Much of the recent discussion revolves around whether empathy is desirable. Here I address a different question: Is "empathy" a psychological phenomenon worthy of a distinct word to describe it?
People disagree on how empathy should be defined. Here I will treat it as a vicarious experience of the emotions of another. On this definition it is an outgrowth of theory of mind, whereby we not only recognize that another person has a particular thought or feeling, but we actually experience that same thought or emotion directly ourselves.
There is substantial epistemic risk in making claims about the intersubjective similarity of thoughts or feelings. Taken as subjective phenomena, there is no shared access through which they can be objectively compared. When we consider concepts of things in the world, we can make use of the extension of a concept to see whether, at least, that extension overlaps between two individuals. For example, if you and I were to compare our concepts of apples, we would find that in most cases we would agree on whether particular objects fit the category, despite a few fruits on the boundary. We might also describe and agree on some of the diagnostic properties we would expect to see in an object for it to be classified or rejected as an apple.
For an emotion, the exemplars are purely internal. When I refer to being "sad," my only source of information is particular experiences when I have felt sad. On our own, we are mostly able to distinguish these feelings from other sorts of feelings, such as anger or joy, because, well, the experience is quite different, as different as blue is from yellow. The way we map these feelings to words is through indirect references, in particular our visible bodily responses (tears, reddening face) or the circumstances (the death of a relative, someone taking our parking space). It is important to recognize the difference between these indirect references and the actual exemplars - they are comparable to someone heartily eating something or walking through an apple farm to determine that something is an apple.
Note of course that the radical skeptic would not have me make this distinction, because all of it is just phenomenal experience, including what we call the apple. True enough if we want to go there; and in that case we can discard empathy along with many other notions. But if we reject that position and posit a shared world, then emotions are distinctly different - the phenomenal experience is the referent, whereas in the case of the apple the experience is a means of access to the referent.
Now, advances in neuroscience over the past thirty years or so offer us a somewhat more objective means of comparison. fMRI and other technologies give us insight into some of the things that are going on inside the brain. It is important to be careful with conclusions from these technologies: usually the differences between a stimulus and control in any particular region are small, and the "lighting up" of particular brain regions that we see in academic work is a "difference image" often representing as little as 1% of the total activity. Still, these subtle differences are sometimes sufficiently diagnostic to predict what an individual is thinking about, using general categories under constrained circumstances. Further, since human brains have strong regional homology across individuals, we can certainly compare these activations intersubjectively. Let us grant then that we can identify general categories of emotions (sadness, anger, etc.) using such methods. Still, there is no homology of individual neurons in humans, and we know that the particular neurons selected through development and learning to represent particular experiences cannot be mapped intersubjectively.
With those caveats, experiments of this kind have shown that people can (depending on conditions) experience differential activations homologous to those of another. In other words, when one subject feels something we call "sad," the other subject seems to also feel something like sad, and this is not just a self-report but an actual neural correlate. So isn’t this a demonstration that empathy is real?
Sure. When someone we care about is sad, or angry, we feel a bit of sadness or anger ourselves. But this is true of virtually every intersubjective cognitive experience. When someone stubs their toe, we feel a little pain in our toe. When someone swings a golf club, some of the neurons in our own brain that activate when we swing our own golf club will activate. When someone eats an apple, our representation of apples and of eating them activate. These are sometimes called "mirror neurons," and there is nothing particularly special about them other than the fact that they form that portion of our representation of an activity that is insensitive to who is performing that activity.
Because these "mirror neuron" activations are produced through observation of another rather than through direct experience, they are easily blocked or superseded by other attentions, just as any sort of imaginative experience requires focused attention to sustain. So it is not guaranteed that we will experience these vicarious emotions; still, all we need to do is attend to the other person’s experience to have those general categories of emotions in some degree or another.
Can we, though, with further focus and attention, do better than that? Can we come closer to experiencing the same emotion that the other person has? I contend that this very quickly reaches diminishing returns and that the "sameness" of the emotion (beyond the very general categories that happen to all of us naturally) is illusory.
One of the key features of strong emotions is that they interact with episodic memories. When we are sad, we recall other times that we were sad. Further, if we are trying to feel sad, we can often drive it by recalling particular times when we were sad. These memories trigger widespread neural activity that relate to our own experience but have nothing to do with the particulars of another person’s current experience. These memories are part of the feeling, and the memories and their downstream penumbras are necessarily different intersubjectively.
But we might focus on particular properties of the circumstances. Suppose a friend is in the midst of a complex, painful breakup with a love partner. There are many mixed feelings and complex emotions. You might discuss and consider all the various circumstances and feelings. If you are emotionally attuned, you may have experienced some or most of the individual components. But these components do not comprise the experience - they are your friend's attempt to break down and convey what it feels like. It is an extremely succinct adumbration, the barest hint of what is really going on. Think of the roiling, chaotic experience you had in your own last breakup. Would you be able to characterize it in words, or by referencing other experiences as components, and feel that you thoroughly transmitted the experience? Probably not.
Also, though, the combinatoric nature of these experiences essentially rules out the possibility that either your or the friend has previously experienced a situation in the requisite combination. Each such experience, beyond very general categories, is unique, and not just superficially so. A purported empathizer does not vicariously experience the same emotional state by any reasonable definition of "same." Instead, she simply experiences vicarious emotions of her own that are what she imagines she might feel under the same circumstances, given what she knows about those circumstances.
What about in a relative sense? If two people grew up in the same culture, in a nearby geography, under similar socioeconomic circumstances, would the responsive emotions be more similar than if these factors did not match? We have good reason to doubt it. Consider the different, sometimes opposite, emotional responses siblings have to the same circumstances. So much of emotional response depends on individual psychology. Now, it is likely that similarity of background leads to more complete and accurate understanding of another’s situation. But such an understanding does not provide detailed insight into how the person feels (beyond general categories and perhaps simple combinations of them), much less offer us the ability to reproduce those feelings with any fidelity.
Consequently, the content of the word "empathy" refers to a person’s willingness to attend to or not to block the emotions that arise naturally when imagining being in similar circumstances. Any similarity with the emotional experience of the other, to the extent that such a similarity relation has any meaning at all, is very rough and general.
A common view is that empathy is a driver of compassion. If we can feel the way someone else does, so the logic goes, we are more likely to treat them kindly and help them. Given the analysis above, it seems to be the other way around. If we are feeling compassion toward someone, i.e., caring about and attending to their circumstances, then we are likely to enable our own emotions to flow, which probably includes some very general similarity of experience along with other, unrelated emotional content from our own experiences.
There is also a common notion that empathy is something we are born with or that develops early, and it is not something we can intentionally learn later. This suggests some sort of special talent akin to synaesthesia (which actually has been shown to be a real, and likely not learnable, ability), and that only certain "empathetic" people have. This discourages perfectly normal people from using the ability that they do have. People who see themselves as not empathetic may simply be disinclined to experience emotions or to attend to the emotional state of another - and these are factors that can actually be changed.
In the case of people who do apparently have a strong empathetic ability, they are mixing several ingredients - including normal emotional response to others, making the emotional state overt, asking questions, listening genuinely to the answers, attempting to express what it might feel like, and narrowing in on descriptions that seem to hit the target. This has a tendency to make the other, the original experiencer, feel like the person is sharing his or her emotions. But it is an illusion. The emotions the two people feel are not the same, and there is no special ability. Instead, again, it is better described as caring, listening, and paying attention. This is potentially a good thing, and if we were to modify the definition of empathy to mean a particular action instead of an experience, then the word might be justified.
In conclusion, to the extent that "empathy" refers to the vicarious experience of a general category of emotion based on the circumstances of another, it seems that this occurs but it is nothing cognitively special or unusual; to the extent that it refers to a deeper intersubjective similarity of emotion, this is unjustified and almost certainly illusory. When we distinguish someone who is inclined to have the vicarious experience, it is mainly because they are inclined to allow themselves to freely experience emotions generally and are attending to them in a particular case. Such people are simply caring and emotionally open, i.e., they are compassionate in context. They have no special talent or ability, nor are they doing anything unusual, that justifies using a distinct term "empathy."