People with schizophrenia recognize speech poorly under multiple-people-talking (informational masking) conditions. In reverberant environments, direct-wave signals from a speech source are perceptually integrated with the source reflections (the precedence effect), forming perceived spatial separation (PSS) between different sources and consequently improving target-speech recognition against informational masking. However, the brain substrates underlying the schizophrenia-related vulnerability to informational masking and whether schizophrenia affects the unmasking effect of PSS are largely unknown.

Using psychoacoustic testing and functional magnetic resonance imaging, respectively, the speech recognition under either the PSS or perceived spatial co-location (PSC) condition and the underlying brain substrates were examined in 20 patients with schizophrenia and 16 healthy controls.

Speech recognition was worse in patients than controls. Under the PSS (but not PSC) condition, speech recognition was correlated with activation of the superior parietal lobule (SPL), and target speech-induced activation of the SPL, precuneus, middle cingulate cortex and caudate significantly declined in patients. Moreover, the separation (PSS)-against-co-location (PSC) contrast revealed (1) activation of the SPL, precuneus and anterior cingulate cortex in controls, (2) suppression of the SPL and precuneus in patients, (3) activation of the pars triangularis of the inferior frontal gyrus and middle frontal gyrus in both controls and patients, (4) activation of the medial superior frontal gyrus in patients, and (5) impaired functional connectivity of the SPL in patients.

Introducing the PSS listening condition efficiently reveals both the brain substrates underlying schizophrenia-related speech-recognition deficits against informational masking and the schizophrenia-related neural compensatory strategy for impaired SPL functions.