We may not like to admit it, human beings engage in "like with like" mating, what population geneticists call assortative mating.  Individuals are more likely to mate with other individuals[1] with characteristics like their own (i.e., Brangelina).  Humans practice assortative mating based on many characteristics, such as height, IQ, skin color, social status, and religion.  And, according to this Hot or Not^TM study, humans also practice assortative dating based on looks. 

We expect good looking folks to get together with other good looking folks.  We wonder about the physically attractive male who is with a physically unattractive female.  We make unpleasant assumptions about the pecuniary motivations of the beautiful woman who is with an unattractive man. 

In natural populations, assortative mating can lead to the separation of a population into two subgroups.

The effect on the genotype frequencies among the progeny is
to increase the frequencies of homozygotes and reduce that of
heterozygotes.  In effect the population becomes partially subdvided
into two groups, mating taking place more frequently within than
between the groups.

- Introduction to Quantitative Genetics, 4th Ed. by DS Falconer and Trudy Mackay

Those subgroups will, as a result, experience a reduction in genetic variation in all characters, referred to as a reduction in heterogeneity as with inbreeding, but with less extreme effects in most cases.  As the number of genes involved in the trait A reduction in heterogeneity is bad for several reasons:
1. Increased inbreeding coefficient leading to more West Virginia jokes.
2. Increased probability of getting individuals that are homozygous for deleterious, recessive alleles.
3.Reduced ability to respond to changes in environment, because the response to selection in a population is directly proportional to the amount of variation in the population.

If only there was a mechanism to counteract the potentially baleful affects of assortative mating[3] based on looks.

Did you ever wake up sober after a one night stand, and the person you're next to is layin' on your arm, and they're so ugly, you'd rather chew off your arm then risk waking 'em? That's coyote ugly.
-Lil in Coyote Ugly

[Coyote ugly] is the external and visible manifestation of [beer goggles].
-Adapted from Richard Dawkins' The Extended Phenotype

beer goggles: phenomenon in which one's consumption of alcohol makes physically unattractive persons appear beautiful
-Urban Dictionary

While beer goggles may not be good for your pride, is it good for the species?

In order to examine the effects of beer goggles, we first need to establish a simple, generalizable description of human assortative mating in the Saturday night, singles bar environment.  The rugbyologist is not the first to attempt an analytical analysis of beer goggles, but may be the first to make an attempt that actually makes sense[4].

Let's think of mating as a binary decision being made by each individual involved in a particular pairing.  Mating only occurs only in the situation where both individuals are in a YES state.  Situations in which one individual is in a YES state and the other is in a NO state can lead to major awkwardness. 

Since a discussion of beer goggles focuses on attractiveness, we shall be extremely superficial and only focus on attractiveness as a determinant of mating decision.  In a traditional assortative mating framework, we would define the probability of a YES state as being dependent on the similarity between the phenotypes of the potential mates.  Population geneticists can statistically identify assortative mating by demonstrating that the phenotypes of mates are significantly more correlated than those of random pairs in the population in general.

This, however, is not a satisfyng model for human mating.  As the aforementioned Hot or Not^TM study demonstrated, attractive and unattractive individuals rate the attractiveness of others the same (i.e., no attractiveness rating preference for phenotypes similar to their own) and that individuals are more willing (increases asymptotically) to date a more attractive person than someone with the same rating.       

So, how do we model this behavior?  Let's call the individuals i and j.  Each of these individuals has a measure of both their self-perceived attractiveness (V for vanity) and their other-perceived attractiveness (B for beauty).  Then we have the probability of a YES state (Y) that is dependent on the difference between V and B (H_j=V_i-B_j).  Instead of just making things up out of whole cloth, we will base our probability equation on the Hot or Not study using the following logistic function:

 

which, I think you will agree, does a bang up job of modeling the Hot or Not data (m=0.4, ll=1).


Within this model of mate choice, we can also consider the ways in which alcohol consumption might act to increase Y.  Conventional wisdom suggests two general mechanisms for beer goggles:
1. Incorrectly estimate the difference in attractiveness between individuals (H).
2. Change the curve describing the relationship between the difference in attractiveness and the probability of a YES state (Y).

Incorrectly estimate the difference in attractiveness between individuals
The term beer goggles suggests a mechanism that alters one's perception.  There are two mechanisms by which H could be incorrectly estimated.  First, one could underestimate one's own attractiveness (V)  when intoxicated, thus reducing H.  Empirical evidence suggests that this mechanism (i.e., drunks underestimating themselves) is implausible.

Second, one could overestimate the other individual's attractiveness (B) when intoxicated, thus reducing H.  Back in the good old days when they tested myths that did not involve explosions, the MythBusters^TM[5] tested this one.  Unfortunately, due a flawed experimental design[6], wide variation in results, and a small sample size, the results were inconclusive.  Other research from the University of Leicester[7] is claimed to suggest that the estimations made by men are not affected while those of women are.  The actual study (PDF) shows that the gender differences exist for age estimation, not estimation of attractiveness.

Based on these examples, it seems that incorrect estimation of differences in attractiveness is the likely mechanism of beer goggles.  

Change the Curve
The alternative explanation for beer goggles is a reduction of inhibitions, as embodied by the phrase "IN VINO VERITAS."  This phrase implies that intoxication brings out one's true self and, for the purposes of this discussion, assumes that one's true self is the equivalent of a randy chihuahua humping everything in sight.

In our simple model of human assortative mating, reduced inhibitions would be modeled as a change in the curve describing the relationship between the difference in attractiveness (H) and the probability of a YES state (Y).  There are two general variables (m and l) that can be manipulated in the equation describing our curve, which in case you forgot is:

Changing m changes the maximum for Y (when H is high), but does not change the shape of the curve much, especially at the negative values of H that are most relevant to a discussion of beer goggles.

Changing l, however, shifts the entire curve to the left, dramatically increasing Y for lower values of H as l increases.  Although the maximum Y does not change from sober to drunk for even the most attractive potential partner, small changes in l can cause large changes in Y at intermediate values of H due to the shape of the curve.

A reduction in inhibitions, particularly as modeled through an increase in l, seems like the most probably mechanism for beer goggles.  It's not that you don't know that he/she/it is ugly, it's that you don't care.

Effects of Changing the Curve
Does the beer goggles reduction of inhibitions affect assortative mating?  One of the great things about describing human assortative mating mathematically is that it becomes very amenable to computer simulations.  As described above, an increased phenotypic correlation between the parents is the signature of assortative mating.  Also above, we suggested that beer goggles might be a mechanism to reduce assortative mating.

To avoid really raunchy numbers and provide a random control, the phenotypic correlation between parents was normalized to a simulation of purely random mating.  Therefore, the y-axis of the following graphs should be interpreted as "parent phenotypes are X times more correlated than in random mating."

When we simulate the effect of changing m, we find that this has very little effect on the degree of assortative mating (parental phenotypic correlation approximately 87 times greater than random) over the entire spectrum of m.

In contrast, when l is changed, we see a significant reduction in assortative mating as l increases.  The phenotypic correlation between parents decreases from a maximum near 120 times random to a minimum that is effectively equivalent to random.

Within the confines of this simplistic model of human assortative mating, beer goggles is no longer a source of shame, but is a potential mechanism to support the evolutionarily important diversity of the human race.  That's right, population genetics wants you to get drunk and make poor decisions.  All for the sake of humanity.  This is also why people don't ask geneticists questions.

NOTES
1: They are already very likely to mate with themselves, although this rarely produces viable offspring.
2: This also begs the question of how we determine physical attractiveness.  It is quite possible that attractiveness is a relative measure not an objective measure.  To be physically attractive, one must be significantly different than average.  If everyone is beautiful, is no one beautiful.
3: Which, although it is in fact very American, seems freaking un-American and not at all in line with that "liberte, egalite, fraternite" stuff.  Oh, wait, that's France.  Can something be un-American and un-French simultaneously?
3: Choice of sex focus for this example was not motivated by cultural norms regarding male dominance in mate choice.  Instead it was motivated by the rugbyologist's ownership of a penis (although some will suggest that it is actually leased from mrs. rugbyologist).
4: Hopefully.
5: Patent-pending, I think.  This actually comes from one of my personal favorite episodes ever (Alcohol Myths, aired 22 October 2008), featuring not only the beer goggles test, Adam Savage drunkenly wiping out on a treadmill, Jamie Hyneman slapping the bejesus out of Adam, and the bad-assedness (but non-reloadability) of the hwacha medieval arrow rocket launcher.



6: I was going to write a long description of why this experiment was well-meaning, but flawed, and how to do it right; but I am tired.  Ask and I'll be happy to discuss.  You can see why some purported fans of the show thought this test was flawed here.  While many complaints are stupid and wrong, some have merit, and most take the overly critical tone of those who are massively jealous because they are not on television.
7: This research was actually focused on alcohol as a potentially contributing factor to statutory rape as it might cause overestimation of age.