francisco: Jossey-Bass.


Sherman, L. W., Strang, H., Angel, C., Woods, D., Barnes, G. C., Bennett, S., & Inkpen, N. (2005). Effects of face-to-face restorative justice on victims of crime in four randomized, controlled trials. Journal of Experimental Criminology, 1, 367-395.


Strang, H., Sherman, L., Angel, C. M., Woods, D. J., Bennett, S., Newbury-Birch, D., & Inkpen, N. (2006). Victim evaluations of face-to-face restorative justice conferences: A quasi-experimental analysis. Journal of Social Issues, 62, 281-306.


Discussion Questions

  1. Sympathy-based forgiveness is characterized here as an “easy” kind of forgiveness, and forgiveness based on free choice is characterized as a “hard” kind of forgiveness. Can you think of exceptions to this characterization?
  2. Can you think of real-life instances of people who managed to forgive extremely grievous harms without any changed behavior from the offender? Were they able to generate sympathy on their own? If so, did that sympathy result from their choosing to think about their offenders in such a way that sympathy resulted?
  3. Which, in your experience, is more effective or long-lasting: sympathy-based forgiveness or choice-based forgiveness?

9 Responses

  1. plewsogan says:

    MIchael,  great essay and topic!  I wonder if one might think about forgiveness as an iterative process, in which perhaps one person chooses to take a step toward forgiveness, which elicits some modest self reflection on the part of the recipient, which then begins to generate some modest sympathy that was not originally present, and on it goes.  In other words, rather than an all or nothing state, this is often an iterative, evolving state in which each entity changes the other along the way, for the better.  thus perhaps the initial goal for one who must choose some action unilaterally is not forgiveness, or apology, if these steps are too extremen, but merely an openness to engaging. Restorative justice often begins with simply the willingness to engage.

    • Michael McCullough says:

      I could not agree more. Forgiveness very often involves the sort of iterative interpersonal process you describe here. When we can’t will ourselves into forgiving (which is, for the reasons I described in the essay, not the easy route anyway), or count on the offender to make it easy for us, often it is easiest to make forgiveness happen for ourselves by trying to kindle those changed social realities into existence. This is exactly how restorative justice interventions like those I described in my essay (and in my book, Beyond Revenge) work: A crime victim wants to feel better, and so agrees to participate in a victim-offender conference. By doing so, the victim exposes himself or herself to new social realities that he or she would not have experienced otherwise (because he or she and the offender would otherwise have had very little chance of interacting in a similar, non-threatening venue). From what I understand about restorative justice interventions like these, they can be very powerful.

  2. George Gantz says:

    Approaching forgiveness as if sympathy and free will are separate poles seems a bit simplistic.  Is it not more correct to say that forgiveness (post-conflict reconciliation) is a multi-faceted process engaging both emotive and rational capacities?  Ultimately, forgiveness requires that we get our hackles back down – letting the emotions of anger, fear and desire for retribution abate – otherwise we will never be able to forgive.  But for a sentient human, it also requires a rational context – a conclusion (or rationalization) that this is the right choice.  A given act of forgiveness will require varying emotional and rational efforts depending on many factors.

    In some cases, we may forgive for purely emotional reasons, if the cost of confronting a high-status perpetrator is too high, for example.  We make a choice and then our rational thinking will tag along – “oh well, it wasn’t such a big deal anyway.”  In other cases, we need to work rationally through the moral, psychological and consequential implications to outline a course of action – and bend our anger and fear to align with the plan so that we achieve a freely willed resolution.  It is worth noting that our ability to navigate these complex and challenging waters is easier, and the outcomes measurably better (see BQO – “Does Belief in Free Will Make Us Better People” or “Which Beliefs Contribute to Virtuous Behavior?”), if we have a religious or moral ground to stand on.  

    In many cases (but not all), forgiveness will also involve negotiating relationships between two (or more) parties which adds more layers of emotive and cognitive challenge.  Sometimes the negotiation is easy once the “heat of battle” has abated and everyone has had a chance to take a deep breath.  There are also cases of qualified or limited forgiveness based negotiation such as a “trust but verify” resolution of conflict.  Ultimately, forgiveness is quite a complex process – and becoming an optimal “forgiver” will require a delicate balancing of emotive and cognitive effort in order to achieve the goal of true wisdom.

    • Michael McCullough says:

      Thanks for the ideas. Although I agree with you that sympathy-driven forgiveness and choice-driven forgiveness are rarely sealed off from each other in practice, I am not so sure that forgiveness inevitably involves the sorts of rational calculations, which I take to mean something like “deliberative decision-making about the costs and benefits or rightness/wrongness of forgiveness” as you suggest. Judgments of rightness/wrongness or costs and benefits certainly can be part of the decision-making process, but when facts on the ground have compellingly indicated that the transgressor is both valuable and safe (apologies, expressions of desire for forgiveness, promises not to repeat the behavior, emotion expressions consistent with internal states like these), I doubt that forgiveness requires much in the way of reflective, conscious decision-making. We also should not underestimate the power of habit. In close relationships, for instance, it appears that forgiveness becomes quite easy through the power of habit or because the implicit value of the relationship overrides any sense that conscious deliberation is required. Nevertheless, your observation that sympathy-driven forgiveness and choice-driven forgiveness are not opposite poles (along with your observation, toward the end of your essay that forgiveness itself, is a graded phenomenon) adds nice nuance to this discussion.

      • George Gantz says:

        I think my definition of “rational calculation” may be less restrictive than yours.  However we get there (rational decision or rationalizing a choice we have already made), the thinking mind has to be on board with the emotional one to be able to “forgive”.   Even then, forgiveness may be qualified by rational calculations – “yes, I can trust this person, but only as far as I can throw them!” 

  3. Meyer1953 says:

    Forgiveness, in its mature form, is defense-based. If it is not defense-level, the offense not only continues but remains degenerate and thus worsens. So forgiveness which is anything more than complicity to offense must be anchored in defense. The defining term of defense is that the offender’s predilections are exactly fair to turn back upon the offender. Forgiveness then is turning the offender’s predilections back upon the offender in a manner that leaves the offender a strong and whole person.

    The conclusion of WW1 was a sloppy, emotional rant that left the defining offense in place while placing punishment on the offenders. But the punishment did not visit the predilection to slaughter back to the offenders, so the armistice that ended the war did not defend innocence against the slaughterers. WW2 then resulted with its horrendous offenses. WW2 was concluded with unconditional surrender, resulting from a delivered return of predilection to the offender (Germany in this example). The forgiveness that followed WW2 was then swift and sure, such that with the Marshall plan specifically and other allied efforts the former offender rejoined the whole-world as a full and vital member, as it continues to this day.

    So defense is the very basis of forgiveness, and without defense at the core of forgiveness, so-called forgiveness is merely enablement of self-increasing degeneracy.

  4. wondering14 says:

    The article defines forgiveness functionally, in terms of why it is given, forgiveness “for” something..  I will forgive you “if”.  There are all kinds of “forgivers”.  And all kinds of definitions to fit them within, as the author has done.  Conditional forgiveness may be prevalent in some cases, but could not such a pragmatic “forgivenss” not be further refined and named “forgiveness of the head” (or of free will). 

    Is there not unconditional forgiveness, and unconditional forgivers?  Regardless of what is to be forgiven?  I think so.  This subset might be named “forgiveness of the heart” (of of sympathy).  In the hard scientific world, such forgiveness may be seen as “soft forgiveness”, but it is more durable and genuine than the contingent kind.

  5. Michael McCullough says:

    I’m not sure I’m ready to go along with your analysis of WWI’s failures to bring a lasting peace to Europe (I think other powerful causes were at work), but I do get your point about the need for deterring future harms, and I agree with it. In both an evolutionary perspective and an everyday perspective, forgiveness is risky business (that is, both for the fitness of the genes that make forgiveness systems happen and for the welfare of human beings) when it occurs without also signaling one’s readiness to requite future harms. Even so, forgiveness stands to be at its most winsome when it is communicated as an alternative to the retaliatory inflection of harm in response to the initiating harm, because when one is successful in changing an offender’s regard for you without punishment, you also reduce the risk of eliciting counter-retaliation or destroying the relationship entirely. I do think that applying this logic at the level of international relations is a bit problematic, though, because the notions of “raising another agent’s regard for you” and even “destroying the relationship entirely” can mean importantly different things in an international context. But yes; deterrence is critical to making forgiveness work in the long term. Offenders’ incentives need to be changed. Pursuing a forgiving strategy is an attempt to get deterrence without the retaliatory infliction of harm: As Mark Twain is credited (erroneously, most likely) with saying, “Forgiveness is the fragrance the violet sheds upon the heel that has crushed it.” It is an effort to get deterrence and a preserved relationship without the risks associated with using retaliation for deterrence. A dicey business always.

  6. Michael McCullough says:

    Although I want to be careful here not to mix up your conceptual distinction with mine, there is some connection between the two sets of concepts. Your distinction between contingent forgiveness (by which I take you to mean something like “I am prepared to raise my regard for you back to where it was before you hurt me and to foreswear retaliation if you do so and so.”) versus unconditional forgiveness (“I have raised my regard for you and will not retaliate for what you have done, no strings attached.”) is a different way of slicing up the conceptual pie. Of course people do forgive without strings attached, but how do they do it? I suggest that they use their “heads” (reasons) to fuel up their intuitive, built-in forgiveness systems, which evolved to respond to changed social realities (that is, an offender who has indicated that he/she is still valuable and safe). So, when you decide that you are going to forgive an offender for a religious, or ethical, or self-help reason (your religion tells you too, you espouse an ethical principle that tells you that “an eye for an eye leaves everyone blind,” or you simply want some peace of mind), you must give your evolved forgiveness circuitry something to work with. I think that this is why, for example, so much of the modern religious writings about forgiveness are really self-help books: Religious people hold principles (reasons) for forgiving, but to act on those reasons, they still need to put some fuel into their intuitive forgiveness engines to make it happen. To wit, the first modern self-help book about forgiveness (theologian Lewis Smedes’s Forgive and Forget), which some people credit with inspiring modern psychological research on forgiveness, is a self-help book! 


    Now, of your claim that unconditional forgiveness (in which, by my way of thinking about it, reasons motivate you to feed something useful into your own intuitive forgiveness systems, which enables you consequently to forgive without any changed social realities) is a more durable and genuine kind of forgiveness: Is that statement actually true? Personally, I have my doubts, but who knows? Perhaps some empirical data would help to settle that claim.