Sunday, 26 October 2014

The genetics of altruism

Introductory waffle and warnings

This is very much a work-in-progress blog post, written in a bit of a rush. I hope to edit it ‘sometime’ to make it more clear, readable, and engaging, especially for non psychology students. This week’s topic on my “Altruism and helping behaviour” module is the Biology of altruism. Ideally, I would write at least four blog posts for it: this one; one on twin, adoption, and family studies; one on evolution; and one on brain stuff. There’s no way the space-time continuum will facilitate me doing all four this year, so let’s crack on and see if I can manage at least one decent draft.

I used to think that psychology got itself into serious unnecessary muddles because of vaguely defined, inconsistently applied, ill-chosen jargon. I also used to think that biology – a proper science and everything – would not be like that. I no longer hold one of these beliefs.

Below I try to explain some basics of genetics. I have never been taught genetics and haven’t studied biology since I was 12, by which time I was already well out of my depth. Be critical and sceptical about everything I say below – and indeed everywhere. But don’t be nasty. There’s rarely a good reason to be nasty.

I’m not a Gene Genie


Below is a representation of some DNA. The famous “double helix” is shown on the right. The “ladder” picture on the left is more useful for current purposes.

Some DNA

The “rungs” of the ladder are called “base pairs”, mainly because they are pairs of “bases” (a.k.a. nucleotides).

Bases are chemicals and in DNA there are only 4 types. These are summarised by the first letter of each chemical: A, C, G, and T. These are represented in the diagram above in red, yellow, green, and blue, respectively. (If you see a “U” when reading about bases somewhere else then something called “RNA” is being discussed and things are probably getting more complicated than necessary for current purposes.)

Bases almost always team up with a particular complementary base. They’re kind of cute like that. A and T almost always hook up together and so do C and G. If something else happens, biologists call the process and its result an example of “mutation”. Frankly, I’m surprised they had the restraint not to call it an “abomination”.

If any section of the ladder is ripped in half lengthways and some appropriate spare material is about, it is possible to make two identical copies of the original section. Each half-rung is simply completed with its complementary base: Ts are added to As, As to Ts, Gs to Cs, and Cs to Gs. Mutation aside, this happens every time the body makes a new cell. (There are approximately 37 trillion cells in the human body, almost all of which houses its own copy of the entire body’s DNA, which itself contains 3 billion base pairs in a particular order. If that doesn’t impress you when you really think about it, you are hard to impress.)

DNA section split

The term “gene” is used rather chaotically. I will try to differentiate distinct uses of the term in the hope I don’t make the situation even worse. If I fail, please forgive me.

“Protein genes”

What I am calling “protein genes” are specific lengths of DNA half-rungs.

A “protein gene”

The half-rungs of a protein gene act a bit like a blueprint that is used to bring about a protein. Some sets of three adjacent bases, known as “codons” or “triplets”, provide blueprints to make specific “amino acids”. Other codons provide instructions about how to combine chains of amino acids into proteins and about how to combine proteins to make more or less everything of importance in the body (with obvious exceptions such as food and aliens). If every set of three half-rungs coded (provided a blueprint) for amino acids, a protein gene could be represented a bit like this:

 In facilitative conditions, a protein gene will result in a particular protein
The illustration above of a protein gene has a particular sequence of bases. Reading from the top down, its bases are AACCTGACT … GACCTGATT. Each protein gene ‘codes’ for a single protein but varies in how many sequenced bases it has. Human protein genes have sequences of between about 27 thousand and 2 million bases. Special triplets of bases called “stop codons” indicate that either end of a protein gene has been reached.

It takes two

Each half-rung on the DNA ladder is actually a double half-rung. (I know, but try to stay with me. This really is important.) By this I do NOT mean a base-pair of two complementary half-rungs. I mean each half-rung can be thought of as comprising two half-rungs hugging each other. If it helps, imagine gluing together two more-or-less identical ladders laid one on top of the other and then splitting the combined ‘double-thick’ ladder lengthways. Each half-rung of each double-thick half-ladder will comprise a half-rung from one of the original ladders glued side by side with the corresponding half-rung of the other ladder.

The protein gene depicted above is actually better represented by making clear that there are two parts to every half-rung, so it looks a bit like this, with each letter in the top row being connected to each corresponding letter underneath:


If this was your genetic code, one letter would correspond to the equivalent one on the half-rung you inherited from your biological mother and the other would correspond to the equivalent one on the half-rung you inherited from your biological father. For convenience, let’s pretend that all the top-row letters were inherited from your biological mother and all the corresponding bottom-row letters were inherited from your biological father. (A more accurate but difficult representation would jumble them between the top and the bottom.)

Shared genes

In the illustration I just used, all of the letters for each half-rung are identical for the bit inherited from your biological mother’s DNA and the bit inherited from your biological father’s DNA. This is mostly because your biological mother’s DNA is or was very, very, very similar to your biological father’s DNA.

When two or more living things’ DNA is very similar, those living things are often said to “share” their DNA, or have “the same” DNA. To avoid becoming disastrously confused later (when considering evolution), remember that different living things neither share nor have the same DNA as each other, although they do often have very, very similar DNA.

The most likely reason that your biological mother and your biological father each had an “A” representing their first half-rung above is because most people in the world probably have an “A” half-rung at that point in their DNA. The vast majority of people’s DNA is close to being identical to every other person’s DNA.

Estimates vary (for various reasons) but it is common to hear that well over 99% of any human’s DNA is identical to the DNA of any other human. For almost their entire genome (i.e., the whole of their DNA), all humans have the same bases in the same places.

The same is true within all species. Each living thing’s DNA is nearly identical to the DNA of every other living thing within the same species.

Similar is true across species. Compare any human, chimpanzee, bonobo, gorilla, or orang-utan with any other member of these great ape species and their DNA will be at least 93% identical. This makes sense. Lots of DNA will serve similar purposes no matter which great ape it is in, e.g., to serve as blueprints to make two legs, two arms, two eyes, etc.

Large parts of humans’ DNA are more or less identical to comparable bits of the DNA of mice and fruit flies. This makes it possible to study DNA and its effects in human-relevant ways that are much cheaper and less ethically contentious than might otherwise be the case.

“75% of our genetic make-up is the same as [that of] a pumpkin” (Link).

Statistics such as the ones above are often used to reinforce the truth that humans are animals and much more similar to non-human living things than we sometimes realise or remember.

The corollary point is of course also true: living things that have a lot of identical DNA can be enormously different to each other.


More than 99% cent of DNA being identical across humans means that any person’s DNA differs from another person’s by considerably less than 1%.

The bit of DNA that varies across humans is referred to as “polymorphic”, meaning it can take more than one form. Reiterating a point made just above, tiny DNA differences across people can have enormous effects on their bodies and therefore often their behaviour.

When differences in DNA sequences across people affect the form or function of their protein genes, people are often said to be “genetically different” from each other.  This can be confusing. A better phrase would be to say that people in this situation have different types or alleles of the same gene (which results from differences in the specific patterning of the DNA sequencing within the gene in question).

Differences across humans in a single half-rung of DNA can have important consequences. When 1% of humans or more have been found to have importantly different half-rungs, those half-rungs are said to cause (or be) “Single Nucleotide Polymorphisms”: “SNPs” or “SNiPs”, pronounced “snips”.

Here is a representation of four people’s DNA sequence for a protein gene that has alleles (different versions) resulting from a SNP:

Illustration of an allele resulting from a SNP
Notice that the 6th base from the left differs across all 4 people, sometimes because of the bit inherited from the mum, sometimes because of the bit inherited from the dad, and sometimes both. Because this has an effect on the protein coded for, this means that each person has a different allele of that protein gene. Using the base differences to identify the alleles, these are GG, GA, AG, and AA. (Remember that, mutations aside, these pairs of letters refer to the two stuck-together bits that make a single half-rung of the thick half-ladder; one from mum and one from dad.)

SNPs and behavioural differences

Many of the physical differences between humans result from differences in bases at particular DNA locations. Some of these occur because of the effects of one or a few SNPs, e.g., eye colour. Others result from the combined effects of lots of SNPs, e.g., height. But don’t forget that the environment always plays some role and it can play an enormous role. I don’t care how many “height genes” you have; swimming with hungry sharks is likely to take you down a peg or two.

Some SNPs (or combinations of SNPs) have mediated effects on behaviour because they affect physical characteristics that affect behaviour. SNPs that make people relatively likely to be relatively tall also increase those people’s likelihood of playing basketball at certain times in their lives, especially if they also have the “male gene” and the “American gene”. (I’m joking at the end of the previous sentence, of course. You did realise that, right? I am not doing so “randomly”. I do so to make a point. SNPs that can affect one or more characteristics are not SNPs “for” any of those characteristics. SNPs that can affect height do not result in genes “for” height. Still less do they result in “basketball genes”.)

The more closely tied a behaviour is to physical differences between people, the more likely it is that differences across people engaging in that behaviour are partly the result of SNP differences across those people. People who do and don’t do yoga are likely to have identifiable differences in their DNA (and also in their age, geographic location, social status …).

But is something like this true for “altruism”, as I use the term? Are differences in genetic code across species partly responsible for cross-species differences in having and expressing concern for the positive welfare of others? Similarly, are differences in SNPs partly responsible for differences across people in the extent to which they have and express concern for the positive welfare of others? I think the answer to both questions is almost certainly “yes” but I must leave saying why for other posts.

The question I shall move towards answering here though is, “Are there any identified SNPs which affect individual differences (among humans) in concern for the positive welfare of others?” Are there any genes such that people with one allele of the gene are more likely to manifest some discernible form of concern for the positive welfare of others than are people with a different allele of that gene? Has anyone proposed a candidate for the (inappropriate) title of “The Altruistic Gene”?

“The altruistic gene”

Seekers of the altruistic gene have typically not worried too much about what altruism is or how it manifests itself. Instead, they have tended to identify individual differences in some behaviour and claim that it is a good marker of altruism, e.g., giving away some of the money an experimenter gives you after she invites you to share it with a stranger and then watches to see what you do.

When looking for “giving money away in dictator game” “altruism” genes, researchers also often look for alleles that cause individual differences in something physical that (at least sometimes) correlates with their behavioural marker of altruism. Thus, researchers look for SNPs associated with individual differences in the production or regulation of some or other alleged “altruism drug”, such as dopamine, serotonin, testosterone, or vasopressin. The most heralded potential altruism drug in recent years has probably been oxytocin.

I haven’t the time (or frankly the inclination) to look at all the research in this area. Instead, I shall discuss a single study that seems to be fairly representative of such research.

“The (allegedly-altruistic) OXTR gene”

Oxytocin is a hormone with all sorts of effects. How to best categorise those collective effects is a matter of some debate but oxytocin seems mainly to sensitise people to social cues and often accentuate or otherwise modify their responses to them.

People have oxytocin receptors (OXTRs) throughout the body, as might be expected when oxytocin levels have such divergent effects. One SNP involved in the construction or regulation of oxytocin receptors is polymorphic: it can take different forms that influence the nature of the gene that results. This means that the OXTR gene has alleles.

Because the DNA variance under discussion occurs in a single SNP, the OXTR gene’s alleles can be described using the possible base combinations in that SNP. These are AA, AG, GA, and GG.

Participants in a study by Kogan et al. (2011) watched short and silent video segments of each of 23 people whom I shall refer to as the “lovers”. In the video clips, each of the lovers was shown listening to his or her romantic partner who was talking about “an experience of personal suffering” (p. 1910). Research participants then rated how “trustworthy, compassionate, and kind” each lover was (p. 1911). Let’s keep things simple and call this a rating of how compassionate or altruistic the lovers were thought to be.

Lovers were tested for which OXTR allele they had.

Those with the GG allele received slightly higher average ratings for compassion than did those with any of the other alleles. Furthermore, of the 10 lovers with the highest compassion ratings, 6 had the GG allele. Of the 10 lovers who received the lowest compassion ratings, only 1 had the GG allele.

The experimenters had an extra couple of people watch the video segments and rate each lover’s non-verbal behaviour: how many times they nodded, extent of eye contact, openness of arm posture, and whether or not they smiled. These rating were then combined into a single “affiliative cues composite”. 

Lovers with the GG allele were judged to display more affiliative cues than those with the other alleles.

Mediation analysis suggested that lovers with the GG allele were perceived to be more compassionate than people without the GG allele because the former displayed more affiliative cues than the latter.

Thus, Kogan et al. (2011) provided evidence that some differences across people in how compassionate they are may result, in part, from “genetic differences” between people which affect their display of affiliative cues.

A critical point

One of the things I care most about as a university tutor is encouraging students to evaluate evidence and develop arguments towards a justified conclusion. One major obstacle to this occurs when people believe without question what they read in the (mainly titles, abstracts, and summaries of) scientific literature. Sometimes it feels like all I do is shatter my students’ beliefs. This can be wearing; for them and for me. Nevertheless, I would rather they adopted a Socratic acceptance of ignorance than a set of Emperor’s false beliefs. To that end, here are some things that people might like to think about if they are tempted to report findings from Kogan et al. (2011):

1.      How confident can we be that the tales of personal suffering lovers were listening to were more or less comparable, e.g., in terms of how intense the described suffering was or how distressed people were when talking about it?
2.      Is it possible that people with different OXTR alleles attract different sorts of romantic partners, who perhaps have different experiences or ways of talking about such experiences?
3.      Was perceived compassion measured on a single rating of “trustworthy, compassionate, and kind” or the average of three separate ratings of “trustworthy”, “compassionate”, and “kind”? Does it matter either way and does it matter if you can’t find out from the paper?
4.      Did this scale/these scales run from 0 – 6 or 1 – 7? (Answer given below – but not in the paper.)
5.      Did the scale/these scales indicate how compassionate the lovers were thought to be in the specific situation they were judged, in general, both, or neither? How reliable and valid a measure do you think the scale was/these scales were of whatever your previous answer suggested?
6.      Lovers with the GG allele received an average compassion rating of 4.21and lovers with the other alleles received an average compassion rating of 3.80, a difference of 0.41 on a rating scale which ran from 1 = “Not at all” to 7 = “Extremely”. Impressed?
7.      What do you make of the fact that, of the 10 lovers with the highest compassion ratings, 4 did not have the GG allele and of the 10 lovers who received the lowest compassion ratings, one did have the GG allele?
8.      Assuming that the study’s results were persuasive in all respects, would they persuade you that people with the GG allele are relatively likely to be highly compassionate, that people without the GG allele are relatively likely to be without compassion, both, or neither? If you think the first two options are identical, think again.
9.      Do different OXTR alleles lead to differences in compassion, differences in how compassion is expressed, or differences in affiliative behaviour?
10.  Allele effects were examined among a sample of (only) 23 lovers. How reliable and generalizable does that seem to make the study’s main findings? Hint: See here
11.  How many lovers had each allele? When you’ve answered this question, maybe revisit the previous one.
12.  Were differences in scores on the “affiliative cues composite” mirrored by differences on each component of that composite? Why might it matter?
13.  If differences in oxytocin levels affect differences in a particular behaviour on a particular occasion, does this mean that oxytocin differences (a) always result in comparable behaviour differences or (b) are always involved when there are differences in the behaviour? (Hint: Adding a little salt can change the way something tastes.)
14.  Hopefully you will by now be asking yourself all manner of additional pertinent questions.
15.  When I emailed Kogan with questions similar to some of those above (and told him that I was doing so in preparation for this post), he graciously replied and admitted, “In reality, I'd say the [OXTR] gene is associated with slightly less prosociality among A carriers who are caucasian--it gets a lot dicier when you look at other ethnicities. But the effect is almost certainly small (we got lucky in our paper that it was as big as it was!).”

I’m not picking on this paper, particularly. Most of the papers I look at in detail when they claim a link between some or other “gene” and “altruism” are often even less compelling. As I said before, this can be wearing for all involved. Sorry.


The phenomenon I am interested in is a sprawling shape-shifter. People can have and express concern for the positive welfare of others in a multitude of ways and accordingly it can be difficult to identify reliable and valid indicators of even specific instances of such concern. Lots of characteristics influence whether and how a person will engage in even a particular instance of altruism.

Being compassionate when listening to a loved one report a distressing incident is not a single, simple, or uniform activity. It involves attending to the other; trying to understand what they are saying; trying to understand how they felt during the recollected incident; trying to understand how they feel right now; perhaps trying to work out how they might feel in the near and more distant future, perhaps as a result of things that you consider doing; wondering what you can do that would be in their best interest and not make things worse for them; managing one’s own thoughts and feelings in service of the other’s welfare; etc. How such compassion is expressed and how it can be most effective depends on one’s own abilities and preferences; what the other person likes and responds well or badly to; the specific setting (e.g., who else is around and what role they seem likely to play); etc.

I therefore do not expect simple relationships between any SNP and any single manifestation of altruism; still less any and all instances of altruism. This is not because small DNA differences can’t have huge physical and behavioural consequences. I think they can. It is because I think that altruism is too complex a phenomenon to be well predicted by a single simple indicator, whether that is an SNP or levels of some biochemical such as oxytocin.

Any search for “the altruism gene” seems to be to be highly suspect and any claim to have found it considerably more so. Whenever I have looked in detail at the evidence for such a claim, I have always found it to be seriously wanting. Given what I have already said, it would be astonishing if it were otherwise. From my point of view, studies in this area cause confusion more than they further knowledge. They give people the illusion of evidence for their false intuitions and they require a lot of time and energy to try to counter (or even to ignore).

Maybe time and good science will prove me wrong. Until then, I advise extreme caution (to the point of quite radical scepticism) in response to any claim to have identified “a” let alone “the” altruistic gene.

Prelude to later posts

As I predicted, I have run out of time long before I have run out of things to say. For now, I recommend that any of my students thinking about writing an essay using material from this week should seriously think about the following:

·        Many “genetic relatedness” studies show links between (unspecified) “genes” and individual differences in “altruism” within the sample studied. Don’t mix up potential causes of a specific instance of individual differences in altruism with reasons for group (including temporal) differences in altruism; and still less with causes of altruism. Do consider how much “non-genetic” variation there was within any given sample and how well that represents “non-genetic” variation beyond it. (If there is limited environmental variance and individuals’ differing life-experiences are not measured, of course differences in “genes” are going to explain the lion’s share of whatever differences there are in the criterion variable.) Do think carefully about the indicator of altruism and its relation to concern for the positive welfare of others.

·         Most discussions about “biological altruism”, “evolutionary altruism”, “reproductive altruism”, and the like are about a completely different phenomenon to the one studied on this module. Roughly speaking, that phenomenon concerns a living thing having a characteristic which, over the lifespan, has the consequence of being ‘costly’ specifically in the sense of curtailing how many copies of its own “genes” exist in subsequent generations. (“Genes” is in quotes because the term is used in a different way than described above for “protein genes”.) Such a hypothetical phenomenon could, if it existed, seem to cause a problem for Darwin’s theory of (individual level) natural selection. Evolutionary arguments that claim to solve “the problem of altruism” do so by explaining (away) that sort of so-called altruism (which would be better termed something like “DNA-reproduction curtailing”). Because they use such a stupid term for the phenomenon they are most interested in, the evolutionarily-inclined often get themselves and others into terrible and massively costly (in real terms) messes. Discussions about such things as “group selection”, “kin selection”, “reciprocal altruism” and the like are largely arguments about how natural selection works. Those theories were not intended to say anything about altruism as normal people understand the term: (‘psychological’) aggression is at least as likely a candidate as (‘psychological’) altruism for being DNA-reproduction curtailing. If you want to use such theories in that way (although I think you would need a good justification for doing so), you must address various issues.
o   What specific characteristic are you talking about? Describe it as objectively as possible, ideally without importing any “motive-inviting” terminology. “Makes a distinct sound when a predator approaches and in the presence of relatives” is much less likely to lead to confusion than is the phrase “Gives an alarm call” (“gives” and “alarm” both inviting confusion). If you must use something like the latter, define it well first (paying particular attention to the contexts in which the characteristic manifests) and put at least the first use of the phrase in scare quotes.
o   What evidence is there that individual differences in this characteristic are influenced by individual differences in DNA? If there is none, you are making an assumption and you need to provide justification for doing so. A belief that “they must be” because “everything is genetic” is not enough. Unless individual differences in the characteristic you are considering are influenced by individual differences in DNA, you are looking in the wrong place for a good explanatory theory for the former.
o   What evidence is there that individual differences in the characteristic can be DNA-reproduction curtailing over the life-course (seeing here, for example, and allowing for possibilities suggested by reciprocal altruism, kin selection, group selection, etc.)? If you just tell a story about how the characteristic ‘must be’ costly for DNA-reproduction, I bet I can usually come up with an equally plausible story for why it might not be. I will want reasons for thinking your Just-So story is better than mine. (For introductions into Just-So stories in evolutionary theory, you could do a lot worse than looking here or here.)

·         Evolutionarily-speaking, “altruistic punishment” is, if anything, a more misleading term than “biological altruism”. It is necessarily “altruistic” only in the “apparently curtails DNA reproduction” way, just as “evolutionary spite” would - were it to exist. (Relevant research often muddies the water by examining situations in which people pay economic costs apparently to punish violation of norms that, if followed, would benefit others in some way. The mish-mash of different types of costs and benefits – let alone inferred motives - in such paradigms almost guarantees confusion.) Again, such phenomena and possibilities may (but I would guess probably don’t) cause problems for one or other theory of natural selection. They have nothing obvious to say about people’s concern for the welfare of others.

Picture credits

All gene diagrams [Link]
Bullshit poster [Link]
Jean Genie [Link]

How to cite this blog post using APA Style

T. Farsides. (2014, October 26). The genetics of altruism. Retrieved from

Saturday, 18 October 2014

Conceptualisation and measurement of altruism

Introductory waffle

This is very much a work-in-progress blog post, written in a bit of a rush. I hope to edit it ‘sometime’ to make it more clear, readable, and engaging, especially for non psychology students. This week’s topic is the Conceptualisation and Measurement of altruism. My guess is that it will be less interesting to non psychology students than are some of the other posts on this blog: it may be at times a bit too specific and technical. If you choose to read on, though, I hope my guess proves to be wrong.


We are examining the putative phenomenon of people being concerned about the positive welfare of someone or something other than themselves. We are calling that putative phenomenon “altruism”. At this stage, we are not concerned with how such altruism fits with any other conceptualisation of “altruism”.

Expressions as imperfect indicators of altruistic concern

If concern for the positive welfare of others exists, it cannot be examined directly. We can only examine what may be markers or expressions of such concern. We can examine how people (including ourselves) seem to think, feel, and behave, and we can argue about how consistent that is with a possibility that such people are to some extent concerned about someone else’s positive welfare.

We can be wrong about how people (including ourselves) are actually thinking, feeling, and behaving and we can also be wrong that particular thoughts, feelings, and behaviours are, in fact, markers or expressions of concern about the positive welfare of others.


People’s apparent (and likely actual) concern for the positive welfare of others varies in important ways. Among other things, it can be rare or frequent; fleeting or enduring; narrow or inclusive; intense or feeble; dominant or attenuated or silenced by other concerns; and, when enacted, effective, ineffective, or counter-productive.

Ideally there would be a single measure that captured all of these features so that any instance of and change in apparent altruism could be accurately measured. We are far from having such a measure, even if one were possible and practical. Instead, we have measures which attempt to capture apparent altruism that is either highly specific in manner, space, and time (e.g., attempting to give a particular form of help to a particular person on a particular occasion) or is relatively broad in manner, space, and time (i.e., attempting to provide lots of type of help to lots of others on lots of occasions, e.g., dedicating one’s life to becoming the most helpful person one can be).

Measurement matching

Research in attitude psychology (Ajzen & Fishbein, 1980) and personality psychology (Epstein, 1979) has compellingly demonstrated that measures work best if they are used to measure the phenomenon they are intended to measure. (Well, duh!) A measure that accurately predicts whether or not a person will put money in a charity tin when it is waved under her or his nose may not be very accurate at predicting how helpful that person is, on average, across a wide range of helping opportunities. Similarly, a measure that accurately differentiates people according to how helpful they are ‘on average’ may not be particularly good at predicting whether or not one of those people will put money in a charity tin when it is waved under his or her nose on their way to work one day. What people do on specific occasions may be a poor indicator of what they typically do and vice versa.

States and dispositions

As I am using the terms here, “states” are things that exist in a particular moment and “dispositions” are tendencies. As just mentioned, concern for others on a particular occasion may or may not reflect a tendency to be concerned about others on average over time and someone having an altruistic disposition may or may not accurately predict how altruistic they will be on a specific occasion.

Focus and expansiveness

As I am using the terms here, “focus” and “expansiveness” roughly refer to how many others’ positive welfares people care about. Similar to what has just been said, concern for a specific other (e.g., one’s child) may or may not accurately predict concern for others more generally (e.g., all people), just as concern for others in general may or may not accurately predict concern for particular others (e.g., an astonishingly irritating colleague).

Particular or general concern

As I am using the terms here, “particular” concern is expressing concern for another’s welfare in a very specific way, (e.g., by experiencing a particular feeling or adopting a particular body posture or by providing emotional support or by handing over money, etc.) while “general” concern is expressing concern for another’s welfare in range of ways (e.g., with feeling, manner, and behaviour, etc.) or by tailoring the expression of one’s concerns to the specific perceived needs of the other (e.g., by flexibly providing whatever seems to offer the best hope of helping the other).

One-off measurement

To try to keep this post even remotely manageable, I will consider in it only measures that are administered on a single occasion. Nevertheless, the measures reviewed will differ according whether they are claimed or appear to capture state or dispositional altruism, focused or expansive altruism, and particular or general altruism.


The terminology I have just laid out is not commonly used and the distinctions I have just made are often not considered in discussion of particular altruism measures. This may temporarily add to confusion but my hope is that it will eventually lead to improved clarity and understanding (and better measures and interpretation of the results obtained from them!)

Measurement of relatively dispositional, expansive, general altruism

Rushton’s “Self-Report Altruism Scale” (SRAS)

People completing the SRAS (see p. 297 of Rushton et al., 1981) report how frequently they have engaged in each of 20 behaviours likely to have been helpful to individuals they did not know very well. Example items are, “I have helped push a stranger’s car out of the snow”; “I have delayed an elevator and held the door open for a stranger”, and “I have given a stranger a lift in my car”.

Respondents are usually given a score of 1 each time they say “Never”; 2 each time they say “Once”; 3 each time they say “More than once”; 4 each time they say “Often”; and 5 each time they say “Very often”.

People usually therefore obtain a total score of between 20 (if they report never having engaged in any of the behaviours) to 100 (if they report having very often engaged in all 20 behaviours).

The SRAS is rare among altruism measures in asking about people’s ‘helping behaviours’ without asking or speculating about their reasons for engaging in those behaviours.

It is possible, of course, to help strangers for reasons other than caring about their positive welfare. One may help without any strong motive (e.g., from habit or convention) or for only instrumental selfish reasons (e.g., in the hope of being rewarded). Nevertheless, higher scores on the SRAS indicate helping strangers in lots of different ways, with considerable frequency, or both. It does seem likely that this is an indicator of dispositional, relatively expansive, relatively general altruism. Altruistic people are likely to be helpful in various ways according to strangers’ perceived needs and people who are frequently and variously helpful to apparently needy strangers are likely to be relatively dispositionally altruistic.

Answers to the 20 questions tend to correlate positively with each other. This means that the scale has good (enough) internal reliability. (It could be better, though!) This is important because the majority of the questions and the scale as a whole each seem to be measuring more or less ‘the same thing’ (or at least things that tend to highly correlate with each other). Frequencies of helping push strangers’ cars out of snow, holding doors open for strangers, and giving strangers lifts seem not to be independent of each other. Rather, they seem to be measuring something common that each specific instance of helping is representative of: presumably something like “a tendency to be helpful to strangers”.

If the SRAS is given to the same people on different occasions, people’s scores at one time correlate with their scores at other times. People who get relatively high (or low) scores on one occasion tend to get relatively high (or low) scores on the other occasion(s). This means that the scale has good (enough) test-retest reliability. This is important in a dispositional measure. Unless something fairly radical happens to change things, dispositions are relatively stable across time. That is what it means for something to be dispositional. Someone who is relatively low (or high) in dispositional altruism should obtain relatively low (or high) dispositional altruism scores whenever it is measured (using an appropriate disposition measure).

People’s own scores on the SRAS correlate with scores other people obtain when rating them. This means that (in addition to being easily converted for use as an other-report measure) the SRAS has good inter-rater reliability. This is particularly important in self-report measures which can have the weakness of being susceptible to various well-known errors and biases. Having decent inter-rater reliability means that people’s self-reports tend to agree with reports others make of them, which gives some assurance that scores are accurate representations of what is being measured, i.e., relative frequency of helping strangers, on average, across a variety of contexts.

People’s scores on the SRAS correlate with other measures of their helpfulness and with other things that would be expected to correlate with it. For example, people’s SRAS scores correlate (r = .20) with other people’s rating of those people as generally caring, helpful, considerate of others’ feelings, and willing to make a sacrifice (Rushton et al., 1981, p. 296). This means that the scale has some (I repeat, some) claim to being valid. To some extent, it seems to be indicating what it claims to be measuring, i.e., frequent and general helping of strangers because of dispositional concern for the welfare of others, or similar.

In combination, data about the scale’s reliability and validity provide evidence that it is (at least) psychometrically adequate as an indicator of what it claims to be measuring.

In my opinion, the RSAS has two main weaknesses as a measure of dispositional, expansive, general altruism. It is has a serious confound and its frames of reference are unspecified, unclear and, I suspect, inconsistently guessed.  

A measure is confounded when something other than the thing being measured systematically affects what the measure indicates. SRAS scores are cofounded by opportunities to help. (This is one reason SRAS scores often correlate with age.) Individuals who have never seen snow and do not own a car cannot obtain a score as high as potentially equally altruistic car-owning individuals who live in snowy countries. If people across ‘groups’ (e.g., genders, countries, etc.) have different opportunities to help, SRAS scores comparing those groups will show differences that reflect those different opportunities as well as any differences in concern for the positive welfare of others, including their willingness to act on such altruistic concerns. There are likely to be sex differences in giving strangers lifts in cars in Saudi Arabia, not because Saudi men are more altruistic than Saudi women but because Saudi women are not allowed to drive cars.

The SRAS asks how often people have engaged in various behaviours but it does not tell them how to understand the terms “often” or “very often”. There are various ways people might do this. They can choose over what time period they are making their decision and they can choose with whom they compare themselves. These are frame of reference problems. I gave strangers lifts in my car a lot in my 20s but I have probably done so only a handful of times in the last two decades. This is certainly “more than once” but is it “often” or “very often”? If I take a lifetime view and compare myself with “most people in my country”, I guess it is often and maybe very often. If I take a “nowadays” view and compare myself to “drivers of the world” it probably doesn’t even count as often.

In combination, issues such as these make SRAS scores imprecise and difficult to compare (across time, across individuals, and across groups) with confidence.

Nevertheless, the SRAS is one of the most commonly used self-report measures of dispositional, expansive, and general altruism and it is certainly better than many of the alternatives, especially if used on samples whose members have had similar opportunities to engage in the behaviours asked about.

(Johnson et al., 1989 add to the SRAS questions with many more to try to address other limitations they perceive with the SRAS. I believe this approach has had mixed success.)

Aquino and Reed’s Internalised subscale of the Self-importance of Moral Identity Scale (ISMIS)

People completing the ISMIS (see Appendix on p. 1286 of Reed & Aquino, 2003) are asked to think of a real or imaginary person who is “Caring, Compassionate, Fair, Friendly, Generous, Helpful, Hardworking, Honest, and Kind.” They are then asked five questions about being such a person, e.g., “It would make me feel good to be a person who has these characteristics”, “Having these characteristics is not really important to me”, and “I strongly desire to have these characteristics”. Each answer is scored between 1 and 5, where 1 = “strongly disagree” and 5 = “strongly agree”. An average of those scores is then taken (once scores for which low numbers represent high internalisation have been 'reversed').

Thus, scores on the ISMIS seem to most directly indicate differences in people’s strength of self-reported aspiration to be a particular type of person: “moral” for Aquino and Reed but seemingly almost equally interpretable as “altruistic”, i.e., presumably relatively enduringly,  expansively, and generally concerned about the positive welfare of others.

It is possible that some people concerned about the positive welfare of others would not strongly aspire to have all of the characteristics in Aquino and Reed’s list. They might want to be caring, kind, and compassionate, for example, but not especially hard-working or honest. It is also possible that some people might aspire towards only limited altruism; being passionate about the positive welfare of their friends and family, for example, but indifferent to anyone else’s welfare. If such people exist (and I think they do), the ISMIS might not capture the full extent of their altruism.

Other people may not notice or may ignore the “and” towards the end of the attributes listed in the ISMIS. They might give themselves relatively high scores if they have or aspire to any package of traits listed, e.g., if they strive to be fair, friendly, hard-working, and honest even if they do not care too much about being particularly compassionate and helpful.

The two paragraphs immediately above raise the possibility that even accurate and honest self-completion of the ISMIS might lead to under-representation of some people’s altruism and over-representation of others’.

Nevertheless, the ISMIS has impressive psychometric qualities. It has the usual indicators of reliability and validity (including as a measure of relatively enduring, expansive, general altruism). It is also short (just five questions) and only requires people to identify how much they care about being or becoming a particular type of person – a relatively easy task.

The biggest limitation to the ISRIS is arguably that people may have different ideas about what they actually have to do to legitimately consider themselves aspirationally caring, compassionate, etc. Research suggests that nearly everybody thinks they are ‘good’ and indeed better than most other people – even if they could show very little evidence to support such claims (e.g., Sedikides et al., 2014). Whereas honest and accurate completion of the SRAS requires that people regularly act in multiple helpful ways to obtain high scores, the ISRIS may allow people to sincerely claim to be altruistic without them having to do very much by way of actually helping others.

It should be noted that there is some research which suggests that the higher peoples’ ISMIS scores, the more they will tend to engage in certain forms of prosocial behaviour (e.g., donating and volunteering). Nevertheless, more validating studies would provide greater reassurance that this is indeed the case. Until (and indeed probably even) then, it would probably be wise for researchers administering this measure to also administer measures of actual helping behaviour.

In sum, Aquino and Reed’s measure seems to provide a quick and simple indicator of how committed people are to having characters in which enduring, expansive, and general altruism is prominent.

Sprecher and Fehr’s Compassionate Love for Strangers – Humanity (CL4SH)

Sprecher and Fehr (2005) developed a measure of Compassionate Love that can be applied to more or less any target. In this section I will consider the Compassionate Love for Strangers – Humanity scale.

People completing the CL4SH (which I pronounce “Clash”) measure report how true of them they feel 21 statements are, e.g., “If a person (a stranger) is troubled, I usually feel extreme tenderness and caring”, “I tend to feel compassion for people, even though I do not know them”, and “One of the activities that provides me with the most meaning to my life is helping others in the world when they need help.”

Respondents are given a score between 1 and 7 for each answer, where 1 = “not at all true of me” and 7 = “very true of me”.

Thus, scores on the CL4SH measure indicates people’s self-reported emotional, motivational, and behavioural concern for the positive welfare of strangers - humanity.

A few of the questions on the CL4SH scale arguably measure things that are not necessarily associated with altruism. These include perspective-taking (which is one possible route to and aid for altruism, but it is possible to be altruistic without taking others’ perspectives and vice versa) and empathic joy (which often happens alongside altruism but need not; we can feel sad knowing that a loved other’s happiness will be short-lived). Few of the questions are like this, though, and answers to them frequently will be correlated with the other answers (because other-concern, perspective-taking, and emotional contagion often do co-occur).

As with the other measures reviewed so far, the CL4SH scale has at least adequate psychometric properties, i.e., reliability and validity.

In brief, the CL4SH scale provides a simple, psychometrically-adequate, direct measure of people’s enduring altruism towards the expansive group of strangers – humanity.

Perception and learning

People who care about the positive welfare of others ‘generally’ may see and interact with the world in different ways than do less altruistic people. They may be specially attuned to seeing (Den Daas et al., 2013) and learning (Kwak et al., 2014) things likely to be relevant to assessing and enhancing valued others’ welfare. I predict that altruism measurement will soon develop in ways that reverse the order of reporting this relationship: future studies will examine people’s perceptions and learning processes in an attempt to reveal their concerns for the welfare of others. I think this is an exciting prospect.

Poorer measures of relatively dispositional, expansive, general altruism

Caprara et al.’s (2005) measure of ‘Adult Prosocialness’
This is a 16-item self-report measure the questions of which sample a wide-range of arguable ‘prosocial’ behaviours (sharing, lending, helping, volunteering, consoling) and characteristics (perspective taking, emotional contagion, empathic concern, emotional intelligence).
This measure may or may not be an excellent measure of adult prosociality (broadly, individual differences in ‘prosocial personality’) but it has two main weaknesses as a measure of altruism. First, the various ‘bits’ measured (perspective taking, empathic concern, etc.) may well correlate highly with each other for people who are extremely high or low in adult prosociality but need not always do so among most people. It is easily possible for a person to be good at perspective taking but low in empathic concern, for example. Second, the scale as a whole may not correspond too closely with dispositional, expansive, general altruism as I am using that phrase. It may be possible, for example, to be high in concern for others’ positive welfare without manifesting high levels of perspective taking, empathic concern, etc. This might happen if someone is concerned about others’ welfare (as one understands it) for moral reasons or from attempted conformity to religious edicts. A zealot trying to save others’ souls may be genuinely interested in their welfare but be tragically short of empathic concern.

Penner et al. (1995)’s ‘Prosocial Personality Battery’ (PSP)
The PSP combines and distilled a lot of potentially related self-report measures, the items of which formed into two ‘factors’ (groups). One is labelled ‘other-oriented empathy’ and comprises questions about perspective taking, empathic concern, ascriptions of responsibility, and various forms of moral reasoning. The other is labelled ‘helpfulness’ and comprises measures of helping behaviour and the absence of personal distress when witnessing others’ suffering.
My evaluation of this measure is much the same as my evaluation of the Caprara et al. measure just above.

Romer et al.’s (1986) Helping Orientation Questionnaire (HOQ)
The HOQ asks respondents which of 4 provided options they would be most likely to do in each of 23 scenarios. In each case, one alternative is deemed “altruistic”, usually more or less in the way that I am using the term. The other options are deemed to reveal different motives, e.g., helping only as a means of obtaining purely selfish benefits, e.g., being rewarded. In one scenario, for example, people will be scored as altruistic if they say that they would lend a friend some clothes if the friend really wanted to borrow them - rather than refusing to lend them the clothes or doing so only on condition that they could borrow something in return. (Most people choose a predominance of altruistic answers.) Respondents are given an overall altruism score depending on how many more allegedly altruism-revealing options they select compared to other people being measured. The psychometric qualities of the measure are lamentable and it is mentioned here only because it has been influential in the development of (somewhat) better instruments, e.g., Penner et al.’s (1995) Prosocial Personality Battery.

Van Lange et al.’s (1997) Social Value Orientation (SVO)
The measure of SVO in this paper asks people to make 9 choices between 3 options, A, B, or C. Each choice involves allocating some points (that people were told to imagine were valuable) to the self and some points to an anonymous other individual. Options A, B, and C differ in systematic ways so that, for example, A results in self and other getting identical points (e.g., 520 each), B results in the self getting the same amount of points as in Option A but the other getting far fewer (e.g., 520 to self and 120 to other), and C results in the self getting more points than in Option A and the other getting fewer points than in Option A but more points than in Option B (e.g., 580 to self and 320 to other). Option A is considered to be a “prosocial” choice because it maximizes the points given to the other at some cost to the self (relative to choosing Option C). People are said to have a “prosocial value orientation” if they select the “prosocial” option in at least 6 of their 9 allocation decisions. Measures like this are very popular. I have more issues with them than makes sense to go into now. The main one is that I see no tension between someone being extraordinarily altruistic ‘in real life’ but not making the most prosocial choices in this particular setting, i.e., when allocating points between the self and a random stranger in a game-like context. Slightly more formally, a single measure of how people act on a single highly particular occasion is likely to be an unreliable, invalid indicator of how they act over a wide range of situations over an extended period. In other words, I doubt Van Lange et al’s claim that such measures are strong indicators of people’s “general tendencies toward others” (p. 736). They seem to me to be measures of something much more specific (and not necessarily altruistic). See my comments below on “Economic Game Behaviour” for some related issues. See Murphy and Ackerman (2013) for a more thorough consideration of SVO measurement.

Measurement of relatively dispositional, expansive, particular altruism

Davis’ Empathic Concern (EC)

People completing this measure report how much they agree or disagree with each of seven statements concerning their regular, mainly emotional, reactions to others’ perceived welfare, e.g., “I often have tender, concerned feelings for people less fortunate than me”, “I would describe myself as a pretty soft-hearted person”, and “When I see someone being taken advantage of, I feel sort of protective towards them”.

For each answer, respondents obtain a score between 1 and 5, where 1 = “does not describe me well” and 5 = “describes me very well”. An average of those scores is then taken.

Thus, scores on the EC measure indicate people’s enduring tendencies to experience sympathetic or compassionate feelings when perceiving others in need.

It is possible to care about others’ welfare without experiencing sympathy or compassion. Moral beliefs can motivate altruism without any accompanying emotions and emotions other than sympathy can also evoke altruism, e.g., anger. Nevertheless, feeling sympathy, etc., is one indicator or form of positive other-concern.

As I said in relation to the CL4SH, there is validating data showing that EC scores correlate with various forms of actual helping behaviour, probably because empathic concern is one cause of attempted helping. The EC scale has at least adequate psychometric properties in other respects, too.

Davis’ Empathic Concern therefore provides a brief and simple measure of people’s enduring tendency to experience compassion when considering others’ apparent misfortune.

Measurement of relatively dispositional, focused, general altruism

Sprecher and Fehr’s Compassionate Love for Close Others (CL4CO)

Sprecher and Fehr (2005) developed three versions of their Compassionate Love scale. One of these was the Compassionate Love for Close Others (“Claco”) scale, also known as the Compassionate Love for Friends and Family (CL4FF) scale. This is essentially exactly the same as the CL4SH scale (see above) but replaces references to “strangers”, “humankind”, and similar with ones to “friends”, “family”, “loved ones”, and the like, e.g., “If a family member or close other is troubled, I usually feel extreme tenderness and caring”, “I tend to feel compassion for people who are close to me”, and “One of the activities that provides me with the most meaning to my life is helping others with whom I have a close relationship.”

Everything else I might say about the CL4CO measure is more or less what I would say about the CL4SH one. Thus, the CL4CO scale provides a simple, psychometrically-adequate, direct measure of people’s enduring emotional, motivational, and behavioural altruism towards the relatively specific ‘group’ of close others.

Sprecher and Fehr’s Compassionate Love for Specific Others (CL4SO)

Sprecher and Fehr (2005) third scale measures Compassionate Love for Specific Others (“Claso”). Again, this is essentially exactly the same as both the CL4SH and the CL4CO scales but rather than making explicit reference to strangers and humanity or to close others, questions have blanks in them for participants to fill-in with the name of someone or something they love, e.g., “If ______ is troubled, I usually feel extreme tenderness and caring”, “I tend to feel compassion for ______”, and “One of the activities that provides me with the most meaning to my life is helping _______.”

Most of what I might say about the CL4SO measure is more or less what I would say about the CL4SH or CL4CO ones, with two exceptions. First, the CL4SO actually subsumes the other two scales; it is just that the researchers have ‘pre-filled’ the spaces with either ‘strangers and humanity’ or ‘close other’ references. Secondly, the CL4SO allows measurement of altruism towards almost anything that can be loved and perceived as having a welfare, including non-human targets such as, for example, “my pet” or “the environment” This gives the scale an impressive potential range of applicability. (This range is limited somewhat by some of the items, largely the confounding ones, which also require that the recipient of altruism has to be able to experience things so that, for example, people can take its perspective. This prevents the unadulterated scale being used as a measure of altruism towards things without such capacities, such as one’s nation or the planet.)

Thus, the CL4SO scale provides a simple, psychometrically-adequate, direct measure of people’s enduring emotional, motivational, and behavioural altruism towards almost anything that participants might think of as having welfare.

Measurement of relatively state, focused, particular altruism

Responsive (apparent) helping

A number of researchers (e.g., Levine, 2003) have used observations of actual or attempted helping as an indication of altruism. These measures are often taken in ‘real world’ situations where people are confronted with the possibility of helping when someone else appears to have done something like:

  • Lost or forgotten to post a letter
  • Dropped, spilled, or be struggling to carry one or more objects
  • Had a fall or other accident
  • Experienced a seizure
  • Some other thing which means they would likely benefit from assistance

Helping responses are often measured in one or both of two ways: whether or not help is given and either the extent (e.g., how many spilled pens picked up) or the speed (e.g., delay before picking up the first pen) of helping.

Such measures have a number of considerable strengths. They are realistic, observable, and straightforward. They have two main potential weaknesses as measures of altruism.

First, any behaviour can be engaged in for reasons other than wanting to improve another’s welfare. (I can pick up an addressed envelope out of curiosity and post it because my wife commands me to.) Second, absence of a particular behaviour is not compelling evidence for a lack of concern about others’ positive welfare. (I may have simply not noticed a need for or an opportunity to help.) For such reasons, no single act is an infallible indicator of altruism and nor is the absence of that act.

Potential helping behaviours differ in the extent they have such weaknesses. Some behaviours can clearly be engaged in for reasons other than particularly wanting to improve another’s welfare, e.g., putting money on a church collection plate when everyone else has and when people are clearly watching. Some helping opportunities are easy to miss or to clearly identify as such, e.g., a fellow rush-hour commuter leaving on her seat a newspaper she planned to read later. Nevertheless, some helping opportunities can be difficult to miss or to misconstrue and aspects of the situation can be such that people who help can often be fairly safely assumed to do so in large part because of concern for another’s positive welfare, e.g., comforting or protecting someone who is obviously in need when doing so is personally costly. (Just to be clear – cost to the helper makes it easier to identify their altruism; it is not a required element of altruism.)

To the extent that performing a behaviour seems clearly motived by concern for the positive welfare of another and that not performing the behaviour is difficult to consider in any way other than as revealing a lack of such concern, responsive helping can be an excellent measure of state, focused, particular altruism. Further, to the extent that the revealed altruism (or lack thereof) can be confidently predicted to generalise to other times, situations, and specific ways of helping, the behaviour may also serve as an indicator of more dispositional, expansive, or general helping. The onus of course is on anyone who wishes to claim that a specific behaviour reveals altruism to provide evidence that conditions such as these have been met, especially if there seems good reason to doubt this.

Economic Game behaviour

Economic games are an important sub-set of responsive (apparent) helping. It is often claimed that some of these games reveal altruism, perhaps especially the Dictator Game (DG).

In the DG a person receives a sum of money or a similar ‘good’ or resource. They are then asked if they want to give any of it to someone else. How much they give is commonly used as a measure of altruism. My view is that this is justified (only) to the extent that the conditions specified at the end of the previous section are satisfied. I think that these conditions are usually very poorly satisfied.

The DG is often used in situations in which a person is invited to divide a sum of money, let’s say £10, between themselves and an anonymous stranger, usually having received explicit assurances that they will never meet the other person and that the other person will not know who their dictator was. While it is of course possible that some people are motivated by concern for the positive welfare of the other in such circumstances, it seems likely that such concern will be rare, slight, and play an extremely limited role in most people’s decision making and subsequent action. Why would or should anyone in the standard DG scenario care too much about the positive welfare of an anonymous other who will be better off if they receive any money, especially when the ‘assurances’ received almost certainly provide strong hints about ‘how to play the game’, e.g., “Keep as much money as your conscience and standing in the eyes of the experimenter will allow” or “Clearly you should give the other person a decent wodge – unless you want to reveal yourself as an immoral, selfish toad” (depending on how people interpret the rules of the game on any given occasion; Liberman et al., 2004).

More formally, there are lots of possible motives for people to give money to another in the DG other than altruistic concern and players keeping money is probably a better indicator of their concern for personal monetary gain in a zero-sum situation than it is an absence of concern for another’s welfare (although there probably isn’t much of the latter, either, but the main problem is that keeping money in the DG confounds concern for another with concern for self).

To be clear, it seems likely that giving in economic games can be motivated by and indicate concern for the positive welfare of another (Batson et al., 1995; Batson & Moran, 1999; Bettinger & Slonim, 2006). It just probably usually isn’t and doesn’t (Bardsley, 2008; Krupka & Weber, 2013). Similarly, behavior in the highly artificial and managed situation of economic games can sometimes reflect more general inclinations and behaviours (Peysakhovich et al., 2014). It just probably usually doesn’t (Laury & Taylor, 2008; Levitt & List, 2007).

Heal’s Alien Thought Experiment

Jane Heal (1991, p. 164-165) created a captivating thought experiment which essentially serves as a fairly crude measure of altruism. Here it is:

Imagine that you are kidnapped by some aliens. They present you with a console on which are two buttons, A and B. Your choice is to press one of the buttons. The aliens explain to you that they have your loved ones in their power. Whatever you do, you will never see them again, but their future fate rests in your hands. If you press button A they will be well looked after and live long and happy lives; you, on the other hand, will be rendered unconscious as soon as your finger touches the button and when you come round you will have had a set of delusive memories planted in your brain, according to which you betrayed them to suffering; you will be plagued with guilt and misery. If you press button B, however, your loved ones will be tortured, starved and degraded while you live happily, equipped with the delusive conviction that your nobility has preserved them. The aliens persuade you that they can do what they say. You see with your own eyes the effects of their brain manipulations on others and you become convinced of the extent of their power. The moment of choice is now; what will you do?

Choosing Button A indicates altruism towards your loved ones. In at least one fantastical situation, you claim willingness to bear significant personal cost to ensure specific others’ (i.e., loved ones’) positive welfare. You certainly claim to prefer this to the alternative option which involves condemning your loved ones to a miserable fate to ensure your own future welfare.

There are several limitations with using Heal’s thought experiment as a measure of altruism. One is its hypothetical nature and the problems of self-knowledge and self-report. What people say they would do sometimes bears little relation to what they actually would do. Without any validating studies, one has to be sceptical. Another is the required contrast between the two Options. It not clear whether someone pushing Button A is doing so because they like the claimed consequences of doing so, because they hate the claimed consequences of pushing Button B, or both. Are they actively pursuing positive welfare for those they love or ‘merely’ seeking to avoid a living hell for them? A third and parallel limitation is that each Button leads to two consequences, one for the self and one for loved others. However unlikely it may be, pushing Button A may also be motivated by not wishing to have false memories of personal nobility. A fourth is the all-or-nothing nature of the measurement: one must either be 100% altruistic or not. I could go on but it seems unfair to criticise something that was not intended to be a measure for not being a perfect measure!

I include Heal’s thought experiment here because it is a lovely illustration of a principle that could underpin a good altruism measure. While, as I have already mentioned, sacrifice is not an essential component of altruism, the extent of one’s willingness to make a sacrifice on another behalf is an indication of one’s concern for their positive welfare. The more altruistic one is towards someone or something, the more costs one will be willing to incur to improve their welfare by a set amount. Similarly, the more altruistic one is, the bigger the cost one will be willing to incur to improve another’s welfare by a set amount.

Anyone interested in developing a measure along these lines might also like to consider the following papers: Bélanger et al. (2014), Gómez et al. (2011), Jung et al. (2014), and Read and Loewenstein (1999).

Video Game Behaviour

Some video games give players an opportunity to help others. One example is provided by Lieberg et al.’s (2011) Zurich Prosocial Game (ZPG). In the ZPG there are occasions when players’ avatars will die unless a task is completed. At the same time, players have opportunities to help the avatars of other players not in competition with them. Similarly, in Zanon et al. (2014) players in virtual reality scenarios have the opportunity to help other players’ avatars when all the avatars are portrayed as being in mortal danger.

These specific examples of video game helping have similar weaknesses to many economic game measures of altruism. Helping may indicate things other than altruism; not helping may indicate things other than altruism being absent; altruism within the game may not reliably and validly indicate altruism outside the game; and so on. Moreover, to the extent that altruism does occur, it is not clear towards whom that altruism is directed. When players do altruistically help, are they concerned with the positive welfare of other players, of other players’ avatars, or both?

I nevertheless wanted to include video game altruism here, for two reasons. First, the most important aspect of altruism in video games is usually missed by those who discuss it. Players are deeply altruistic towards their own avatars, i.e., towards the characters who represent them and/or with whom they are identifying. Players often invest considerable resources (e.g., time, money, opportunity costs) in promoting and protecting their avatars’ perceived (because portrayed) welfare and react in multiple expressive ways according to avatars’ (portrayed) actual and possible fate, e.g., tensing up when the avatars are (portrayed as being) in peril; basking in avatars’ (portrayed) achievements; etc.

The second reason I wanted to include video game altruism measures here is because they have great promise. They are currently in their infancy but are likely to improve rapidly. They do not need to develop very much before they offer measures that are popular with participants (people like playing good games!), valid (well-designed games will able to differentiate concern for others’ positive welfare from other concerns, e.g., for the ‘self’, for morality, etc.); reliable (players’ behaviours are likely to be relatively consistent in all the ways that are important to demonstrate measurement reliability); and generalisable (people are likely to  play well-designed games in ways that predict how they think, feel, and behave in the world outside those games).

Batson’s State ‘Empathy’

People completing this measure report how strongly they are or were feeling each of the following states at a particular moment: sympathetic, moved, compassionate, tender, warm, and soft-hearted.

People obtain a score of 1 each time they say ‘Not at all’ and a score of 7 each time they say ‘Extremely’. An average of those scores is then taken.

Batson’s measure captures how sympathetic or compassionate people are feeling at a particular moment, usually in response to witnessing someone in distress. He describes this feeling (which he calls “empathy”) as an important precursor to altruism but for current purposes it is an emotional expression of altruism: there is plenty of evidence that such feelings are reliable and valid indicators of concern for the other’s positive welfare.

Batson’s State ‘Empathy’ therefore provides a brief, simple, and effective measure of a particular form or expression of altruism (i.e., an emotional one) evoked in particular situations (i.e., usually when perceiving others to be in personally-unwelcome unenviable circumstances).  

Facial and Bodily Expressions of Other-Concern

Nancy Eisenberg, Richard Fabes and their colleagues pioneered techniques to identify non-verbal expressions of other-concern beyond actual helping behaviour. They took pains to be able to differentiate such expressions (which reveal concern for the other’s welfare) from ones indicating mere personal distress when people confront situations they find upsetting (without being concerned about the others’ welfare, per se). So, for example, personal distress was thought to be revealed when a person’s eyebrows were “somewhat raised and pulled together” or the person “exhibited non-functional, nervous mouth and chin movements such as tightening or biting the lips”. Other-concern, on the other hand, was distinguished by a person having their “eyebrows pulled down flat and forward toward the bridge of the nose, furrowing in the centre of the brow (for some children), eyelids not pulled in tight or raised, head and body oriented forward, and the bottom eyelids raised slightly” (p. 138).

Eisenberg and Fabes (1990) provide an excellent review of this and related research. In it, they caution that experimental situations sometimes evoke self-presentation concerns, people can fake experience of particular emotions/concerns, and people differ in how emotionally expressive they are (note the “in some children” above). These researchers’ ambitions were also arguably hampered to a certain extent by the dual not-entirely-compatible goals of (a) wanting to identify markers of other-concern, and (b) wanting to do so in a way that maximally distinguished between other-concern and (solely) personal distress. This is potentially a limitation because people who are other-concerned are often likely to experience personal distress as well when witnessing or contemplating others’ dire situations or experiences.

A similar limitation dogs McEwan et al. (2014)’s claim to have identified what a compassionate expression looks like to such an extent that they been able to produce a set of photographed faces depicting such expressions. Their methodology is such that, at best, they can claim to have photographs which depict expressions that are more compassionate than are photographs depicting neutral or critical expressions. Participants in their study had to score sets of three photographed faces according to how each photo within each set expressed compassion/warmth, criticism, or neutrality. In each set, one face was clearly critical and another was wholly inexpressive. The photographs allegedly expressing compassion were rated as more compassionate than were the critical or neutral ones. It is a long leap from this to say that the former photographs actually conveyed compassion! That claim would require that people looking at the photographs would recognise compassion without any prompts or clues about what they were ‘supposed’ to be seeing. That is a much harder task and I suspect McEwen et al.’s faces would not pass it (see Widen et al., 2011)

It is also worth noting that many people may be skilled at recognising other-concern in people’s faces, posture, and gestures, even (or maybe even especially) if they are not given detailed recipes for how to recognise it.

Physiological Measures

Various physiological markers of altruism or related constructs have been proposed. These include heart rate (Eisenberg and Fabes, 1990), oxytocin (Barraza & Zak, 2009; Israel et al., 2012; Zak et al., 2007), vagal activity (Kogan et al., 2014), and changes in blood flow in particular brain regions (Zanon et al., 2014). I will consider each of these in depth in later posts. For now I shall simply express scepticism that a very close and discriminating relationship will be found between any of these things and altruism, particularly if the latter is poorly specified (Bartz, et al. 2011; DeWall et al., 2014; Jarrett, 2014; Love, 2014).


It is difficult to avoid a conclusion that there is and probably never will be a perfect measure of altruism. It would be wrong to conclude in turn that we should give up trying to measure it and that every measure of altruism is completely unfit for purpose. Rather, we should recognise that each measure has its strengths and weaknesses and select from among them accordingly. We should also recognise that when we really want to know if someone is altruistic, we are likely to have to seek coherence among a number of different indicators.

I often invite my students to think about a time when they wondered if someone loved them. I suggest that what they did to try to answer this question was to think about all the possible evidence and seek to understand what the pattern of these indicators suggested. Doing this again yesterday made me think of fairy tales and Disney films. Princes wanting to marry princesses are often set quests as proof of their dedication (and worth). But, arduous and discriminating though such tests can be, they are not infallible. Many a plot rests on the possibility that a suitor seeks not a loving relationship with the princesses but marriage as a purely instrumental means of obtaining land, power, and other personal profits. Rare is the princess (or the father thereof) who is unable to ever reach a decision because of this possibility. Instead, they gather data across a range of times and situations, particularly those that might provoke suitors to reveal any caddish motives. Eventually, the only possible explanation for some noble man’s actions is their true love for the princess.

It ’aint all Disney (see plot)

True life can be trickier than cartoon life but the process is similar even if the outcome is not quite so certain. If someone genuinely cares about the positive welfare of another, such care should manifest itself in various or tailored ways. Although any given indicator may be individually unreliable, collectively they can often prove matters beyond reasonable doubt. Zahn-Waxler et al. (1992) provide a good example of such a multi-method approach, looking as they did at potential altruists’ hypothesis testing (trying to find out if need existed and how help might be provided), non-verbal indicators of empathic concern, and helping behaviours.


Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behaviour. Englewood Cliffs, NJ: Prentice-Hall.
Aquino, K., & Reed, A. II. (2002). The self-importance of moral identity. Journal of Personality and Social Psychology, 83, 1423-1440.
Bardsley, N. (2008). Dictator game giving: altruism or artefact? Experimental Economics, 11, 122-133.
Barraza, J. A. & Zak, P. J. (2009). Empathy toward strangers triggers oxytocin release and subsequent generosity. Annals of the New York Academy of Sciences, 1167, 182-189.
Bartz, J. A., Zaki, J., Bolger, N., & Ochsner, K. N. (2011). Social effects of oxytocin in humans: context and person matter. Trends in Cognitive Sciences,15(7), 301-309.
Batson, C. D. (1987). Prosocial motivation: is it ever truly altruistic? In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 20, pp. 65-122). New York: Academic Press.
Batson, C. D., Batson, J. G., Todd, M., Brummett, B. H., Shaw, L. L., & Aldeguer, C. M. R. (1995). Empathy and the collective good: Caring for one of the others in a social dilemma. Journal of Personality and Social Psychology, 68, 619-631
Batson, C. D., & Moran, T. (1999). Empathy-induced altruism in a prisoner's dilemma. European Journal of Social Psychology, 29, 909-924. 
Bélanger, J. J., Caouette, J., Sharvit, K., & Dugas, M. (2014). The psychology of martyrdom: Making the ultimate sacrifice in the name of a cause. Journal of Personality and Social Psychology107(3), 494.
Bettinger, E., & Slonim, R. (2006). Using experimental economics to measure the effects of a natural educational experiment on altruism. Journal of Public Economics90(8), 1625-1648.
Caprara, G. V., Steca, P., Zelli, A., & Capanna, C. (2005). A new scale for measuring adult prosocialness. European Journal of Personality Assessment, 21, 77-89.
Costa, P. T., & McCrae, R. R. (1991). Facet scales for agreeableness and conscientiousness: A revision of the NEO personality inventory. Personality and Individual Differences 12(9): 888. doi:10.1016/0191-8869(91)90177-D
Davis, M. H. (1983). Measuring individual differences in empathy: evidence for a multidimensional approach. Journal of Personality and Social Psychology, 44, 113-126.
Den Daas, C., Häfner, M., De Wit, J. (2013). Sizing opportunity: Biases in estimates of goal-relevant objects depend on goal congruence. Social Psychology and Personality Science, 4(3), 362-368.
DeWall, C., Gillath, O., Pressman, S., Black, L., Bartz, J., Moskovitz, J., & Stetler, D. (2014). When the love hormone leads to violence: Oxytocin increases intimate partner violence inclinations among high trait aggressive people Social Psychological and Personality Science, 5 (6), 691-697
Eisenberg, N., & Fabes, R. A. (1990). Empathy: Conceptualization, measurement, and relation to prosocial behavior. Motivation and Emotion, 14, 131-149.
Epstein, S. (1979). The stability of behavior: I. On predicting most of the people much of the time. Journal of Personality and Social Psychology37(7), 1097.
Gómez, A., Brooks, M. L., Buhrmester, M. D., Vázquez, A., Jetten, J., & Swann Jr, W. B. (2011). On the nature of identity fusion: Insights into the construct and a new measure. Journal of Personality and Social Psychology100(5), 918.
Heal, J. (1991). Altruism. In R. A. Hinde & J. Groebel (Eds.), Cooperation and prosocial behaviour (pp. 159-182). Cambridge: Cambridge University Press.
Israel, S., Weisel, O., Ebstein, R. P., & Bornstein, G. (2012). Oxytocin, but not vasopressin, increases both parochial and universal altruism. Psychoneuroendocrinology, 37 (8), 1341-1344.  
Jarrett, C. (2014, August 5). What’s different about the brains of heroes?
Johnson, R., Danko, G. P., Darvill, T. J., Bochner, S., Bowers, J. K., Huang, Y-H., Park, J. Y., Rahim, A. R. A., & Pennington, D. (1989). Cross-cultural assessment of altruism and its correlates. Personality and Individual Differences, 10 (8), 855-868
Jung, M. H., Nelson, L. D., Gneezy, A., & Gneezy, U. (2014). Paying more when paying for others. Journal of Personality and Social Psychology, 107(3), 414-431.
Kogan, A., Oveis, C., Carr, E. W., Gruber, J., Mauss, I. B., Shallcross, A., ... & Keltner, D. (2014). Vagal activity is quadratically related to prosocial traits, prosocial emotions, and observer perceptions of prosociality. Journal of Personality and Social Psychology,
Krupka, E. L., & Weber, R. A. (2013). Identifying social norms using coordination games: Why does dictator game sharing vary? Journal of the European Economic Association11(3), 495-524.
Kwak, Y., Pearson, J., & Huettel, S. A. (2014). Differential reward learning for self and others predicts self-reported altruism. PloS one9(9), e107621.
Laury, S. K., & Taylor, L. O. (2008). Altruism spillovers: Are behaviors in context-free experiments predictive of altruism toward a naturally occurring public good? Journal of Economic Behavior and Organization, 65, 9-29.
Levine, R. V. (2003). The kindness of strangers. American Scientist, 91, 226-233
Levitt, S.D., & List, J. A. (2007). What do laboratory experiments measuring social preferences reveal about the real world? The Journal of Economic Perspectives, 21 (2), 153-174.
Liberman, V., Samuels, S., & Ross, L. (2004). The name of the game: Predictive power of reputation versus situational labels in determining prisoner's dilemma game moves. Personality and Social Psychology Bulletin, 30 (9), 1175-1185.
Lieberg, S., Klimecki, O., & Singer, T. (2011). Short-term compassion training increases prosocial behavior in a newly developed prosocial game. PLoS One, 6 (3), e17798.
Love, T. M. (2014). Oxytocin, motivation and the role of dopamine. Pharmacology Biochemistry and Behavior119, 49-60.
McEwan, K., Gilbert, P., Dandeneau, S., Lipka, S., Maratos, F., Paterson, K. B., & Baldwin, M. (2014). Facial expressions depicting compassionate and critical emotions: The development and validation of a new emotional face stimulus set. PloS one9(2), e88783.
Murphy, R. O., & Ackermann, K. A. (2013). Social Value Orientation: Theoretical and measurement issues in the study of social preferences. Personality and Social Psychology Review, 18(1), 13-41.
Penner, L. A., Frizche, B. A., Craiger, J. P., & Freifeld, T. S. (1995). Measuring the prosocial personality. In J. N. Butcher, & C. D. Speilberger (Eds.), Advances in Personality Assessment: Vol. 12 (pp. 147-163). Hillsdale, NJ: Erlbaum.
Peysakhovich, A., Nowak, M. A., & Rand, D. G. (2014). Humans display a ‘cooperative phenotype’ that is domain general and temporally stable. Nature Communications5(4939).
Read, D., & Loewenstein, G. (1999). Enduring pain for money: Decisions based on the perceptions of memory of pain. Journal of Behavioral Decision Making, 12, 1-17.
Romer, D., Gruder, C. L., & Lizzadro, T. (1986). A person-situation approach to altruistic behavior. Journal of Personality and Social Psychology, 51, 1001-1012. 
Rushton, J. P., Chrisjohn, R. D., &Fekken, G. C. (1981). The altruistic personality and the self-report altruism scale. Personality and Individual Differences, 2, 293-302.
Sedikides, C. Meek, R., Alicke, M. D., & Taylor, S. (2013). Behind bars but above the bar: Prisoners consider themselves more prosocial than non-prisoners. British Journal of Social Psychology,
Sprecher, S., & Fehr, B. (2005). Compassionate love for close others and humanity. Journal of Social and Personal Relationships, 22, 629-651.
Van Lange, P. A. M., De Bruin, E. M. N., Otten, W., Joireman, J. A. (1997). Development of prosocial, individualistic, and competitive orientations: theory and preliminary evidence. Journal of Personality and Social Psychology, 73, 733-746.  
Widen, S. C., Christy, A. M., Hewett, K., & Russell, J. A. (2011). Do proposed facial expressions of contempt, shame, embarrassment, and compassion communicate the predicted emotion? Cognition & Emotion25(5), 898-906.
Zahn-Waxler, C., Robinson, J. L., &Emde, R. N. (1992). The development of empathy in twins. Developmental Psychology, 28, 1038-1047.
Zak, P. J., Stanton, A. A., & Ahmadi, S. (2007). Oxytocin increases generosity in humans. PLoS ONE, 2, e1128.
Zanon, M., Novembre, G., Zangrando, N., Chittaro, L., &Silani, G. (2014). Brain activity and prosocial behavior in a simulated life-threatening situation. NeuroImage, 98, 134-146.

Picture credits

Cartoon [Link]
Hug [Link]
Nun [Link]

How to cite this blog post using APA Style

T. Farsides. (2014, October 18). Conception and measurement of altruism. Retrieved from