. . .but not about natural selection. 

In The Expression of the Emotions in Man and Animals, Charles Darwin argues for the universal recognition of emotion based on facial expression in birds[1] and mammals:

...the young and the old of widely different races, both with man and
animals, express the same state of mind by the same movements.

Researchers at the University of Glasgow[2], decided to test the hypothesis that relationship between facial expression and emotion is universal among humans.  To do so, they asked 13 East Asian men and 13 Western men (the study's demographic terms) to identify the emotions being expressed by a standard test set of faces used for research in the West.  Previous studies are generally interpreted toh suggest that the relationship between facial expression and emotion is universal for humans, but some studies suggested cultural heterogeneity.  This study examined this question with improved detail and technology.  In contrast to those previous studies, the University of Glasgow research team found that the Westerners had no problems identifying the emotions from facial expression, but the East Asian subjects performed worse, misidentifying disgust as anger (and possibly fear as surprise).

They also suggest a potential explanation for this result (correlation, not causation) in an analysis of eye tracking data showing where the subjects were looking during the test.  The Westerners spread their attention evenly over the entire face, while the East Asians spent more time on the eyes.

showing that Eastern observers use a culture-specific decoding strategy that is inadequate to reliably distinguish universal facial expressions of ‘‘fear’’ and ‘‘disgust.’’ Rather than distributing their fixations evenly across the face as Westerners do, Eastern observers persistently fixate the eye region.  Using a model information sampler, we demonstrate that by persistently fixating the eyes, Eastern observers sample ambiguous information, thus causing significant confusion.
-from the abstract of "Cultural confusions show that facial expressions are not universal"

They conclude that such cultural differences in the encoding and decoding of emotion could lead to difficulties in cross-cultural communication.

Does this study show that Darwin was wrong about the relationship between facial expression and emotion being universal?  Let's take a look at some aspects of the results and the conclusions. 
1. Study design
2. Meaning of universality
3. Use of information
4. Within-group variation vs between-group variation

Study Design
Actually, there is not much to complain about here, except that the study is not particularly large, with only 13 individuals in each group (maybe only twelve, if you exclude an outlier individual amongst the East Asians-more on this bad boy later).  Nothing negative here, except to say that this makes the study more suggestive (not in the fun way) and preliminary, rather than definitive.

Meaning of Universality
Does universality require inerrant identification of emotion from facial expression?  That seems a bit extreme.  So, how good is good enough?  The true metric is the utility of my reaction to the perceived stimulus of the expression.  If I interpret the aggressive growl of a grizzly bear as happiness and desiring a hug, well my reaction to the perceived stimulus is likely to be very non-utilitarian for me (perhaps more so for the bear's stomach).  If, however, I misinterpret my dog's look of fear for one of surprise, I might still have a relatively utilitarian response.  Without immediately thinking "THREAT!", my first reaction would be to look for the source of surprise, which might lead to the discovery of an aggressive grizzly bear that would allow me to reevaluate my conclusion about the dog's emotions and respond even more appropriately.  Similarly, disgust and anger could cause similar results.  I'm not going to pet a dog making either face, nor will I eat any of whatever it ate right before making such an expression.  The confusions identified in this study were not between extremely divergent emotions.  The statistical variation between cultures may not lead to a functional lack of universality at the utilitarian level.     

Use of Information
The researchers took a decoding/information approach to their results.  Because the emotion expressing faces were Western identified a statistics skewing "unusually accurate observer".  Could this individual, through familiarity, have a better understanding of the "Western" coding system than the other East Asian subjects?  This study was very focused (as it should be for a question of this difficulty), but, as a result, cannot address how non-facial information might be incorporated.

If I might offer a personal anecdote on this topic (remember anecdotes are not evidence).  A while ago, I worked with an Indian rotation student.  While we talked about his project, he would shake his head horizontally.  In my cultural idiom, this gesture signals disagreement or confusion.  In his cultural idiom, this gesture signals agreement or understanding.  This situation seemed ripe for the kind of cultural miscommunication the researchers fear.  Fortunately, I was incorporating additional information in my efforts to decode his expression, such as knowledge of his geographic and cultural origin.  Although use of this additional, non-intuitive (to me) information required a titch more cognitive effort on my part, I found that both my brain cells were able to handle the situation relatively seamlessly.

Although cultural differences in emotional coding can lead to ambiguous information, the existence of additional, correlated information (e.g., awareness of other cultures, sound, environment, etc.), which may not be used as consciously with individuals from the same culture, could greatly improve the accuracy, but was not examined in this research.

Within-group Variation vs Between-group Variation 
The presence of variation within a group does not necessarily that there is variation between groups.  Within-group variation can be observed in a characteristic that appears to be shared between groups for several reasons.  First, there is simple chance.  If one happens to pick variants from different groups that happen to share a characteristic, but do not share it with everyone in the group, one will perceive universality between groups.  For example, from the images below, one might conclude, incorrectly, that red hair is universal among mammals: 

Second, minor negative effects due to variation from the universal condition may be so small as to allow variation over small time scales (within-group), but universality over longer time scales (between-group).

A molecular example of this phenomenon can be found in the so-called ultraconserved elements.  An ultraconserved element was arbitrarily defined as a string of at least 200 base pairs of DNA that is identical in the human, mouse, and rat reference genomes.  As this study shows, these universal DNA sequences (at the scales Darwin was sequencing) vary within the human population.  Even DNA sequences that are perfectly conserved between groups (i.e., universal) can vary within groups.  In this case, the cause of this perceived issue appears to be combination of general misunderstandings of the effects of even weak selection and chance when choosing the individuals to be sequenced for hte reference genomes.  

Conclusion
Was Darwin wrong?  I think we have to give this one a pretty robust, "Meh."  It really depends on how strictly you define "universal."  And, we all know how tricky definitions can be.  At least until somebody conducts my man-grizzly bear-dog experiment.

*Hat tip to Steven Novella of NeuroLogica.

NOTES
1: In fairness, the study discussed deals with facial expression only.  Darwin is clearly referring to facial expression and body language, cause who can tell what the frak a bird is thinking from looking at its face.  Damn thing doesn't even have lips.
2: Speaking of Scotland, I highly recommend the single malt scotch Ardbeg Airigh Nam Beist, of which I happily dropped a dram the other day.