This study will examine how the brains of depressed people function during learning and respond to feedback. Participants perform a "probabilistic reversal learning task" to determine whether depressed people perform worse on a task than non-depressed people when they sometimes receive misleading negative feedback.

Right-handed healthy volunteers, people with major depression who are currently depressed or have previously been depressed and people with bipolar depression between 18 and 50 years of age may be eligible for this study. Candidates are screened with a medical and psychiatric history, physical examination, electrocardiogram, blood and urine tests. They are also interviewed to evaluate mood, sleep, energy, work and school performance, and social relationships, and asked to answer questions to investigate whether any history of paranoia, panic attacks, obsession, compulsions, suicidal thoughts, eating disturbances, and alcohol or drug abuse is present. They complete rating scales for depression, anxiety, and negative thinking; history of alcohol and tobacco use; physical movement; socioeconomic status; overall level of functioning; and depression type. Finally, they undergo a battery of neuropsychological tests to assess general intelligence, handedness, and specific cognitive abilities, including memory and concentration.

Participants perform the probabilistic reversal learning task either in a testing room seated in front of a computer or lying down while undergoing functional magnetic resonance imaging (FMRI). FMRI is a diagnostic test that uses a strong magnet and radio waves to obtain pictures of brain structure and function. For the scan, the subject lies on a narrow bed with a plastic-encased metal coil close to the head. The bed slides into the scanner - a small tunnel about 6 feet long. All subjects, whether in the testing room or in the MRI scanner, undergo the learning task as follows:

Two patterns are presented on a computer screen. One pattern is designated "correct" and the other "incorrect." Subjects are asked to choose the correct pattern on each try and are provided feedback as to whether the response was right or wrong. Sometimes the rule changes, and the pattern that was correct is now wrong, and vice versa, so that the new correct pattern must be chosen. In addition, misleading feedback is sometimes given intentionally, indicating the subject chose the wrong pattern when in fact the response was correct. Subjects should change their response only when they are sure that the rule has changed, and not because they were incorrectly told that they were wrong. For patients undergoing MRI, blood flow in the different areas of the brain is measured during the test.

After the test, outside the scanner, participants undergo additional tests of attention, memory, and concentration like those that were administered during the screening procedures.

When depressed subjects perceive they have failed while performing a neuropsychological task their performance on subsequent trials becomes impaired. Elliott et al. showed that MDD subjects were far more likely to fail on subsequent tasks after failure on one task than controls. This effect has been described as an "abnormal response' to negative feedback or a 'catastrophic response to perceived failure.'

Murphy et al. investigated this further using a probabilistic reversal learning task in which subjects were faced with occasional negative feedback that was actually false. On each trial of the task subjects choose one of two stimuli on screen, one considered correct and the other incorrect, this is followed by feedback. Instructions prior to starting the task inform the subject that sometimes the rule will change and the previously correct stimulus will be incorrect, but sometimes the computer will give them incorrect feedback. They are instructed to switch responding only when they are sure the rule has actually changed. MDD subjects demonstrate intact acquisition and reversal of probabilistic learning but are impaired in their ability to maintain a response in the face of misleading negative feedback. This profile of impairment may be relatively specific to MDD as, for example, it is not found in subjects with other neuropsychiatric conditions. This 'abnormal response' to negative feedback may represent a significant link between negative affect and cognitive impairments in depression.

During the proposed study functional magnetic resonance images will be acquired while subjects perform a version of the probabilistic reversal learning task. Four groups will take part: MDD, bipolar depressed, unmedicated MDD in remission and healthy control subjects. These images will be used to investigate differences in neural activation between the four groups, to clarify whether this specific profile of behavioral response extends to bipolar depressed subjects and to assess the mood state dependences of this effect. This would increase understanding of the specific profile of performance on this task and neural basis of this abnormal response to negative feedback that contributes to impairment of cognitive performance in depression.

• INCLUSION CRITERIA:

All subjects will be between 18 and 50 years old. Male and female subjects will be included. All subjects must be able to give written informed consent prior to participation in this study.

MDD Depressed Sample: Twenty five subjects (ages 18-50) male and female will be selected, with primary MDD currently depressed as defined by DSM-IV criteria for recurrent MDD, in a current major depressive episode and who have a first degree relative with MDD but no first degree relatives with mania, alcoholism, or antisocial personality disorder.

MDD Remission Sample: Twenty subjects (ages 18-50) male and female will be selected. Remission is defined as a period of at least three months during which the subject has not taken an antidepressant agent, with Hamilton Depression Rating Scale (HDRS; 21-item; Hamilton 1960) scores in the non-depressed range (less than 8), and with no more than one clinically significant depressive symptom.

Bipolar Depressed Sample: Twenty five subjects (ages 18-50) male and female will be selected who meet DSM-IV criteria for bipolar I or II disorder and are currently in a depressed phase.

Healthy Control Sample: Thirty subjects (ages 18-50) male and female who have not met criteria for any major psychiatric disorder will be selected. From this large sample a control subject will be matched to each depressed subject for age, gender, handedness and IQ. The control subjects will have no known first degree relatives with mood disorders.

EXCLUSION CRITERIA:

Individuals with any major medical or neurological disorder or who have received psychotropic drugs or medication likely to influence cerebral blood flow or metabolism within 3 weeks (8 weeks for fluoxetine [Prozac]), of scanning.

Bipolar subjects with rapid cycling or with current psychosis will be excluded from the protocol.

Effective treatment will not be discontinued for the purposes of this protocol, subjects will be identified who have never been treated or who have discontinued medication due to lack of efficacy, noncompliance, physician order or other reasons prior to study entry.

Individuals who meet DSM-IV criteria for alcohol and/or substance abuse within 1 year prior to screening or lifetime history of substance dependence.

Women of childbearing potential who are known to be pregnant or who have a positive pregnancy test.

Individuals who have experienced serious suicidal ideation within the past 2 months.

General exclusions for MRI imaging, such as having had a cardiac pacemaker or ferromagnetic object implanted through surgical intervention or accident (for example: shrapnel).