17 May 2009. Like researchers who study other diseases, those who examine schizophrenia tend to accentuate the negative as they try to understand what goes wrong in the disease. However, a symposium at the 2009 International Congress on Schizophrenia Research brought some fresh air into a conference room in San Diego on 29 March. It focused on successful cognition in schizophrenia and the patterns of brain activation that underlie it.
Chair Henry Holcomb, of the University of Maryland School of Medicine in Baltimore, organized the session to offset the focus on finding the lesion(s) responsible for cognitive deficits in schizophrenia. He sought to highlight adaptation and plasticity in the illness.
Holcomb presented findings from his own research that examined subjects’ performance on a visual match-to-sample task, which involves recognizing a previously shown stimulus. Subjects with schizophrenia and healthy volunteers first judged whether a rectangle was the same size as a prior one; in the next stage, they received feedback, and finally, they underwent testing in the absence of feedback. Holcomb evaluated activation in the subject groups in the two kinds of trials with correct responses: reinforcing trials preceded by a correct response and adaptive trials preceded by an incorrect response. He wanted to see if volunteers with schizophrenia and control subjects would tap the same circuits to learn the hard task.
Both groups of subjects included learners, defined as those who improved by 7 percent or more during stage two, and non-learners. Functional magnetic resonance imaging revealed patterns of brain activation that differed between patients and controls as a function of the kind of trial and whether they learned the task. During learning, particularly in the adaptive trials, controls showed increased subcortical and cortical activation. In contrast, volunteers with schizophrenia showed little subcortical activity, although they did show high involvement of the dorsolateral prefrontal cortex.
The next speaker, Kathrin Koch of the University of Jena, Germany, explored the mechanisms involved in successful working memory processing in schizophrenia. She had found that mostly similar networks underlie successful performance in patients with schizophrenia and healthy controls, but Koch wondered whether looking at static performance adequately captured the situation. Therefore, she decided to study brain activation and functional connectivity associated with increasingly successful learning.
Using blood oxygen level dependent (BOLD) MRI, Koch found that, as patients and controls overlearned the task, activation decreased in a fronto-parieto-cerebellar network. This may indicate that practice led to more efficient use of working memory resources. Although patients at first showed excess frontal activation at the start of learning, this normalized as learning continued. She further found that this normalization mirrored increased functional connectivity between brain regions involved in the task. The results underscore the potential for practice to reduce working memory impairment in schizophrenia.
Both Holcomb and Koch looked at medicated patients with schizophrenia, making it difficult to be sure that the findings stem from the disease rather than drug effects. In contrast, Florian Schlagenhauf of the Charité Universitätsmedizin Berlin, in Berlin, Germany, was able to study antipsychotic drug-free subjects with schizophrenia.
Noting that prior studies finger dopamine in reward-processing abnormalities and in positive symptoms of schizophrenia, Schlagenhauf compared patients with the illness and healthy controls on a reversal-learning task that enabled them to earn money by adapting to changing task rules. They had to choose the stimulus that offered the best likelihood of reward, even though the probability of reward associated with the stimulus sometimes changed.
Subjects with schizophrenia performed worse than controls on this task. In addition, they showed less activation, as measured by BOLD MRI, in the ventral striatum as well as in ventrolateral and medial regions of the prefrontal cortex during learning. Furthermore, relative to controls, they showed a dampened prediction error signal in the ventrolateral prefrontal cortex; this signal conveys information about a mismatch between the expected versus received reward. These findings lend weight to the notion of disturbed dopamine-related incentive processing in schizophrenia.
In his second talk, Holcomb stood in for his colleague Laura Rowland, also of the University of Maryland School of Medicine in Baltimore. They had sought to uncover the neural changes associated with successful relational learning in schizophrenia. Rowland and associates tested chronic medicated patients with schizophrenia and healthy controls on a transverse patterning task, which requires learning a series of arbitrary distinctions. Subjects with schizophrenia tend to do poorly on such tasks, but when the researchers used a stepwise approach to providing feedback, 75 percent of the patients learned the task, albeit more slowly than controls.
The researchers used functional magnetic resonance imaging (fMRI) to examine correct trials of subjects who learned the task. In control subjects, but not subjects with schizophrenia, activity in the medial temporal lobes became more centralized with learning. On the other hand, parietal activation increased in both groups as they learned the task. Since patients and controls performed similarly in the long run, the patients may have used alternate neural pathways or approached the task differently than controls. Diffusion tensor imaging revealed lower functional anisotropy in the fornix of patients versus controls, an outcome that correlated with task performance. The fornix links the hippocampus to other brain areas.
Closing out the session, Graham Murray of the University of Cambridge, UK, presented findings from two studies of normal and abnormal learning in first-episode psychosis. The first study looked at reversal learning in a set-shifting test, which required subjects to adapt to changing stimuli and rules in response to feedback. Some patients with psychosis did as well as controls on reversal learning. However, as a group, patients made more reversal mistakes than controls did; those with the greatest number of negative symptoms made the most errors. These findings, coupled with prior studies that point to a role of the orbitofrontal cortex and ventral striatum in reversal learning, implicate pathways involving these regions in the lack of motivation in schizophrenia.
Murray’s second study pursued this idea further by performing fMRI while subjects with positive symptoms of psychosis and control subjects performed an instrumental-reward learning task. They had to choose which of two stimuli would maximize their chance for reward during trials in which they received money, neutral feedback, or no feedback. Patients generally learned the task. However, controls showed greater activation than patients throughout a wide network of brain regions associated with the prediction of reward: the midbrain, striatum, cingulate, and insula.
Murray concluded that patients with early psychosis show evidence of abnormalities in the midbrain, striatum, and limbic system. Since these are the targets of dense dopaminergic innervation, the results back the notion that dopamine abnormalities hinder the processing of information about the reward potential of stimuli and may contribute to psychosis. Murray’s talk, coupled with others in this session, suggests that patients with schizophrenia can learn more kinds of tasks than previously thought, given the right conditions. They also hint that studying what goes right in schizophrenia might help researchers to figure out what goes wrong.—Victoria L. Wilcox.