Automatic Inhibition of Auditory Processing During Speech Production in Schizophrenia

JV Pardo, JT Lee, JC Uecker, CL Jensen, Cognitive Neuroimaging Unit, Psychiatry Service, VA Medical Center, and the Division of Neuroscience Research, Department of Psychiatry, University of Minnesota, Minneapolis, MN


Auditory hallucinations are a cardinal and disabling symptom of schizophrenia; their pathophysiology remains unknown. Several studies of patients with schizophrenia have correlated the intensity and content of reported auditory hallucinations with subvocal speech.

The functional anatomy of language processing in normal control subjects has been studied during single word processing. These investigations show that control subjects do not "hear" themselves when speaking aloud. In other words, the activity within auditory cortices during the reading aloud of single nouns presented on a video monitor ("Read") is no different than the auditory activity while passively viewing the same nouns ("Look"). In control subjects, other brain regions activated while reading as compared with looking include the following: supplementary motor cortex (SMA--involved in motor programming); mouth motor cortex (involved in motor control); and Broca's area (related to articulatory programming and subvocal rehearsal).

We tested the hypothesis that a mechanism of auditory hallucination in schizophrenic patients results from the loss of automatic inhibition of auditory processing of self-generated speech. So, schizophrenic patients were predicted to activate auditory cortices during reading as compared to looking at words.



Schizophrenic patients (N=8) were right-handed and met DSM-IIIR criteria for schizophrenia. All were psychotic (auditory hallucinations and/or delusions) at the time of the study. All were able to read fluently.


Single, common concrete nouns were presented in lower case letters at the rate of 1 every 3 seconds just above a central fixation mark on a video monitor directly in front of the patient. In the "Look" condition, subject were asked to fixate on the mark and to not read the words. In the "Read" condition, subjects were asked to read the nouns as clearly and loudly as possible.

PET Measures

Normalized regional cerebral blood flow was measured using the MVAMC PET camera, Siemens ECAT 953B. Approximately 40 mCi of O(15)-water was injected intravenously. The accumulated cerebral activity over 60 seconds (following arrival of activity to the brain) was assayed with measured attenuation correction. Images were reconstructed with filtered back-projection to a final image resolution of approximately 10 mm FWHM.

Image Analysis

The individual patients' subtraction pairs (i.e., "Read minus Look") were averaged together across subjects after stereotactic normalization.


Read Minus Look No Auditory Activation
Read Minus Look No Auditory Activation
Broca's Area Anterior Cingulate
Broca's Area Anterior Cingulate

1. Schizophrenic patients, like normal controls, have intact automatic inhibition of auditory processing during speech production. Since this suppression is intact, this mechanism can not be pathophysiologic in auditory hallucinosis.

2. Schizophrenic patients activate during reading many areas in common with normal controls: motor mouth(SMI), Broca's area, cerebellum, supplementary motor area(SMA).

3. Unlike normal control subjects, the most active brain region in schizophrenic patients during reading aloud is the anterior cingulate cortex. This brain region is activated in normal subjects only during the highest levels of linguistic processing, particularly when attention is required for task execution.


1. Automatic inhibition of auditory processing during reading aloud remains intact in patients with schizophrenia. Therefore, disinhibition of auditory processing during subvocal speech can not account for auditory hallucinations.

2. Patients with schizophrenia, unlike normal controls, recruit the anterior cingulate cortex, a component of the attention system, when reading aloud. The brain activity in these patients during reading aloud mirrors that of norm controls during difficult, attention demanding task such as verb generation.

3. These data indicate that patients with schizophrenia deploy attention to execute tasks which are typically automated and without attentional demands in controls.

Supported by the Scottish Rite Schizophrenia Research Fund, Matt Kaul Fund for Schizophrenia Research, and the Department of Veterans Affairs.
Copyright © 1996 Cognitive Neuroimaging Unit, VA Medical Center, Minneapolis, MN
Send correspondence to José V. Pardo
<Cognitive Neuroimaging Unit Home Page>