Learning Differences May Predict Vulnerability to Psychoses
9 July 2006. People with psychoses can have a very different outlook on the world from those who are spared the pain of such psychological problems—delusions are a well-characterized symptom of schizophrenia, for example. Though it is still a mystery as to exactly how psychoses precipitate delusional thinking, some new research reinforces the view that difficulty with associative learning, a process that is crucial for establishing a rational view of what goes on in one’s daily life, may be at least partly responsible for delusions. The study, carried out at Aldenbrooke’s Hospital, Cambridge, England, and published in the June issue of Archive of General Psychiatry, also fingers the prefrontal cortex as the region of the brain where such associative learning problems may begin to manifest themselves.
The research, led by Paul Fletcher, University of Cambridge, set out to examine one particular facet of associative learning: prediction error. There have been suggestions that schizophrenic patients react differently from normal individuals when something they believe to be true actually turns out to be false. In the normal brain, such prediction errors are believed to play an important part in molding our view of the world because by reinforcing or negating our perception of causal associations, they help us to separate the meaningful from the insignificant as we try to put daily happenings in some kind of logical perspective.
To examine the relationship between prediction error and psychoses, first author Philip Corlett and colleagues in the University of Cambridge, and also at University College London, measured brain activity in normal people faced with a prediction error, then correlated that activity with their susceptibility to psychoses. Brain activity was measured by functional magnetic resonance imaging (fMRI), while the drug ketamine, a glutamate antagonist, was used to mimic a psychotic state.
To challenge the subjects with a prediction error, the researchers used a simple computer-based associative learning task. On the screen the volunteers saw an image of a mystery person’s meal, for example, a juicy-looking hamburger and two bananas. They learned that the meal would lead to an allergic reaction in the unfortunate mystery character. In a second stage, the volunteers learned that the burger by itself did not cause a reaction. Ergo, it must have been the bananas. This test thus creates in the volunteers an association, with a resulting strong expectation that when they see the bananas, they’ll see an allergic reaction. But next came the maddening part—the old bait and switch.
The computer experiment was set up to deliver that answer only half the time, so when the volunteer would make a logical prediction that bananas were the allergenic culprit, half of their predictions would be wrong. While the volunteers frustratingly found themselves making prediction errors, the researchers recorded fMRI scans of brain.
Corlett and colleagues found that the right prefrontal cortex (rPFC) became highly activated when the volunteers made a prediction error compared to when their prediction was confirmed (see also Fletcher et al., 2001). But if the volunteers did the test under the influence of ketamine, then the rPFC was activated even when they did not make an error, while prediction error elicited much less of a response than normal. So the psychotomimetic ketamine seems to both augment the rPFC response to normal stimuli and attenuate its response to prediction errors. “Taken together, our observations suggest that the perturbation of prediction error signaling may manifest as both decreases in appropriate signaling and increases in inappropriate signaling,” write the authors. While they also found that prediction error versus confirmation elicited different responses in the striatum, substantia nigra, and hippocampus, none of these differences were altered by administering ketamine.
These findings suggest that associative learning responses are compromised in a model of a psychotic state, but what about the development of psychoses and associative learning problems? It appears there may be a connection there, too. The authors addressed this question by correlating rPFC activation in normal individuals on placebo, with subsequent susceptibility to ketamine-induced psychosis. Significantly, they found that those individuals who had the greatest activation of the rPFC in response to a prediction error also reported the most severe psychotic episodes when given high doses of ketamine later on (plasma levels averaging 210 ng/mL). Using the Clinician-Administered Dissociative States Scale (CADSS) and the Present State Examination, which are designed to evaluate the extent of dissociation and of psychosis, respectively, Corlett and colleagues found direct relationships between prediction error-elicited rPFC activation and perceptual illusions and delusional thoughts.
The latter finding leads to some interesting chicken-and-egg scenarios. “That is, does disrupted prediction error produce the perceptual change or vice versa?” question the authors. In this case, they speculate that it is the prediction error signaling that leads to the altered perception because at the doses of ketamine used, the volunteers do not suffer from any significant perceptual change per se, but rather “feel” different about external stimuli.
One aspect of delusions this study does not address, emphasize the authors, is the “fixity” of delusions. While the study indicates links between prediction error signaling and sensory experience, it does not reveal how untenable views of the world become entrenched in the delusional state. The authors do suggest, however, that “repeated experience of abnormal prediction error signal will be an insidious process in which patients are frequently and repeatedly surprised by their experiences.” Thus, fully formed delusions may be the ultimate attempt to account for the uncertainty that seems to permeate their world.—Tom Fagan.
Corlett PR, Honey GD, Aitken MRF, Dickinson A, Shanks DR, Absalom AR, Lee M, Pomarol-Clotet E, Murray GK, McKenna PJ, Robbins TW, Bullmore ET, Fletcher PC. Frontal responses during learning predict vulnerability to the psychotogenic effects of ketamine. Arch Gen Psych. June 2006;63:611-621. Abstract