In his essay Paul Bloom argues that empathy, like anger, is not a good guide for behavior unless “modified, shaped, and directed by rational deliberation.” But he equivocates on the meaning of empathy. The real target of his criticism is unfettered affective empathy, not empathy in general. As it is used in policy and popular discourse, the concept of empathy contains an element of deliberate perspective taking inconsistent with the purely affective connotation Bloom tries to foist on it. To “put yourself in her shoes” describes not a stimulus-driven, affective response, but a deliberate use of both cognitive and affective empathy.

A major weakness of Bloom’s argument is that he assumes the cognitive and affective components of empathy are cleanly separable. He claims it is not just possible but morally right to suppress affective empathy in favor of cognitive empathy. “The two are distinct,” he writes, “they emerge from different brain processes; you can have a lot of one and a little of the other.” While these two components may be conceptually distinct, theoretical arguments and empirical data suggest actual neural and cognitive processes are more integrated than Bloom appreciates.

The cognitive and affective components of empathy cannot be cleanly separated.

Indeed, an emerging body of research finds that neural systems for affective and cognitive empathy heavily influence each other. In our work, for example, we have shown that the brain systems involved in affective empathy correlate with those at work in a pro-social decision-making task. The cognitive, in short, is not sealed off from the affective.

Another weakness of Bloom’s argument is that he ignores considerable evidence showing that affective empathy is a precursor to compassion, which he thinks more morally respectable. It is true that a person incapable of empathy could learn to behave pro-socially, but someone capable of empathy has a better chance of doing so, all other things being equal. The fact that individuals with autism spectrum disorders (ASD) are not known for antisocial behavior, despite their deficits in affective empathy, is hardly grounds for dismissing the importance of affective empathy in social competence. Most individuals with ASD do complain that their condition negatively affects their social relationships and general well-being.

For these reasons, we think it is disingenuous to characterize Bloom’s line of reasoning as “against empathy.” Such a characterization implies that we can arrive at mature, rational compassion without any help from the affective domain. It ignores the strong connections between the rational processes of cognition and the emotional processes of feeling. It makes it seem that emotional empathy can only hinder intelligent compassion rather than promote it. And at worst, it could be construed as an argument against the efficacy of clinical treatments or counseling interventions based on affective, low-level forms of empathy such as imitation and gaze following. What does Bloom make of the fact that these approaches have had clinical success in promoting social cognition?

Finally, while we agree with Bloom’s criticisms of self-reporting as a tool for studying empathy, we find it peculiar that he dismisses research on these methodological grounds yet discusses in some detail an imaging study of a single subject self-reporting on his mental states—a practice at odds with current brain imaging standards that rely on group data and analysis.