Defining Altruism

What happens when you scratch an altruist?

Are humans really altruistic?

Yesterday, David Pinsof (at Everything is Bullshit) posted an interesting, provocative piece on this timeless puzzle, asking if altruistic people truly want to make the world a better place—as they often say they do—or if they actually want to help allies and cultivate a good reputation.

If you “scratch an altruist,” do you “watch a hypocrite bleed,” as Michael Ghiselin suggested?

Altruism has vexed philosophers and poets since time immemorial. This post steps back from this debate and focuses on what altruism is, exactly, from a biological standpoint and humbly, but definitively, offers a solution to this mystery.

The debate among biologists has, in my view, historically been muddled because the definition of the word “altruism” has been muddled.

See, definitions are important. Right now, there is a fight about the definition of “woman” big enough to merit an entire movie about the question. And of course the question of what a “person” is—a zygote? a corporation?—is similarly fraught, and the answer can determine crucial issues such as when it’s legal to terminate a pregnancy and when you can drive in the High Occupancy Vehicle Lane.

In fairness, definitions are hard to get right.

According to Diogenes, when Plato—no fool he—defined man as “a featherless biped,” Diogenes brought him a plucked chicken to provide a counterexample. (And to make Plato feel silly, I assume.)

The Platonic man.

Counterexamples, in the form of plucked chickens or whatever else, matter. If your definition includes something that it shouldn’t, or misses something that it should, you don’t get to respond by saying c’mon, you know that’s not what I meant. A definition should mean what it says and say what it means, just like my great uncle Kolia always did.¹

Ok, so let’s look at reputable sources for the definition of altruism as used in biology. 

1. The Stanford Encyclopedia of Philosophy (the first result when you search on Google with the terms altruism, definition, biology) is as follows: “In evolutionary biology, an organism is said to behave altruistically when its behaviour benefits other organisms, at a cost to itself. The costs and benefits are measured in terms of reproductive fitness, or expected number of offspring.”

2. Wikipedia provides the following definition, distinguishing the biological meaning from the philosophical meaning: “In biology, altruism refers to behaviour by an individual that increases the fitness of another individual while decreasing the fitness of themselves. Altruism in this sense is different from the philosophical concept of altruism, in which an action would only be called "altruistic" if it was done with the conscious intention of helping another.”

3. Stu West, one of the most authoritative scholar in this area (in my biased opinion), says that “Altruism is when a behaviour reduces the fitness of the actor, but increases the fitness of the recipient.”

4. ChatGPT, when given the prompt “Please define altruism from the perspective of biology,” says: “From a biological perspective, altruism refers to behaviors that benefit other members of a species at a cost to the individual performing the behavior.”

From these definitions we notice a few consistencies. First, when discussing altruism, for some reason people favor the British spelling of the word behaviour. Second, and I cannot stress this enough, altruism is defined in terms of the behavior’s effects on fitness.

Ok, now look at the next picture and keep in mind Diogenes’ chicken.

Is something fishy about the definition of altruism?

This picture is supposed to be the classic shot of a fish swimming upstream to spawn and leaping into a bear’s mouth, but, you know, digital media rights, so I had an AI do it. You know the picture (video) I’m talking about.

Ok, with that preamble, let’s look at how the definitions above categorize the drama in that image.

The fish has leapt into the mouth of a bear. Let’s tot up the fitness consequences:

1. Fish Fitness – Decreased. Fitness reduced to zero as this fish will be digested and produce no more offspring.

2. Bear Fitness – Increased. Calories are fitness-good for bear.

Now, let’s check the conditions against the definitions above. Has the salmon behaved in a way that benefited the bear at a cost to itself? It seems so. The bear is better off and the fish is worse off. I don’t see any way around it. This behavior by the fish meets the definitions above.² We have ourselves an altruistic fish.

Here is another picture. Suppose that, as is likely, the events preceding this picture are the fly buzzing around and getting caught in the spider’s web. Insect worse off, spider better off. We have ourselves an altruistic fly.

An altruistic fly. (Photo by zibik on Unsplash)

Now, you can say that this was not the intent of the behavior, but that’s the whole point. The biological definition does not include intent. This is why I emphasized that the definition was about behavior. Because if the definition refers only to the fitness effects of behavior, it allows Diogenes’ chicken in.

Something is, clearly, wrong with the definition.

How to Fix.

Let us, for a moment, forget all about that and think instead about screwdrivers.

Here are two ways to define a screwdriver.

1. Something used to drive a screw.³

2. Something designed to drive a screw.

As you can see, one definition is in terms of what it’s used for—which is an analog of the behavioral definition—and one definition is in terms of what it’s designed to do.

Now, suppose you need to drive a screw into the wall to hang a new picture and, not wanting to fetch a tool from the basement, you take a dime out of your pocket and drive a screw into the wall with it. Under the first definition, you are now holding a screwdriver, which is absurd. If you go down to the basement and fetch a screwdriver, then you are (correctly) holding a screwdriver in your hand under definition (2), even if you then use the screwdriver to pry up floorboards. From this, we see that definition (1) is just wrong. More importantly, we have learned that using the notion that what a thing is designed for can help properly define hand tools. And not just hand tools, obviously. Anything that was designed—by a human artificer or the process of evolution by natural selection—to serve some function can be defined this way.

See where this is going?

My claim here is that it is possible to define something in terms of what it does or, instead, to define something in terms of what it is designed to do. This is the fulcrum on which my argument rests.

And, of course, you absolutely believe that it is possible to infer something’s function from its shape, whether the something is an evolved biological trait or a tool. If someone asked you if the eye is for seeing or grasping objects, there is just no way you would say, “I don’t know. I am, sadly, unable to infer function from form.”⁴

Ok, let’s get back to altruism.

Suppose we define it this way. An altruism device is one that is designed to deliver benefits to another organism. Just like a screwdriver is a tool designed to drive screws. An organism is said to behave altruistically when its behaviour—I’ll retain the British spelling, for fun—is produced by an altruism device, i.e., a system designed to deliver benefits to another organism.

This definition could be tweaked here and there, but let’s see how it does with the fish and fly cases.

In the salmon case, the design of the underlying systems—the muscles, fins, and other bits used to swim—are emphatically not designed to benefit bears. So this case is correctly ruled out. Similarly, the navigation systems of insects are just-as-emphatically not designed to lead to being trapped in a web. Also correctly ruled out.

Ok, so what systems are correctly included under this definition? Let’s take the most basic sorts of systems, parental investment systems. Penguins feed their chicks by regurgitating, a mammal delivers milk to a pup, cub, or baby with a specialized gland and tube, and so on across the animal world. These systems, which benefit offspring, are correctly captured by the definition.

Now note that this way of looking at things gives rise to a happy consequence. Whether something is an altruism device that causes altruistic behaviour is an empirical question. We can do research and try to discover the function of this behavior or that trait. We now have a nice, clean agenda for answering questions about altruism. Pish posh!

Why?

If this persuades you that design rather than behavior is the right way to build a definition of altruism, you might be thinking, hey, wow, that makes so much sense! Why hasn’t biology—and psychology—already come to this conclusion? The bear/salmon cases by themselves are just like Diogenes’ chicken; they illustrate the problem with the behavioral definition. What has been the problem here?

That is such a great series of questions. The short version is that I don’t know. If I had to guess, I would start with this joke about sheep.

An engineer, a physicist, and a philosopher are hiking through the Scottish Highlands. Coming to the top of a hill, they see a solitary black sheep standing before them. The engineer says, "Huh. Scottish sheep are black." The physicist says, "Well, some of the sheep in Scotland are black." The philosopher says, "Unclear. All we know is that at least one of the sheep in Scotland is black on one side."

The black sheep of the family.

In some circles, there is a status competition that is won by making the most conservative inference from observations.⁵ If you went to an archeology museum and saw an Egyptian comb from 1000 BC on display, you might say, “Wow, a comb” and your philosophy of biology friend might say, “Unclear. It might be a vase but a really, really crappy one.”⁶

I mean, sure. The philosopher could be right. A child might have been trying to create a vase but, being terrible with clay, wound up with a comb-shaped object. But the philosopher almost certainly isn’t right and everyone knows they probably aren’t. What are the odds? Of all the ways you might make a crappy vase, it happens to look like this thing that is useful for this other purpose? Very likely, it’s a comb and it was shaped into that form to do comb things to hair. If a demon came to earth and forced you to bet on whether the thing in the display case was a vase or a comb, which way would you go? Still, the status bonus from being super conservative in your inferences can be tempting, even to the point where you posit bicolor sheep and comb-shaped, crappy vases.

For some reason, this game is a special favorite in the world of evolutionary biology. I think a lot of this is due to a famous paper by Stephen Jay Gould and Richard Lewontin back in 1979. They make an argument about an architectural bit called a spandrel and afterwards everyone had to say, sure, there are five fingers and they seem to be really good as grasping things but how do we know that they aren’t there because people who had ten fingers were better at base ten math and they just happen, as an accidental but happy side effect, to be good for grasping?⁷

This little post is not the place to relitigate the fight that began back in the 70s. I do, however, suspect that the reason that the functional definition of altruism suggested here—which correctly excludes salmon/bear cases—has not been popular is because of the general aversion to functional language. Some follow Gould’s lead and exercise one-sided-sheep level skepticism about evidence of design/adaptation. Which means that “design” becomes something of a dirty word.

Conclusion

By defining altruism in terms of behavior and its effects, biologists run into the same problem that Plato did: obvious counterexamples. The adaptationist/functional definition avoids this problem.

The relentless logic of these simple facts will have no effect on how altruism is defined in the field in the next fifty years, mostly because ideas, once entrenched in science, are devilishly hard to expunge.

And now we know if humans are, at least in part, altruistic. We are, of course, selfish: we have many adaptations designed to benefit ourselves and advance our fitness interests. Additionally, we have lots of systems on board designed to deliver benefits to others (in certain ways and under certain conditions). Physiologically, as mammals, we have parental investment systems. Then we have systems designed for loyalty, friendship, love, and so on.

Humans are, then, altruistic in the sense that we’re designed to aid (certain) others. The fact that in the past the genes that gave rise to these systems did better because of this generosity does not change the fit with the definition.

As Plato would say, so it is demonstrated.⁸

Coda: Exercise for the Reader

Let’s take one more example, the well-known case of brood parasitism, most famously observed in cuckoos. Brood parasitism in cuckoos is a reproductive strategy in which cuckoos lay their eggs in the nests of other bird species, outsourcing rearing of their offspring to foster parents. This strategy relieves the parasitic cuckoo parents from the duties of building a nest, incubating eggs, and feeding and caring for the chicks, which are all carried out by the host birds. The cuckoo achieves this by stealthily laying its egg in the host's nest when the latter is away. The egg of the cuckoo often closely mimics the appearance of the host bird's eggs to reduce detection. After hatching, some cuckoos eject the host's eggs or chicks from the nest to eliminate competition and gain more parental attention and resources. The host species, deceived by the cuckoo chicks, feeds the parasites, meeting the definition of altruism. This case is included by the definition I’ve proposed above. Should it be? If not, how does the definition of altruism need to be tweaked?

1. Total straight-shooter, he was.

2. If you’re paying attention, you noted it doesn’t fit ChatGPT’s definition. GPT added that the act has to help another member of the same species. Here, getting eaten helps a member of a different species. Kudos on your attention to detail.

3. We can do this without using the words “drive” or “screw” if we want. For (2), we could say that it is something designed to turn the head of a small piece of hardware that fastens one thing to another,” or some such. My point here is about the used/designed distinction, so I’m not worried about the circularity here.

4. In some circles, drawing this inference is a whole thing—about which I’ve written in various places—but the best place to start is always Adaptation and Natural Selection by George Williams. If you don’t care about this inference, you can just move along.

5. A technical note. The tendency to help kin—offspring or other relatives—increases the replication rate of the genes causing this tendency. The logic of kin selection, then, explains why such systems evolved and, importantly, predicts that these systems will be designed to aid kin, narrowly, rather than other individuals, broadly. To evolve, adaptations must, as Pinsof highlights, generate fitness benefits for the genes that give rise to the traits. They can do so by many routes, one of which is to help others. This way of putting it sheds light on the “limits of altruism” question that David Pinsof discussed. Altruism systems must promote the replication of the altruist’s genes, one way or another, or they won’t evolve by natural selection. Whether that answer is grounds for cynicism depends on your definition of cynicism—another example of the importance of definitions.

6. Another dumb thing they might say is that the artifact is what’s left over when you make something else. It just happens to be really useful, by accident.

7. See George Williams’ remarks in the introduction to Adaptation and Natural Selection about foxes making tracks in snow for the OG take on side-effects and inferring adaptation when a side-effect explanation is available.

8. He might have pronounced it, “quod erat demonstrandum.”