Abstract
Background
Associative learning is fundamental to how we construct reality, providing a mechanism to create predictive constructs about the world. People with schizophrenia and other psychoses have been shown to have ‘abnormalities’ in associative learning, which may help to explain how reality distortions arise in patients. Recent studies have also indicated that associative learning may be used to distinguish genetic subpopulations in schizophrenia. The most investigated measures have been Latent inhibition, where prior exposure to a stimulus without outcome retards subsequent learning about that stimulus and Kamin blocking in which learning about one stimulus of a compound of stimuli is attenuated through prior learning with the other stimulus. Findings however, particularly for latent inhibition, have been variable, which may in part be due to the use of task designs that do not control for abnormal learning confounds such as learning rate acquisition or learned irrelevance.
We investigated latent inhibition and Kamin blocking in patients with psychosis using tasks designed to remove some confounds from previous studies such as learned irrelevance. These tasks both involve predictions of outcomes based on prior exposure to stimuli. We predicted that if patients have a generalized abnormality in forming new associations based on prior experience, then they would show abnormalities in both tasks.
Methods
Latent inhibition was measured using a method developed by us to remove a learned irrelevance confound (Granger et al, 2016, Personality and Individual Differences91: 31–39) where participants make a keyboard response to a series of letters presented on a computer screen. Slower reaction time to pre-exposed stimuli versus non-pre-exposed stimuli constitutes latent inhibition. Kamin blocking was measured using a task previously developed in (Haselgrove & Evans, 2010, Q. J. Exp Psychology, 63: 1127–1149). In this task, participants are presented on a screen with the names of foods and rate whether they will make them ill. 54 patients with psychosis and a control group were tested. Symptoms were measured using PANSS, anxiety using STICSA and IQ using WASI. Theory of mind was additionally assessed using jokes and hinting tasks. Tasks were presented using a fully counterbalanced design. Current symptoms were measured using positive and negative symptom scale (PANSS).
Results
Latent inhibition was significantly reduced in patients compared to controls, the control group showing faster reaction times to the non-pre-exposed stimuli, while patients showed no difference in reaction times between pre-exposed and non-preexposed stimuli, which was particularly evident in the early test trials. (split plot ANOVA Pre-exposure X trial X patient status interaction F 19,741 =1.76,p<0.05, between group effect of patient status F 1,39=5.2, P<0.05). Kamin blocking was also significantly reduced. There was a difference in rating of blocked versus control stimulus in controls (Wilcoxon, signed rank: Z=-2.7,P<0.005) versus no effect in patients (Z=-.975,NS). Chlorpromazine equivalent doses were not correlated with global measures of latent inhibition and Kamin blocking. Kamin blocking was associated with PANSS negative symptoms while reduced latent inhibition was associated with positive symptoms.
Discussion
These data show that using tasks that control for learning confounds such as learned irrelevance and learning rate, patients with psychosis show clear reductions both in latent inhibition and Kamin blocking. These reductions are not correlated with each other within patients and are associated with different symptoms profiles, Kamin blocking being associated with negative symptoms and latent inhibition with positive symptoms. This suggests that these tasks will be useful to measure associative learning phenomena in patients and may have potential to distinguish between different symptom profiles.