Narcissism and selfishness are the human default
Narcissism is difficult to define, as definitions have some inconsistencies across sources:
Wikipedia defines it as a “self-centered personality style characterized as having an excessive preoccupation with oneself and one's own needs, often at the expense of others.”
Oxford languages defines it as “excessive interest in or admiration of oneself and one's physical appearance.”
The encyclopedia Britannica claims that narcissism is “pathological self-absorption” and that “Narcissism is characterized by an inflated self-image and addiction to fantasy, by an unusual coolness and composure shaken only when the narcissistic confidence is threatened, and by the tendency to take others for granted or to exploit them.”
The DSM V defines narcissism personality disorder as “a pattern of grandiosity, need for admiration, and lack of empathy”.
The Wikipedia and Britannica definitions include a tendency to exploit others but not the others, and the DSM includes a lack of empathy in its definition, though the other three do not.
Like with intelligence, there is a tendency to define narcissism by what it correlates with instead of what it really is. Low empathy is best characterized as a facet of psychopathy, and selfishness/exploitativeness is a facet of machiavellianism. From the four definitions I posted and others I have looked at, seeing the self as more valuable and important than it really is seems to be the most consistent definition of narcissism.
II.
Research on how individuals rate themselves relative to other people has resulted in the discovery of the better than average effect: people tend to consider themselves superior to the average person by about .78 standard deviations, which corresponds to roughly the 78th percentile. This doesn’t mean that 78% of people engage in self-enhancing behaviour, as the average and the distribution are not the same thing. This could be skewed by somebody considering themselves 3 standard deviations above or below the mean - what is most relevant is how many people consider themselves to be above average, which seems to be about 55-95% depending on the trait.
80% of psychology students at Victoria University consider themselves to be a better than average driver.
94% of teachers at the University of Nebraska consider themselves to be above average teachers.
77% of Swedish drivers and 88% of American drivers consider themselves to be more safe than the average driver.
In the add health dataset, about 6% of teenagers rate themselves as below average in intelligence, while about 54% rated themselves as above average in intelligence.
It is difficult to measure exactly how narcissistic the average person is based on just this data - it’s possible that somebody may consider themselves a below average driver but an above average thinker or prostitute. Some people may not consider these traits to be valuable, so they don’t feel the need to overreport them. Because of this, I believe that the percentage of people who possess a functionally narcissistic personality is even higher than the data suggests it is on paper. The individual differences still matter, of course, somebody who is predisposed to self-enhance by 0.5 SD in desirable traits will have an incredibly different personality to somebody who is predisposed to self-enhance by 2.5 SD.
I hypothesize that the reason why narcissism evolved is because of mate selection. If people were to give themselves realistic self-assessments of mate value, they would go after people who are in their league. Instead, people tend to go after people who are moderately more desirable than themselves. While this strategy reduces the chances of being selected by a mate, this strategy only needs to work once to succeed.
Self-reported attractiveness and other-rated attractiveness correlate weakly - the observed correlation in a sample of 528 teenagers was only .20 to .30. This is lower than the correlation between self and peer reported personality of .47, and the agreement between self-reported verbal ability and measured verbal ability (r = .40). If it is the case that narcissism evolved to warp people’s perceptions of their mate value, it would be expected that self-reports of traits strongly associated with mate value such as attractiveness would be reported the most inaccurately.
The low agreement between self and other rated attractiveness is not due to attenuation, as both the ratings of self-reported attractiveness (rxx = .85) and other-rated attractiveness (rxx = .9) were fairly reliable.
Subjective Attractiveness
Adolescents were asked to “rate how attractive (good looking) you think each of the following parts of your appearance are: your face, your hair, your body, and your overall appearance” on a Likert scale from 1 (very unattractive) to 5 (very attractive). Responses across the four items were averaged to create an overall subjective attractiveness score ranging from 1 to 5. The reliability of the measure was acceptable across the four time points (T1 α = 0.87, T2 α = 0.87, T3 α = 0.86, T4 α = 0.84).
…
As with the subjective measure above, the raters were asked to score the attractiveness of participant’s face, hair, body, and overall appearance on a scale from 1 (very unattractive) to 5 (very attractive). Mean scores were calculated for each photograph by averaging scores across the four items. The attractiveness ratings of the neutral and animated photographs were found to be highly correlated (T1 = 0.90, T2 = 0.91, T3 = 0.93, and T4 = 0.90) so scores were averaged across the two photographs to create a single objective attractiveness mean score at each measurement occasion.
II.
Seflishness can also be measured using dictator games - Joseph Bronski has already done a review on this literature which I will summarize shortly. The dictator game involves one person being assigned the dictator and them choosing how money gets allocated to both of them. This is the distribution of allocation (0 - taking all the money, 1 - giving it away), according to a meta-analysis:
The trait distribution also depends on demographics.
While it’s likely that individuals who chose to take more money than give are functionally selfish, it’s unclear whether those who chose to split or donate are. They may be instinctively attempting a cooperative strategy for selfish purposes, or they may not value money enough to choose a selfish strategy. Some people might do selfless things because it makes them feel good.
The extended moderator analysis does support that alternative strategies may be the cause behind some of the observed “selflessness”, as the conditions of the studies affected people’s decisions.
Theoretically, if natural selection and sexual selection occurs, then humans that optimize for their self interest will win out. It would be incorrect to say that
Some, including myself, initially found the theory that most humans are fundamentally selfish to be implausible and distasteful, especially if they view themselves and their peers as altruistic. These perceptions are ultimately incorrect, and are likely caused by self-deception, which evolved to allow humans to socially operate more effectively. This is Robin Hanson’s homo hypocritus paradigm - it’s not really something that can be argued for using just an essay, study, or a few anecdotes. It’s something that people must realize for themselves with the bulk of evidence that life may or may not throw at them.
Consider this - if humans were allowed to be consciously aware of their antisocial behaviour, then it would be more difficult for them to convince others that they are prosocial and trying to help the group.
There is the question of group selection - some have argued that altruism could develop because altruistic groups outcompete selfish ones by having more efficient group dynamics. A similar case could be made for selfish groups being selected over altruistic ones by being more brutal in warfare. Without hard data, which will never be collected, it is impossible to know which ad hoc evolutionary theory is true.
Between races, it was the Northwestern Europeans (average agreeableness) who dominated the world; not the Asians (high agreeableness), Southern Europeans (mid-low agreeableness), or the Slavs (low agreeableness) for that matter. This does not fit the group selection for altruism model - personally I suspect the distribution of agreeableness within humans is not influenced by group selection.
I would question your conflation of agreeableness with altruism. Tests of agreeableness conflate different traits, I always thought Europeans were just as high in altruism as East asians but just less conformist giving us lower agreeableness scores. But I've never seen comparisons of them doing the dictator game so I could be wrong I guess. I wouldn't say that Europeans dominating the world over Asians is strong evidence against group selection for altruism, it's probably just other traits that Asians have that stopped them from starting an industrial revolution like low Openness. Both seem high in altruism compared to the primitive low trust societies that failed to develop.
With regard to group selection, I think it needs to be disambiguated, so that we know which form you're referring to - because there's more than one sense of the term "group selection". A plausible form of group selection is "cultural group-selection": that refers to culture evolving in response to some form of competition between groups. In which case - and here I'm being speculative - it may be that moderate agreeableness is optimal. Perhaps because it simultaneously allows people to cooperate sufficiently with group members but to discount the interests of members from outside groups. In any case, an explanation in terms of cultural group-selection doesn't necessarily need to invoke selection pressures on alleles for agreeableness per se, via the feedback process that culture can have on the gene pool. Rather, the average or modal level of agreeableness seen in a society could be to a large extent moderated simply by that society's cultural norms and practices. And perhaps societies in Europe evolved to be closer to that cultural optimum, for various historical reasons. Furthermore, this kind of cultural group-selection-fuelled optimization could have regulated the evolved tribal psychology of humans: it could help regulate just how cooperative the average person is with an in-group member, and how hostile or ambivalent they are with respect to out-groups.