Rational Adaptation in Lexical Prediction: The Influence of Prediction Strength

Recent studies indicate that the processing of an unexpected word is costly when the initial, disconfirmed prediction was strong. This penalty was suggested to stem from commitment to the strongly predicted word, requiring its inhibition when disconfirmed. Additional studies show that comprehenders rationally adapt their predictions in different situations. In the current study, we hypothesized that since the disconfirmation of strong predictions incurs costs, it would also trigger adaptation mechanisms influencing the processing of subsequent (potentially) strong predictions. In two experiments (in Hebrew and English), participants made speeded congruency judgments on two-word phrases in which the first word was either highly constraining (e.g., “climate,” which strongly predicts “change”) or not (e.g., “vegetable,” which does not have any highly probable completion). We manipulated the proportion of disconfirmed predictions in highly constraining contexts between participants. The results provide additional evidence of the costs associated with the disconfirmation of strong predictions. Moreover, they show a reduction in these costs when participants experience a high proportion of disconfirmed strong predictions throughout the experiment, indicating that participants adjust the strength of their predictions when strong prediction is discouraged. We formulate a Bayesian adaptation model whereby prediction failure cost is weighted by the participant’s belief (updated on each trial) about the likelihood of encountering the expected word, and show that it accounts for the trial-by-trial data.

Uncovering complex disease subtypes by integrating clinical data and imputed transcriptome from genome-wide association studies: Applications in psychiatry and cardiovascular medicine

Classifying patients into clinically and biologically homogenous subgroups will facilitate the understanding of disease pathophysiology and development of more targeted prevention and intervention strategies. Traditionally, disease subtyping is based on clinical characteristics alone, however disease subtypes identified by such an approach may not conform exactly to the underlying biological mechanisms. Very few studies have integratedgenomic profiles(such as those from GWAS) with clinical symptoms for disease subtyping.In this study, we proposed a novel analytic framework capable of finding subgroups of complex diseases by leveraging both GWAS-predicted gene expression levels and clinical data by a multi-view bicluster analysis. This approach connects SNPs to genes via their effects on expression, hence the analysis is more biologically relevant and interpretable than a pure SNP-based analysis. Transcriptome of different tissues can also be readily modelled. We also proposed various new evaluation or validation metrics, such as a newly modified ‘prediction strength’ measure to assess generalization of clustering performance. The proposed framework was applied to derive subtypes for schizophrenia, and to stratify subjects into different levels of cardiometabolic risks.Our framework was able to subtype schizophrenia patients with diverse prognosis and treatment response. We also applied the framework to the Northern Finland Cohort (NFBC) 1966 dataset, and identified high- and low cardiometabolic risk subgroups in a gender-stratified analysis. Our results suggest a more data-driven and biologically-informed approach to defining metabolic syndrome. The prediction strength was over 80%, suggesting that the cluster model generalizes well to new datasets. Moreover, we found that the genes ‘blindly’ selected by the cluster algorithm are significantly enriched for known susceptibility genes discovered in GWAS of schizophrenia and cardiovascular diseases, providing further support to the validity of our approach. The proposed framework may be applied to any complex diseases, and opens up a new approach to patient stratification.

Prediction strength modulates responses in human area CA1 to sequence violations

Emerging human, animal, and computational evidence suggest that, within the hippocampus, stored memories are compared with current sensory input to compute novelty, i.e., detecting when inputs deviate from expectations. Hippocampal subfield CA1 is thought to detect mismatches between past and present, and detected novelty is thought to modulate encoding processes, providing a mechanism for gating the entry of information into memory. Using high-resolution functional MRI, we examined human hippocampal subfield and medial temporal lobe cortical activation during prediction violations within a sequence of events unfolding over time. Subjects encountered sequences of four visual stimuli that were then reencountered in the same temporal order (Repeat) or a rearranged order (Violation). Prediction strength was manipulated by varying whether the sequence was initially presented once (Weak) or thrice (Strong) prior to the critical Repeat or Violation sequence. Analyses of blood oxygen level-dependent signals revealed that task-responsive voxels in anatomically defined CA1, CA23/dentate gyrus, and perirhinal cortex were more active when expectations were violated than when confirmed. Additionally, stronger prediction violations elicited greater activity than weaker violations in CA1, and CA1 contained the greatest proportion of voxels displaying this prediction violation pattern relative to other medial temporal lobe regions. Finally, a memory test with a separate group of subjects showed that subsequent recognition memory was superior for items that had appeared in prediction violation trials than in prediction confirmation trials. These findings indicate that CA1 responds to temporal order prediction violations, and that this response is modulated by prediction strength.