On Jan. 20, 2026, researchers announced they have identified two biomarkers — verbal learning and identifying other peoples’ emotions — that could distinguish between people with and without schizophrenia, according to a study published in Nature Mental Health. 

“The fact that we only needed two tests was really quite shocking,” said Dr. Robert Chen, a psychiatry resident at the University of Washington School of Medicine and the paper’s lead author. Those biomarkers will not only help diagnose the condition but also contribute to its understanding and treatment, he and coauthors said. “In addition to clinical evaluations, our results suggest that a couple of easily administered cognitive tests can enhance the distinctions between people with and without schizophrenia,” said Dr. Debby Tsuang, a psychiatry professor at UW Medicine and the paper’s senior author. 

Schizophrenia is a condition characterized by hallucinations, delusions, and disorganized thinking and behavior.  There are currently good treatments for “positive symptoms” such as hallucinations and delusions. But there aren’t good treatments for “negative symptoms” such as reduced emotional expression, motivation, social engagement, and communication, which impair daily function and recovery even more.  This study focused on finding biomarkers that correspond with underlying biology and may be related to negative symptoms. 

The research team wanted to identify a subset of neurocognitive domains — areas such as the ability to pay attention, remember things and to make and execute plans — that could distinguish between patients with and without schizophrenia in the clinic. The research team analyzed a dataset collected through the Consortium on the Genetics of Schizophrenia, which Tsuang said was “only made possible with the participation and dedication of people with schizophrenia and their family members.” 

The dataset includes how people with and without schizophrenia performed on different tests that ranged from recalling a list of just-learned words to identifying emotions in images of faces. Chen used machine learning to find the minimum set of tests that distinguished between patients with and without schizophrenia. Then he tested whether that set of tests could do that using a separate set of patient data and ended up with the same findings.

The two diagnostic measures that the research team found matter most — verbal learning and emotion identification — are relatively quick to test, taking a total of five minutes. As a result, Chen said, “Down the line, you could imagine that being actually implemented in clinical workflow.” These findings indicate that, among people with schizophrenia, those two tests could uncover the most impaired neurocognitive domains and the most important areas to target treatment. 

The data used in this study represent just one moment in time. With more longitudinal data including how these tests change over time with and without treatments, the researchers could identify which therapies would help specific patients the most. The researchers are also interested in incorporating genetic markers into distinguishing people with and without schizophrenia.