The story of racial prejudice in the U.S. over the past several decades is a tale of good and bad news. On the mostly positive side, surveys of the American public suggest that overt prejudice—biases to which people are willing to admit—has been on the steady decline (although some data suggest an uptick following the presidential election of Barack Obama). On the negative side, prejudice, even in its ugliest forms, is far from eradicated. In the weeks preceding my writing of this column racial slurs surfaced on the gates of the home of basketball superstar LeBron James, and nooses were found hanging at museums in our nation’s capital.

What’s more, such overt prejudice might only be the tip of a massive iceberg. A number of prominent scholars have maintained that a good deal of racial bias has merely “gone underground,” assuming insidious forms such as implicit prejudice. Although the science of implicit prejudice is controversial, few researchers dispute that bigotry is at times manifested in subtle ways.

Against this backdrop, the concept of “microaggressions” has recently received a flurry of attention. Coined in 1970 by Harvard University psychiatrist Chester Pierce, the term “microggression” refers to a subtle slight or snub directed toward historically stigmatized individuals, especially minorities. The concept lay largely dormant until 2007, when an influential article by Columbia University counseling psychologist Derald Wing Sue and his co-authors brought it to the attention of a mainstream academic audience. According to Sue and his collaborators, the toxicity of microaggressions stems largely from their ambiguity.

Victims of these often largely invisible but pernicious statements and actions find themselves trapped in a catch-22. If they ignore microaggressions directed their way, they risk becoming the target of future transgressions from the same “perpetrators,” the term commonly used in the microaggression literature to refer to people who regularly emit such statements and actions. In contrast, if victims accuse perpetrators of aggressing against them, they risk being accused of hypersensitivity and even paranoia.

Sue and co-authors further contend that each microaggression carries a single implicit message. For example, according to them, the microaggression “America is a melting pot” communicates that minority individuals should assimilate into the broader culture, and the microaggression “You are so articulate” communicates that most minority individuals are inarticulate. When encountered frequently over long stretches of time, Sue and colleagues argue, microaggressions exert a detrimental impact on recipients, contributing to low self-esteem and, in some cases, clinical levels of depression, anxiety and other mental health problems.

Over the past few years, the term “microaggression” has become widely used on college and university campuses as well as in scores of businesses. Hundreds of institutions of higher learning now distribute standard lists of microaggressions, many of them modeled after a compendium that appeared in Sue and colleagues’ 2007 article, and warn faculty and staff members to steer clear of them.

Many colleges and universities have also instituted training programs to educate faculty staff about the hazards of microaggressions; these programs have caught on in numerous corporations, too. And several Facebook pages are dedicated to the reporting of microaggressions. Not surprisingly, in 2015 the Global Language Monitor dubbed “microaggression” its word of the year.

To be clear, Sue and his collaborators deserve credit for helping to bring the prevalence and importance of subtle prejudice to wider public and academic attention. But is the microaggression concept grounded in solid science, and is it helpful? It is certainly possible microaggressions can in some cases be harmful, especially when people are exposed to them repeatedly for years, and we need to study this phenomenon better so that we can encourage difficult conversations, not squelch them. Nevertheless, given the provisional state of the literature, it is all but impossible to determine the degree to which people’s reactions to microaggressions are attributable to these stimuli themselves as opposed to people’s subjective reactions to them.

In an article published earlier this year in the Association for Psychological Science journal Perspectives on Psychological Science, I canvassed the extant literature to address these questions. In general, I found that the sizable program of research dedicated to microaggressions raises far more questions than answers, and is far too preliminary to justify real-world applications, including training programs.

For one thing, microaggressions as Sue and others conceptualize them lie entirely in the eyes of the beholder. Therefore, if a person feels “microaggressed” against, he or she is automatically deemed to be the victim of a microaggression. The problems here are twofold: First, if person A is offended by a statement but person B is not, this would mean it both is and is not a microaggression, a proposition that is patently illogical. Second, science hinges on the ability to corroborate findings using converging sources of evidence. If a concept is entirely subjective, it is exceedingly difficult to study it scientifically, let alone subject it to rigorous tests.

Further, the very term “microaggression” is fraught, as it implies that the person emitting microaggressions is behaving aggressively. Yet, Sue and his collaborators admit most microaggressions are inadvertent. This acknowledgement leads to another logical contradiction, because psychologists almost invariably define aggression as intentional harm directed toward another person (or animal). The worry here is more than semantic. Psychological research demonstrates that if person A believes that person’s B’s actions are intentionally hostile, he or she is more likely to respond aggressively in turn. People’s perceived motives matter.

Like many psychological concepts, microaggression has fuzzy boundaries. That fact by itself is not troubling. Psychologists routinely conduct high-quality research on such concepts as intelligence, impulsivity and depression, none of which lends itself to a strict, dictionary-type (or what scientists call “operational”) definition.

Still, the microaggression concept is so nebulously defined that virtually any statement or action that might offend someone could fall within its capacious borders. For example, according to Sue and his colleagues, saying that “America is a land of opportunity” is a microaggression. Yet many nonprejudiced Americans, including many minorities, would surely endorse that assertion. In a study published in a major journal one researcher argued that a clinical supervisor being overly critical of a trainee counts as a microaggression; yet, he or she being insufficiently critical of this same trainee counts as a microaggression, too.

In some cases, the boundaries of the microaggression concept have become so vast as to invite ridicule. Last year an employee forum at the University of North Carolina at Chapel Hill proclaimed that telling a co-worker, “I love your shoes” or organizing golf outings with fellow colleagues were microaggressions (in the first case, because doing so might be perceived as patronizing, in the second case because doing so might be perceived as presumptuous). And a group of researchers recently labeled the phrase “God bless you” following a sneeze as a microaggression, presumably because it might offend certain nonreligious individuals. To his credit, even Sue has expressed misgivings with the increasingly indiscriminate application of the microaggression concept. Yet without considerably greater clarity regarding the definition of microaggressions, misuse and abuse of the concept seems virtually inevitable.

A final key problem is that the microaggression literature neglects to distinguish between the perceived and the perceiver. For decades, psychologists have recognized that our reactions to the world are shaped by both reality and our interpretations of it. Research demonstrates certain people are marked by consistently high levels of hostile attributional bias: a propensity to perceive aggressive intent in response to ambiguous stimuli.

The science aside, it is crucial to ask whether conceptualizing the interpersonal world in terms of microaggressions does more good than harm. The answer is “We don’t know.” Still, there are reasons for concern. Encouraging individuals to be on the lookout for subtle, in some cases barely discernible, signs of prejudice in others puts just about everyone on the defensive. Minority individuals are likely to become chronically vigilant to minor indications of potential psychological harm whereas majority individuals are likely to feel a need to walk on eggshells, closely monitoring their every word and action to avoid offending others. As a consequence, microaggression training may merely ramp up already simmering racial tensions.

None of this means the microaggression concept is useless. It might be helpful if viewed as the beginning, not as the end, of a constructive and mutually enlightening conversation between two people of differing backgrounds. If instead of saying, “You microaggressed against me. I am offended by your prejudice and you need to stop,” people were instead to say, “You probably didn’t mean to do so, and perhaps I’m taking this wrong way, but I was a bit hurt by what you said. Maybe we’re just misunderstanding each other. Let’s talk,” college campuses and businesses would be far better off. We need to encourage difficult conversations, not stifle them.