A well-established notion in cognitive neuroscience proposes that multiple brain systems contribute to choice behaviour. These include: (1) a model-free system that uses values cached from the outcome history of alternative actions, and (2) a model-based system that considers action outcomes and the transition structure of the environment. The widespread use of this distinction, across a range of applications, renders it important to index their distinct influences with high reliability. Here we consider the two-stage task, widely considered as a gold standard measure for the contribution of model-based and model-free systems to human choice. We tested the internal/temporal stability of measures from this task, including those estimated via an established computational model, as well as an extended model using drift-diffusion. Drift-diffusion modeling suggested that both choice in the first stage, and RTs in the second stage, are directly affected by a model-based/free trade-off parameter. Both parameter recovery and the stability of model-based estimates were poor but improved substantially when both choice and RT were used (compared to choice only), and when more trials (than conventionally used in research practice) were included in our analysis. The findings have implications for interpretation of past and future studies based on the use of the two-stage task, as well as for characterising the contribution of model-based processes to choice behaviour. In this paper, we report a reliability analysis for the estimation of “model basedness”—a psychological construct that informs a wealth of studies in animal, human and clinical research. We consider an exemplar paradigm, the two-step task, widely used in the recent literature. We report low reliability for model-agnostic model-based estimates, as well as computational model parameter estimates. We suggest how a model-based/free trade-off might affect reaction-time variability in this task, and go on to suggest use of model parameter estimates based on a combination of choice and RT. Finally, we demonstrate that combining choice and RT estimates improves both model-agnostic and algorithmic model-based estimates.
Introduction. When deciding about the cause underlying serially presented events, patients with delusions utilise fewer events than controls, showing a “Jumping-to-Conclusions” bias. This has been widely hypothesised to be because patients expect to incur higher costs if they sample more information. This hypothesis is, however, unconfirmed. Methods. The hypothesis was tested by analysing patient and control data using two models. The models provided explicit, quantitative variables characterising decision making. One model was based on calculating the potential costs of making a decision; the other compared a measure of certainty to a fixed threshold. Results. Differences between paranoid participants and controls were found, but not in the way that was previously hypothesised. A greater “noise” in decision making (relative to the effective motivation to get the task right), rather than greater perceived costs, best accounted for group differences. Paranoid participants also deviated from ideal Bayesian reasoning more than healthy controls. Conclusions. The Jumping-to-Conclusions Bias is unlikely to be due to an overestimation of the cost of gathering more information. The analytic approach we used, involving a Bayesian model to estimate the parameters characterising different participant populations, is well suited to testing hypotheses regarding “hidden” variables underpinning observed behaviours.
Self-esteem is shaped by the appraisals we receive from others. Here, we characterize neural and computational mechanisms underlying this form of social influence. We introduce a computational model that captures fluctuations in self-esteem engendered by prediction errors that quantify the difference between expected and received social feedback. Using functional MRI, we show these social prediction errors correlate with activity in ventral striatum/subgenual anterior cingulate cortex, while updates in self-esteem resulting from these errors co-varied with activity in ventromedial prefrontal cortex (vmPFC). We linked computational parameters to psychiatric symptoms using canonical correlation analysis to identify an ‘interpersonal vulnerability’ dimension. Vulnerability modulated the expression of prediction error responses in anterior insula and insula-vmPFC connectivity during self-esteem updates. Our findings indicate that updating of self-evaluative beliefs relies on learning mechanisms akin to those used in learning about others. Enhanced insula-vmPFC connectivity during updating of those beliefs may represent a marker for psychiatric vulnerability.
Viewing the brain as an organ of approximate Bayesian inference can help us understand how it represents the self. We suggest that inferred representations of the self have a normative function: to predict and optimise the likely outcomes of social interactions. Technically, we cast this predict-and-optimise as maximising the chance of favourable outcomes through active inference. Here the utility of outcomes can be conceptualised as prior beliefs about final states. Actions based on interpersonal representations can therefore be understood as minimising surprise - under the prior belief that one will end up in states with high utility. Interpersonal representations thus serve to render interactions more predictable, while the affective valence of interpersonal inference renders self-perception evaluative. Distortions of self-representation contribute to major psychiatric disorders such as depression, personality disorder and paranoia. The approach we review may therefore operationalise the study of interpersonal representations in pathological states.
Impairments in metacognition, the ability to accurately report one’s performance, are common in patients with psychiatric disorders, where a putative neuromodulatory dysregulation provides the rationale for pharmacological interventions. Previously, we have shown how unexpected arousal modulates metacognition (Allen et al., 2016). Here, we report a double-blind, placebo-controlled, study that examined specific effects of noradrenaline and dopamine on both metacognition and perceptual decision making. Signal theoretic analysis of a global motion discrimination task with adaptive performance staircasing revealed that noradrenergic blockade (40 mg propranolol) significantly increased metacognitive performance (type-II area under the curve, AUROC2), but had no impact on perceptual decision making performance. Blockade of dopamine D2/3 receptors (400 mg amisulpride) had no effect on either metacognition or perceptual decision making. Our study is the first to show a pharmacological enhancement of metacognitive performance, in the absence of any effect on perceptual decision making. This enhancement points to a regulatory role for noradrenergic neurotransmission in perceptual metacognition.
OCon behalf of The Neuroscience in Psychiatry Network Research Consortium
on behalf of The Neuroscience in Psychiatry Network Research Consortium
+2
Pavlovian influences are important in guiding decision-making across health and psychopathology. There is an increasing interest in using concise computational tasks to parametrise such influences in large populations, and especially to track their evolution during development and changes in mental health. However, the developmental course of Pavlovian influences is uncertain, a problem compounded by the unclear psychometric properties of the relevant measurements. We assessed Pavlovian influences in a longitudinal sample using a well characterised and widely used Go-NoGo task. We hypothesized that the strength of Pavlovian influences and other ‘psychomarkers’ guiding decision-making would behave like traits. As reliance on Pavlovian influence is not as profitable as precise instrumental decision-making in this Go-NoGo task, we expected this influence to decrease with higher IQ and age. Additionally, we hypothesized it would correlate with expressions of psychopathology. We found that Pavlovian effects had weak temporal stability, while model-fit was more stable. In terms of external validity, Pavlovian effects decreased with increasing IQ and experience within the task, in line with normative expectations. However, Pavlovian effects were poorly correlated with age or psychopathology. Thus, although this computational construct did correlate with important aspects of development, it does not meet conventional requirements for tracking individual development. We suggest measures that might improve psychometric properties of task-derived Pavlovian measures for future studies. Choice behaviour is guided by Pavlovian influences, so that particular features of a situation, e.g. if one seeks to gain rewards to avoid losses, privilege certain decisions over others–here, to be active versus vs. inactive respectively. Such privileging may be useful but may also impair optimal instrumental behaviour. We examined the balance of Pavlovian and instrumental guidance of choice in healthy, 14-24-year-old participants and found that young people with higher IQ relied less on Pavlovian guidance. Experience with a task changed Pavlovian guidance in a rational manner, so that it was used less when unnecessary. On the other hand, the degree of Pavlovian influence was not a highly stable trait and did not depend on age or mood. The degree of unpredictability in choice emerged as a crucial individual characteristic, associated with stronger Pavlovian influences yet more stable than them as a disposition.