Surprise beyond prediction error.

Surprise drives learning. Various neural "prediction error" signals are believed to underpin surprise-based reinforcement learning. Here, we report a surprise signal that reflects reinforcement learning but is neither un/signed reward prediction error (RPE) nor un/signed state prediction error (SPE). To exclude these alternatives, we measured surprise responses in the absence of RPE and accounted for a host of potential SPE confounds. This new surprise signal was evident in ventral striatum, primary sensory cortex, frontal poles, and amygdala. We interpret these findings via a normative model of surprise.

Variational Bayesian mixed-effects inference for classification studies.

Multivariate classification algorithms are powerful tools for predicting cognitive or pathophysiological states from neuroimaging data. Assessing the utility of a classifier in application domains such as cognitive neuroscience, brain-computer interfaces, or clinical diagnostics necessitates inference on classification performance at more than one level, i.e., both in individual subjects and in the population from which these subjects were sampled. Such inference requires models that explicitly account for both fixed-effects (within-subjects) and random-effects (between-subjects) variance components. While models of this sort are standard in mass-univariate analyses of fMRI data, they have not yet received much attention in multivariate classification studies of neuroimaging data, presumably because of the high computational costs they entail. This paper extends a recently developed hierarchical model for mixed-effects inference in multivariate classification studies and introduces an efficient variational Bayes approach to inference. Using both synthetic and empirical fMRI data, we show that this approach is equally simple to use as, yet more powerful than, a conventional t-test on subject-specific sample accuracies, and computationally much more efficient than previous sampling algorithms and permutation tests. Our approach is independent of the type of underlying classifier and thus widely applicable. The present framework may help establish mixed-effects inference as a future standard for classification group analyses.

An indirect treatment comparison (ITC) and matching-adjusted indirect comparison (MAIC) between a 15-valent (V114) and a 20-valent (PCV20) pneumococcal conjugate vaccine among healthy infants.

OBJECTIVES: Immunogenicity between 15-valent V114 (PCV15) and 20-valent PCV20 pneumococcal conjugate vaccines in healthy infants is compared in an indirect treatment comparison and matching-adjusted indirect comparison. Hypotheses: immunogenicity of V114 is non-inferior to PCV20 for all PCV13 serotypes, and superior to PCV20 for serotype 3 based on lower bound margins. METHODS: Two phase 3 pivotal studies on 3 + 1 pediatric vaccination schedule at age 2, 4, 6, and 12-15 months compared V114 (N = 858) to PCV13 (N = 856) and PCV20 (N = 1001) to PCV13 (N = 987). Infant's age and race in V114 study were matched to those in PCV20 study. Primary endpoints were serotype-specific Immunoglobulin G (IgG) response rate difference (RRD) 30 days post-dose (PD)3; IgG geometric mean concentration (GMC) ratios 30 days PD3 and PD4. RESULTS: V114 was non-inferior (marginRRD>-10%-point; marginGMCratio >0.5) to PCV20 (p-value <0.001) for all endpoints. V114 was superior (marginRRD >0%-point; marginGMCratio >1.2) to PCV20 (p-value <0.001) for serotype 3: RRD was 34.5% (95%CI 27.9%-41.1%) PD3, and IgG GMC ratios were 2.39 (95%CI 2.12-2.68) PD3 and 2.15 (95%CI 1.90-2.41) PD4. CONCLUSION: Immune response to V114 administered in a 3 + 1 schedule in healthy infants was considered non-inferior to PCV20 for all 13 PCV13 serotypes and superior for serotype 3 PD3 and PD4. CLINICAL TRIAL REGISTRATION: www.clinicaltrials.gov identifiers NCT03893448, NCT04382326.

Loss Aversion and Risk Aversion in Non-Clinical Negative Symptoms and Hypomania.

In the field of behavioral decision-making, "loss aversion" is a behavioral phenomenon in which individuals show a higher sensitivity to potential losses than to gains. Conversely, "risk averse" individuals have an enhanced sensitivity/aversion to options with uncertain consequences. Here we examine whether hypomania or negative symptoms predict the degree of these choice biases. We chose to study these two symptom dimensions because they present a common theme across many syndromes with compromised decision-making. In our exploratory study, we employed a non-clinical sample to dissociate the hypomanic from negative symptom dimension regarding choice behavior. We randomly selected a sample of 45 subjects from a student population (18-37 years) without self-reported psychiatric diagnoses (n = 835). We stratified them based on percentiles into a low hypomania/low negative symptoms (n = 15), a hypomania (n = 15), and a negative symptoms group (n = 15) using the hypomanic personality scale (HPS-30) and community assessment of psychic experiences (CAPE). Participants completed a loss aversion task consisting of forced binary choices between a monetary gamble and a riskless choice without gain or loss. We found a reduced loss aversion in participants with higher negative symptoms. In addition, risk aversion was reduced in participants with higher hypomania and negative symptoms compared to low hypomania/negative symptoms. This study adds to the understanding of underlying psychological mechanisms of loss and risk aversion. Given the partially opposing nature of hypomania and negative symptoms, further work is needed to examine whether they affect loss and risk aversion via dissociable mechanisms.

False discovery rate revisited: FDR and topological inference using Gaussian random fields.

In this note, we revisit earlier work on false discovery rate (FDR) and evaluate it in relation to topological inference in statistical parametric mapping. We note that controlling the false discovery rate of voxels is not equivalent to controlling the false discovery rate of activations. This is a problem that is unique to inference on images, in which the underlying signal is continuous (i.e., signal which does not have a compact support). In brief, inference based on conventional voxel-wise FDR procedures is not appropriate for inferences on the topological features of a statistical parametric map (SPM), such as peaks or regions of activation. We describe the nature of the problem, illustrate it with some examples and suggest a simple solution based on controlling the false discovery rate of connected excursion sets within an SPM, characterised by their volume.

Attractor models of working memory and their modulation by reward.

This work reports an empirical examination of two key issues in theoretical neuroscience: distractibility in the context of working memory (WM) and its reward dependence. While these issues have been examined fruitfully in isolation (e.g. Macoveanu et al. in Biol Cybern 96(4): 407-19, 2007), we address them here in tandem, with a focus on how distractibility and reward interact. In particular, we parameterise an observation model that embodies the nonlinear form of such interactions, as described in a recent neuronal network model (Gruber et al. in J Comput Neurosci 20:153-166, 2006). We observe that memory for a target stimulus can be corrupted by distracters in the delay period. Interestingly, in contrast to our theoretical predictions, this corruption was only partial. Distracters do not simply overwrite target; rather, a compromise is reached between target and distracter. Finally, we observed a trend towards a reduced distractibility under conditions of high reward. We discuss the implications of these findings for theoretical formulations of basal and dopamine (DA)-modulated neural bump- attractor networks of working memory.

Inhibition of the human two-pore domain potassium channel, TREK-1, by fluoxetine and its metabolite norfluoxetine.

1. Block of the human two-pore domain potassium (2-PK) channel TREK-1 by fluoxetine (Prozac) and its active metabolite, norfluoxetine, was investigated using whole-cell patch-clamp recording of currents through recombinant channels in tsA 201 cells. 2. Fluoxetine produced a concentration-dependent inhibition of TREK-1 current that was reversible on wash. The IC50 for block was 19 microM. Block by fluoxetine was voltage-independent. Fluoxetine (100 microM) produced an 84% inhibition of TREK-1 currents, but only a 31% block of currents through a related 2-PK channel, TASK-3. 3. Norfluoxetine was a more potent inhibitor of TREK-1 currents with an IC50 of 9 microM. Block by norfluoxetine was also voltage-independent. 4. Truncation of the C-terminus of TREK-1 (delta89) resulted in a loss of channel function, which could be restored by intracellular acidification or the mutation E306A. The mutation E306A alone increased basal TREK-1 current and resulted in a loss of the slow phase of TREK-1 activation. 5. Progressive deletion of the C-terminus of TREK-1 had no effect on the inhibition of the channel by fluoxetine. The E306A mutation, on the other hand, reduced the magnitude of fluoxetine inhibition, with 100 microM producing only a 40% inhibition. 6. It is concluded that fluoxetine and norfluoxetine are potent inhibitors of TREK-1. Block of TREK-1 by fluoxetine may have important consequences when the drug is used clinically in the treatment of depression.

Apathy but not diminished expression in schizophrenia is associated with discounting of monetary rewards by physical effort.

Negative symptoms in schizophrenia have been grouped into the 2 factors of apathy and diminished expression, which might be caused by separable pathophysiological mechanisms. Recently, it has been proposed that apathy could be due to dysfunctional integration of reward and effort during decision making. We asked whether apathy in particular is associated with stronger devaluation ("discounting") of monetary rewards that require physical effort. Thirty-one patients with schizophrenia and 20 healthy control participants performed a computerized effort discounting task in which they could choose to exert physical effort on a handgrip to obtain monetary rewards. This procedure yields an individual measure for the strength of effort discounting. The degree of effort discounting was strongly correlated with apathy, but not with diminished expression. Importantly, the association between apathy and effort discounting was not driven by cognitive ability, antipsychotic medication, or other clinical and demographic variables. This study provides the first evidence for a highly specific association of apathy with effort-based decision making in patients with schizophrenia. Within a translational framework, the present effort discounting task could provide a bridge between apathy as a psychopathological phenomenon and established behavioral tasks to address similar states in animals.

A Metropolis-Hastings algorithm for dynamic causal models.

Dynamic causal modelling (DCM) is a modelling framework used to describe causal interactions in dynamical systems. It was developed to infer the causal architecture of networks of neuronal populations in the brain [Friston, K.J., Harrison, L, Penny, W., 2003. Dynamic causal modelling. NeuroImage. Aug; 19 (4): 1273-302]. In current formulations of DCM, the mean structure of the likelihood is a nonlinear and numerical function of the parameters, which precludes exact or analytic Bayesian inversion. To date, approximations to the posterior depend on the assumption of normality (i.e., the Laplace assumption). In particular, two arguments have been used to motivate normality of the prior and posterior distributions. First, Gaussian priors on the parameters are specified carefully to ensure that activity in the dynamic system of neuronal populations converges to a steady state (i.e., the dynamic system is dissipative). Secondly, normality of the posterior is an approximation based on general asymptotic results, regarding the form of the posterior under infinite data [Friston, K.J., Harrison, L, Penny, W., 2003. Dynamic causal modelling. NeuroImage. Aug; 19 (4): 1273-302]. Here, we provide a critique of these assumptions and evaluate them numerically. We use a Bayesian inversion scheme (the Metropolis-Hastings algorithm) that eschews both assumptions. This affords an independent route to the posterior and an external means to assess the performance of conventional schemes for DCM. It also allows us to assess the sensitivity of the posterior to different priors. First, we retain the conventional priors and compare the ensuing approximate posterior (Laplace) to the exact posterior (MCMC). Our analyses show that the Laplace approximation is appropriate for practical purposes. In a second, independent set of analyses, we compare the exact posterior under conventional priors with an exact posterior under newly defined uninformative priors. Reassuringly, we observe that the posterior is, for all practical purposes, insensitive of the choice of prior.