What is Dunbar's Number?

The term "Dunbar's Number" refers to a particular hypothesis by primatologist Robin Dunbar. It is a very simple idea with rather complex implications, and it is one of those simple ideas that gets more complicated than ideal as we look into it more and more. Eventually, the idea is required by many who contemplate it to do more work than was ever intended, and in this way seems to fail, though it really doesn't. I personally think Dunbar's number is useful if it is properly understood, so I want you to give it a chance, and to help you do that I'd like to use an analogy.

I'm thinking of a number called Carrier's Number. Carrier, in this case, refers to the company that installs air conditioners and heaters. Carrier's number is the temperature in degrees F at which you, sitting there in your chair, notice it is too warm so you get up and go turn on the air conditioner. It is best measured as a post hoc number...we watch and wait, flies on the wall, as the room heats up, and when a person gets up and flips on the air conditioner, the temperature at that point was carrier's number for that person at that time.

One might argue that a post hoc measure like this isn't much use in science because in science we like to predict things. But just because carrier's number is best measured pot hoc does not mean that it only exists post hoc. It existed before the test subject got up, we just didn't know what it was. For a large number of test subjects, we should be able to estimate carrier's number (it is probably in the upper 70's F). However, this will vary across cultures, across seasons, humidity, as clothing styles change (in the days of Polyester Leisure Suits, it is said that Carrier's Number went down by about four degrees) and so on. The fact that it varies does not make it a bad number. In fact, its variation and reasons for it can make it an extra good number depending on what one is trying to do with it.

Dunbar's number is the number of full blown social interactions you can manage. This number is lower or higher across species of social primates, as it tracks adaptive suites of sociality and the ability of brains to manage sociality. So, you can measure Dunbar's number across primate groups by looking at how large effective primate groups get across species and figuring that the number is just about that maximum group size. Or, you could estimate Dunbar's number (retrodict it, as it were) by looking at relative brain size, if we assume that brain size is linked to Dunbar's number, all else being equal. In this way, Dunbar's number is a way of linking primate sociality with brain evolution, which was the original idea.

In modern society, and in human historical contexts, we may see Dunbar's number in a lot of places. This is the number at which, more or less, groups start to break down (in some societies) and villages split. Military units max out at about Dunbar's number (companies are about 100 in size) and so on. This does not mean that Dunbar's number and its associated dynamics explain everything. It might mean that the breakdown of social interactions can be more important than, say, resource limitations, on human group fission and fusion. That is exactly what many anthropologists have been suggesting for decades. Dunbar's Number is simply this concept quantified somewhat and expanded to primates.

There are variations and adjustments. Some organisms have apparent smaller brain size because their diets cause a different body size, so that has to be adjusted for (leaf-eating monkeys may small relative brain size because their bodies are large, not because their brain is small, for example). What a fully blown social interaction is may vary. A group of primates may have a subgroup that hardly ever interacts with the others. Perhaps pre-adolescent monkeys don't count for as much as sexually mature monkeys, so if there happens to be a baby boom a couple of years back, the group size if you count everyone is higher than Dunbar's number. Or perhaps the group includes two or three social geniuses who temporarily facilitate an extra large group size, or temporarily force an extra small group size, for some reason.

It makes sense that there is a limit on effective sociality, and thus, on effective social group size. Dunbar's number is nothing other than the number you end up with because when you are making the damn graph you need a damn number to put there on one of the axes. It has been over-interpreted or over-used as a number like many of those from Physics, like the freezing point or boiling point of water, which it is not.

Desiree Schell and I spoke about Dunbar's Number on the Skeptically Speaking that just became ready for you to download. Check it out here.

This video includes, during the last third somewhere, a discussion by Dunbar of all this.

And, here's a few items by Dunbar you might find interesting:


More like this

Dunbar's number is usually given in the range 100-300, which seems both too large and too small. Too small because one can recognize 10,000-100,000 different people, but one can only have a "full-blown" social interaction with a few people, maybe 10, if you're lucky.

If you work at an organization of, say, 200 people, you might have a complex relationship with a handful, but I don't believe that 150 is credible.


BKS, those are rather spectacular claims you've made! I can recognize 30,000 people but I can only manage relationships with a "handful" (one handful = 6-8, right?)

Actually, if you examine closely the primate data which he used to derive the number, you will find that he became more and more selective about which primates to include (New World Monkeys got dropped for example--remember capuchins are the most encephalised primates after humans). So the original base-line for the calculation of the number is, shall we say, groomed.
Then the use of it is rather inconsistent. Other primates have groups which can be counted by the primatologist in the field by the amount of time they spend together and particularly the number of other animals they spend grooming. I suspect that for most people the number of 150 derived from the equations Dunbar created from the primate data is way too large. We do not regularly interact with that number of people. BUT, and it is a big but, we do maintain knowledge of that size of a group of people often by representing their role in our lives as a symbolic fact. This is something we can do because we use language. And that is why Dunbar's argument started off. He was using it in an argument about the emergence of language, but I suspect the logic of that argument is the wrong way around too. But, hey, it is really interesting and I wish I had thought of it.

By Iain Davidson (not verified) on 10 Jun 2012 #permalink

I graduated from a high school with about 110 students. There were 23 in my class and I knew them all. I knew most of the people in the next lower class, but only a few people out of the two lower classes. So I probably knew about half the students in my small high school.

By Jim Thomerson (not verified) on 10 Jun 2012 #permalink

I don't think the claims are spectacular. I don't believe most people have complex relationships with more than three or four people at a time (in the same month, say), But I'm pretty sure that the number of faces that we can recognize (correctly remember having seen before) is around 100,000. I remember this from a Psych class where it was remarked that this is, not coincidentally, about the size of a Greek City State. This is not my field of expertise by any stretch of the imagination, but I don't believe 95% of the population has a "full-blown" relationship with 100 people. No way, no how.

That's my story and I'm sticking to it.


Metacomment: The timestamps in this Blog don't have time zone information. Not uncommon, but a disappointment.


As a mathematician with a pathologically poor memory for names and faces, I'd be more interested in establishing a least upper bound for this number for any given species.

By Ian Kemmish (not verified) on 11 Jun 2012 #permalink

bks, I the spectacular claim is that we recognize 30,000 people. I suppose it depends on what you mean by "recognize."

Dunbar is not talking about the sort of relationship you are referring to, I think.

Iain, one of the problems with Dunbar's first paper in 92 (IIRC), or perhaps the conference papers that got circulated before that, was that everyone complained about the groups of primates he included. I think there's a good argument for treating NWP and OWP separately, given that they have a very deep evolutionary split. With respect to social evolution, most (with adjustments) OWM's and apes are polygynous and have fission fusion group structures, etc. while most NWM's are monogamous or polynadrous and/or have much tighter group structure. I would argue that a "number" that worked for both groups is spurious.

we do maintain knowledge of that size of a group of people often by representing their role in our lives as a symbolic fact. This is something we can do because we use language. And that is why Dunbar’s argument started off. He was using it in an argument about the emergence of language, but I suspect the logic of that argument is the wrong way around too. But, hey, it is really interesting and I wish I had thought of it.

Right, exactly.

Actually, Iain, if you think Dunbar's number is too big, you can propose Davidson's Number of 149. Then take bets on it, you'll probably win!

It is worth adding in here that Dunbar does acknowledge different sizes of "circles of intimacy"--5, 15, 35, 150, 1500 (Dunbar, Barrett and Lycett Evolutionary psychology p.97) and famously showed that conversational groups split in to smaller groups whenever there are six people together. He has also been very inventive in trying to show the 150 group size by looking at christmas card lists and things like that. I am not sure whether FB friend numbers mean anything, though--look how many friends Greg has!

By Iain Davidson (not verified) on 11 Jun 2012 #permalink

Facebook allows one to make smaller lists, as does twitter, etc. I assume Robin has students sampling social networking sites to see if these working lists are about 150 in size!

The details may not be as "precise" as people like to claim but the IDEA of this is interesting. Dunbar has pointed out how the various "circles of intimacy" crop up in all sorts of human organisations. It makes a kind of sense for any creature that lives in a restricted social group to have relationships defined by a series of concentric circles, with it's closest relatives at the centre.

What is really interesting is how, throughout history, broadly similar numbers turn up, time and time again, in the organisations of military units. Even today infantry are organised into fire team / squad / platoon / company / battalion, roughly corresponding to the five lowest "circles of intimacy". Coincidence? Or primate psychology?

By Engineering Animals (not verified) on 12 Jun 2012 #permalink

These authors claim to have verified it using Twitter data:

"In this paper we analyze a dataset of Twitter conversations collected across six months involving 1.7 million individuals and test the theoretical cognitive limit on the number of stable social relationships known as Dunbar's number. We find that the data are in agreement with Dunbar's result; users can entertain a maximum of 100–200 stable relationships."


By Oni Musha (not verified) on 12 Jun 2012 #permalink

An interesting social media app could be constructed exploring the Dunbar's number idea. Something integrating facebook and foursquare-like "check-in" apps, wall-to-wall postings, "likes", etc. Obviously this would have to be incredibly opt-in or voluntary, as most of those idea go a little beyond Orwellian. Can't have big brother mother* watching too closely.

*sorry; I saw Roger Waters' 2012 production of The Wall last night and am feeling a bit anti-establishment.

What is a "full-blown social interaction"? Has it been defined in a way that is generally accepted? I imagine it's highly subject to all sorts of hedges, hems, and haws. What is a meaningful social interaction for an extreme introvert? Is it the same as for an extreme extrovert? At what point does a social interaction become meaningful or full blown? I sense an greased pig.

Thanks! Very helpful description!

By Jane Campbell (not verified) on 24 Jul 2012 #permalink