People care about the minds of others, attempting to understand others' thoughts, attitudes, beliefs, intentions, and emotions using a highly sophisticated process of social cognition. Others' minds are among the most complicated systems that any person will ever think about, meaning that inferences about them are also made imperfectly. Research on the processes that enable mental state inference has largely developed in isolation from research examining the accuracy of these inferences, leaving the former literature somewhat impractical and the latter somewhat atheoretical. We weave these literatures together by describing how basic mechanisms that govern the activation and application of mental state inferences help to explain systematic patterns of accuracy, error, and confidence in mind perception. Altering any of these basic processes, such as through perspective taking or increasing attention to behavioral cues, is likely to increase accuracy only in very specific circumstances. We suggest the most widely effective method for increasing accuracy is to avoid these inference processes altogether by getting another's perspective directly (what we refer to as perspective getting). Those in the midst of understanding the mind of another, however, seem largely unable to detect when they are using an effective versus ineffective strategy while engaging in mind reading, meaning that the most effective approaches for increasing interpersonal understanding are likely to be highly undervalued. Understanding how mind perception is activated and applied can explain accuracy and error, identifying effective strategies that mind readers may nevertheless fail to appreciate in their everyday lives. Through a looking glass, darkly: Using mechanisms of mind perception to identify accuracy, overconfidence, and underappreciated means for improvement.