Regarding the notion of putative “best” practices in social neuroscience and science in general, we contend that following established procedures has advantages, but prescriptive uniformity in methodology can obscure flaws, bias thinking, stifle creativity, and restrict exploration. Generating hypotheses is at least as important as testing hypotheses. To illustrate this process, we describe the following exploratory study. Psychiatric patients have difficulties with social functioning that affect their quality of life adversely. To investigate these impediments, we compared the performances of patients with schizophrenia and those with bipolar disorder to healthy controls on a task that involved matching photographs of facial expressions to a faceless protagonist in each of a series of drawn cartoon emotion-related situations. These scenarios involved either a single character (Nonsocial) or multiple characters (Social). The Social scenarios were also Congruent, with everyone in the cartoon displaying the same emotion, or Noncongruent (with everyone displaying a different emotion than the protagonist should). In this preliminary study, both patient groups produced lower scores than controls ( < 0.001), but did not perform differently from each other. All groups performed best on the social-congruent items and worst on the social-noncongruent items ( < 0.001). Performance varied inversely with illness duration, but not symptom severity. Complete emotional, social, cognitive, or perceptual inability is unlikely because these patient groups could still do this task. Nevertheless, the differences we saw could be meaningful functionally and clinically significant and deserve further exploration. Therefore, we stress the need to continue developing novel, alternative ways to explore social cognition in patients with psychiatric disorders and to clarify which elements of the multidimensional process contribute to difficulties in daily functioning.