Is Compassion Compassionate?

What I'm going to be talking about: 

1) What is compassion, and why does it matter?

2) Brief theoretical background

3) Reasons for thinking these ideas may be incomplete 

w/respect to compassion

4) Offer one possible solution to the problem

and consider its implications for how we're to

understand compassion

5) Delve briefly into what those implications

mean for reasoning

6) Suggest a possible design for testing these ideas

7) Further ideas and avenues for study

Any and all feedback is welcomed!

1) Common conceptions of compassion

Some of us probably remember...

1) Common conceptions of compassion (cont'd)


1) Compassion is necessarily a good thing - the more of it there is, the better the world will be.

2) Compassion is some kind of feeling that's generated when we want to help someone

What I want to focus on today is why I think both of these assumptions are wrong.

1) Why this matters...

If the goal is to make the world a better place by increasing compassion, but compassion has a limited (or conditional) ability to improve things, then our attempts to make the world a better place may be ineffective or even backfire and end up making things worse.

Human history is rife with examples of things we once thought were good but were actually harmful, e.g. tobacco, blood letting, mercury...

Ultimately, what I want to suggest is that we need to have a firmer understanding of what compassion is and how it operates if we hope to manufacture the conditions under which it produces optimal outcomes.

1) A working definition of compassion

So what then is compassion?

That's in part the question to which I'm interested in discovering the answer.

At present, I treat the word 'compassion' as 

a moral emotion designed to encourage one to display behaviors indicative of a higher WTR toward a target or group of targets.

We'll come back to why in a bit. But for now, let's review some of the academic backdrop to this discussion...

1) What is compassion, and why does it matter?

2) Brief theoretical background

"He who was ready to sacrifice his life, as many a savage has been, rather than betray his comrades, would often leave no offspring to inherit his noble nature."

-From "The Descent of Man," by some dude named Charles Darwin

If selection works by propagating those genes that are better able to spread themselves throughout a population, and if acting selflessly inflicts fitness costs on those who engage in such behavior relative to those who don't, how could the genes that code for selfless behavior ever hope to spread? In short, why are people nice to each other at all?

2) The evolution of altruism...

We can treat compassion as an analog to altruism: both presumably inflict fitness costs. Another way to conceptualize this: compassion is the emotion that tends to give rise to altruistic behavior i.e. behavior that supposedly inflicts fitness costs upon those who engage in it.

2) Solution 1: kin selection

John Haldane, R.A. Fisher, later John Maynard Smith and popularized by W.D. Hamilton

"It may be, for instance, that in respect of a certain social action performed towards neighbours indiscriminately, an individual is only just breaking even in terms of inclusive fitness. If he could learn to recognise those of his neighbours who really were close relatives and could devote his beneficial actions to them alone an advantage to inclusive fitness would at once appear. Thus a mutation causing such discriminatory behaviour itself benefits inclusive fitness and would be selected." 

-Hamilton, 1964

Hamilton's Rule: rB > C

All you need need is some method of recognizing kin.

Okay, that explains why people are nicer to kin, but then why are people nice to non-kin?

2) Solution 2: reciprocal altruism

Trivers: altruistic behavior that incurs fitness costs while conferring benefits on non-kin  can be selected for if there is an expectation of reciprocity. 

e.g. male bird nest defense, cleaner and host fish

Need a) opportunities for future exchange and b) some way of remembering those with whom you've exchanged and whether they were cheaters

 (some kind of cheater detection module would be helpful in this. If only there were some evidence for such a thing...)

2) But What about compassion?

The origins of compassion as a specific emotion have received relatively little direct examination, particularly from an adaptationist perspective.

I was able to find a few treatments of it, however...

Compassion "seems to be designed for motivating investing in social partners in need."

- this from a 2012 paper by some schmucks called Petersen, Sznycer, Cosmides, and Tooby

This makes sense: early human environments were composed of small hunter-gatherer bands. The protein calories necessary for survival and mostly produced by the spoils of hunting were scarce and variant -- sometimes hunts were successful, often providing more meat than a hunter or his family could use before it went bad; other times they were unsuccessful, leaving hunters and their families malnourished. Food sharing paradigms between hunter-gatherers likely evolved to buffer these calorie variances. And compassion -- an emotion motivating caring for the needy by sharing resources -- would have protected a hunter in the future had he or his family become the needy party.

2) Brief theoretical background

3) Some questions to consider...

1) If compassion is designed for investing in social partners in need, why are we nice to total strangers in one-shot encounters (well, some of us...)?

2) Suppose I told you two stories: person A recently lost her job when her company moved overseas. Person B worked at the same company and so also lost her job, but was also recently diagnosed with an incurable cancer and has 6 months left to live. And because she lost her job, she can't afford the drugs that will give her some semblance of comfort in her final days....

3) About Question 2...

Our response seems intuitive, and yet from an evolutionary perspective, it's kind of weird. Or at least it seems weird at first glance. 

If compassion is designed to motivate raising one’s WTR toward a needy (potential) social partner in the hope that the favor will be returned in the future, why would compassion increase the less likely that stranger is to be able to return the favor (whether because she dies from her disease or because she's so destitute, she’s never able to reciprocate to an adequate extent)? In other words, how can selection account for the existence of compassion for needy strangers given the Banker’s Paradox?

3) Possible answers

1) Maybe the mere possibility of our future interaction is enough to warrant a compassionate response (i.e. if the cost is low enough, and the benefits of potential future interaction multiplied by the probability of future interaction is high enough...)

But this response can't account for our answer to question #2...

So what can?

3) Reasons for thinking these ideas may be incomplete

w/respect to compassion

4) One possible solution: indirect reciprocity

"Individual selection can nevertheless favour cooperative strategies directed towards recipients that have helped others in the past. Cooperation pays because it confers the image of a valuable community member to the cooperating individual."

-Nowak and Sigmund, 1998

Required: memory and mental reputation-tracking software,  and an environment in which reputational information is easily shared

4) So why be nice to strangers/feel worse for the relatively less fortunate?

Because your behavior is being tracked. The knowledge of how you behave will have reputational consequences - fitness costs or benefits. 

The long-term benefits to being nice that would have tended to accrue from your sterling reputation may have outweighed the costs.

There's evidence this is still the case today:

"Here we show experimentally that donations made in public to a well-known relief organization

resulted both in increased income (that the donors received from the members of their group) and in

enhanced political reputation (they were elected to represent the interests of their group). Donations may

thus function as an honest signal for one’s social reliability."

 -Manfred Milinski, Dirk Semmann and Hans-Ju¨ rgen Krambeck, 2002 

4) Integrating indirect reciprocity with compassion

One of the questions posed earlier was "why do we feel MORE compassion for people the worse off they are?"

The proposed solution under consideration is indirect reciprocity - that the activation of our compassion and the behaviors that emotion tends to generate precisely when the target of those behaviors is LEAST likely to be able to reciprocate them sends the strongest signal to third party observers that we are honest, generous, potentially fruitful future partners worth cooperating with. And, as further motivation for third party observers to engage with compassionate actors, in the event those third party members find themselves in dire straights, they have insurance in the form of their new cooperative allies who've already shown themselves to be willing to confer fitness benefits on the least fortunate.

4) Implications for compassion

Recall the working definition of compassion I provided earlier: a moral emotion designed to encourage one to display behaviors indicative of a higher WTR toward a target or group of targets.

"display behaviors Indicative of a higher WTR," is not necessarily synonymous with "behave in ways that confer benefits."

IMPORTANT: if compassion for needy strangers evolved through indirect reciprocity, then its function was not conferring benefits per se, but was instead to send signals of one's cooperativeness to third parties.

4) Implications for compassion (cont'd)

4) Implications for compassion (cont'd)

Therefore, distinguishing between two kinds of compassion is warranted (I think): 

1) Compassion designed to confer benefits (to kin, close social partners, e.g. spouses) and

2) Compassion for needy strangers - designed to signal generosity (i.e. commonly conferring benefits as a MEANS, not the end in and of itself).

4) Offer one possible solution to the problem

and consider its implications for how we're to

understand compassion

5) Reasoning and compassion

What "reasoning" means is a whole other can of worms..

What I mean by 'reasoning' are the mental states that serve as the connective tissue between feeling compassion and actually engaging in the behaviors that feeling encourages (i.e. conferring benefits, signaling generosity).

If you ask them, people will give you "reasons" (i.e. their reasoning) for why a particular person or group for whom they feel compassion deserves their help and support. Even in the person A and person B example from before, people would probably "reason" that person B is MORE deserving of support than A. And yet that is a feeling (or so my proposal goes) that is generated by indirect reciprocity - by reputational considerations - and so has nothing whatever to do with who is more "deserving" of help. The reasoning that one is more deserving than another is a post-hoc justification for an intuitively generated emotional response.

-The social intuitionist model of moral reasoning.

5) Example

Rent control laws - compassionate reasoning: rent prices are often too expensive for the less fortunate. Can't we do something to help them? Intended to help, made things worse.


1) Compassion seems to be an emotion designed to motivate behaviors indicative of a higher WTR toward a needy individual or group

2) We feel compassion for needy strangers the worse off they are -- this can't be accounted for merely by kin selection or some kind of direct, single-pair dyadic reciprocal framework (A help B, B helps A)

3)  Possible explanatory model: A helps B, despite no expectation of direct returns on A's investment from B. This "compassionate display" serves as a signal to C, D, and E that A is a generous and trustworthy potential cooperator, leading to further cooperative relationships that tended to benefit A more than it cost A to send the signal.



Higher compassion for needy strangers the worse off they are can only be explained by the signal input variable. 

Prediction: If the behaviors generated by compassion for needy strangers are driven in part by how those behaviors appear to third-party observers*, then conceivably compassion could generate harmful behaviors as long as those behaviors appeared to be beneficial (i.e. to signal goodwill).* *

*The worse off the target of my compassion is, the greater likelihood that compassion is generated by the weighting of the signaling component.

**The farther my signaling reaches, the more I'm incentivized to weight signaling, because the more benefits I can potentially obtain from relationships with third parties.


Example: Rent Control


People do reason about/are conscious of feelings that produce "compassionate" outcomes/behaviors. If the compassion mechanism is well-adapted, that reasoning will be a post-hoc justification for a deeply-felt intuition.

 If we create a scenario both to test people's reasoning and invoke feelings of compassion in which the outcome "confer benefits" conflicts with the outcome "signal goodwill," then such a well-adapted mechanism may bias reasoning to choose the latter option.

6) Experimental design

Controls: numeracy (A bat & ball cost $1.10 in total. The bat costs $1 more than the ball. How much does the ball cost?), political ideology, gender, age, education, etc., the strength of several emotions (shame, guilt, anger, happiness, etc.)

Independent variable: Compassion

Dependent Variable: the targets' answers to a math problem designed to generate feelings of compassion (CMP, for compassion math problem)

The answer to math problems are objective, and if there are only two answer choices to choose from - one that confers benefits relative to the other, and one that inflicts costs relative to the other - then we should be able to glean whether higher compassion accounts for those who are more likely to select the wrong answer under conditions in which the wrong answer serves as a "compassionate display of generosity."

6) Experimental design (cont'd)

Separate people into 4 groups: two groups take a standard math problem (SMP) designed to invoke no feelings of compassion and to which the right answers will vary.

6) Experimental design (cont'd)

6) Experimental design (cont'd)

The other two groups take CMPs designed to invoke feelings of compassion and to which the right answers will also vary.

6) Experimental design (cont'd)

6) Experimental design (cont'd)

Other possible CMP vignettes: minimum wage, Hiroshima, Israel-Palestine, sweatshops & drug prices (Molly Crocket), animal welfare...

6) Experimental design (cont'd)

I predict that more compassionate people will be less likely to get the CMP correct when the right answer conflicts with the display of generosity, that they will either be more or equally likely to get it right when it does not.

6) Suggest a possible design for testing these ideas

7) Potential further ideas

If compassion for needy strangers is motivated by signals to third parties, then we might expect for its effects on reasoning to be attenuated by anonymity/raising the signaling threshold.

If the distinction between kinds of compassion I proposed is justified, then supposing we gave targets a CMP wherein the person or group of people affected (i.e. to whom one choice will confer benefits relative to the other) are their kin or close social partners, we should expect to see the bias toward choosing answers that display, rather than confer, generosity disappear.

When the stakes are high (i.e. those who stand to lose or gain are your close social partners on whom you depend or your direct kin with whom you share some non-insignificant percentage of your genes), it matters a whole lot more that you choose the right answer - and confer benefits

So is compassion compassionate?

I think the answer, as with most complicated things, is...

it depends.

It may have been that in our ancestral past, "signaling generosity" was more likely to confer benefits... in modern times, there may be more opportunity to signal compassion without conferring the requisite benefits and/or more opportunity to signal far and wide, leading some of us to adopt a weightier signaling strategy...


Thanks for listening.

Create a presentation like this one
Share it on social medias
Share it on your own
Share it on social medias
Share it on your own

How to export your presentation

Please use Google Chrome to obtain the best export results.

How to export your presentation

Copy - Bunkr Samples

by mitchelllanders


Public - 5/18/16, 3:14 AM