The influence of cognitive ability and instructional set on causal conditional inference

We report a large study in which participants are invited to draw inferences from causal conditional sentences with varying degrees of believability. General intelligence was measured, and participants were split into groups of high and low ability. Under strict deductive-reasoning instructions, it was observed that higher ability participants were significantly less influenced by prior belief than were those of lower ability. This effect disappeared, however, when pragmatic reasoning instructions were employed in a separate group. These findings are in accord with dual-process theories of reasoning. We also took detailed measures of beliefs in the conditional sentences used for the reasoning tasks. Statistical modelling showed that it is not belief in the conditional statement per se that is the causal factor, but rather correlates of it. Two different models of belief-based reasoning were found to fit the data according to the kind of instructions and the type of inference under consideration.     

Subjectivity and unconsciousness

Concepts of the unconscious were crucial to both Jung's and Freud's thinking. Psychoanalytic and analytical psychological views of the unconscious are compared and contrasted, and both are critically reviewed. It is suggested that we need to revise our conceptualization so as to take better account of the role of the analyst's expectations and inferences, and therefore of his or her subjectivity, whenever he or she makes a clinical judgement that unconscious mental processes are in operation. Some technical implications of a revised definition of unconsciousness are considered, especially indications for self-disclosure by an analyst of his or her own experience of events within the treatment.     

On aims and methods in the neuroimaging of derived relations

Ingenious and seemingly powerful technologies have been developed recently that enable the visualization in some detail of events in the brain concomitant upon the ongoing behavioral performance of a human participant. Measurement of such brain events offers at the very least a new set of dependent variables in relation to which the independent variables familiarly manipulated in the operant laboratory may be explored. Two related paradigms in which a start has been made in such research concern the derivation of novel or emergent relations from a baseline set of trained relations, and include the phenomenon of transitive inference (TI), observed in studies of stimulus equivalence (SE) and serial learning (SL) or seriation. This paper reviews some published and forthcoming neuroimaging studies of these and related phenomena, and considers how this line of research both demands and represents a welcome synthesis between types of question and levels of explanation in behavioral science that often have been seen as antithetical.     

Where do "constraints" come from? A comparison between inferences in biology and in word learning

In this paper, as a commentary on the papers in this special issue, the nature of constraints is discussed in terms of the comparison between inferences in biology and in word learning. Young children's inferences in biology could be constrained by three sources of information: factual knowledge, expectations based on a theory, and ontological knowledge. For example, young children's inference about the length of noses could be constrained by the facts the children know about the animals, the similarity between the animals and humans, and the ontological distinction between living things and nonliving things. In the same way, young children might figure out word meanings by linguistic and pragmatic knowledge, expectations of word meanings (e.g., the whole-object assumption), and ontological knowledge. Interactions among these sources of knowledge are documented by the papers in this special issue and related studies. It is argued that learning biases such as the whole-object assumption could not be induced only by linguistic and pragmatic cues in a social context, but should be regarded as a product of the interaction between a social context and children's theories.     

Binary action based estimation of propensities

An observer is to make inference statements about a quantityp, called apropensity and bounded between 0 and 1, based on the observation thatp does or does not exceed a constantc. The propensityp may have an interpretation as a proportion, as a long-run relative frequency, or as a personal probability held by some subject. Applications in medicine, engineering, political science, and, most especially, human decision making are indicated. Bayes solutions for the observer are obtained based on prior distributions in the mixture of beta distribution family; these are then specialized to power-function prior distributions. Inference about logp and log odds is considered. Multiple-action problems are considered in which the focus of inference shifts to theprocess generating the propensitiesp, both in the case of a process parameter known to the subject and unknown. Empirical Bayes techniques are developed for observer inference aboutc when is known to the subject. A Bayes rule, a minimax rule and a beta-minimax rule are constructed for the subject when he is uncertain about.This research was partially supported by the Defense Advanced Research Projects Agency of the Department of Defense and was monitored by ONR under Contract No. N00014-77-C-0095. Any opinions, findings, conclusions or recommendations expressed herein are those of the author and do not necessarily reflect the views of the Defense Advanced Research Projects Agency, the Office of Naval Research, or Carnegie-Mellon University.     

Clinical judgment in forensic neuropsychology: A comment on the risks of claiming more than can be delivered

There is a vast and growing literature in psychology demonstrating the general limits of human judgment and clinical inference. These findings clearly apply in the new specialty of clinical neuropsychology, and there is little empirical research to support the widespread belief that judgmental accuracy correlates substantially with experience, professional stature, or reputation as a neuropsychologist. However, the demand characteristics of expert testimony in the forensic arena may encourage individual neuropsychologists to state or intimate that they have unique or special expertise in understanding brain-behavior relationships, or in predicting outcomes following cerebral insult or injury. These claims will be increasingly difficult to substantiate as attorneys become more conversant with the literature on human judgment.     

Bayesian inference for finite mixtures of generalized linear models with random effects

We present an hierarchical Bayes approach to modeling parameter heterogeneity in generalized linear models. The model assumes that there are relevant subpopulations and that within each subpopulation the individual-level regression coefficients have a multivariate normal distribution. However, class membership is not known a priori, so the heterogeneity in the regression coefficients becomes a finite mixture of normal distributions. This approach combines the flexibility of semiparametric, latent class models that assume common parameters for each sub-population and the parsimony of random effects models that assume normal distributions for the regression parameters. The number of subpopulations is selected to maximize the posterior probability of the model being true. Simulations are presented which document the performance of the methodology for synthetic data with known heterogeneity and number of sub-populations. An application is presented concerning preferences for various aspects of personal computers.     

Towards end-to-end likelihood-free inference with convolutional neural networks

Complex simulator-based models with non-standard sampling distributions require sophisticated design choices for reliable approximate parameter inference. We introduce a fast, end-to-end approach for approximate Bayesian computation (ABC) based on fully convolutional neural networks. The method enables users of ABC to derive simultaneously the posterior mean and variance of multidimensional posterior distributions directly from raw simulated data. Once trained on simulated data, the convolutional neural network is able to map real data samples of variable size to the first two posterior moments of the relevant parameter's distributions. Thus, in contrast to other machine learning approaches to ABC, our approach allows us to generate reusable models that can be applied by different researchers employing the same model. We verify the utility of our method on two common statistical models (i.e., a multivariate normal distribution and a multiple regression scenario), for which the posterior parameter distributions can be derived analytically. We then apply our method to recover the parameters of the leaky competing accumulator (LCA) model and we reference our results to the current state-of-the-art technique, which is the probability density estimation (PDA). Results show that our method exhibits a lower approximation error compared with other machine learning approaches to ABC. It also performs similarly to PDA in recovering the parameters of the LCA model.     

Young children combine sensory cues with learned information in a statistically efficient manner: But task complexity matters

Human adults are adept at mitigating the influence of sensory uncertainty on task performance by integrating sensory cues with learned prior information, in a Bayes‐optimal fashion. Previous research has shown that young children and infants are sensitive to environmental regularities, and that the ability to learn and use such regularities is involved in the development of several cognitive abilities. However, it has also been reported that children younger than 8 do not combine simultaneously available sensory cues in a Bayes‐optimal fashion. Thus, it remains unclear whether, and by what age, children can combine sensory cues with learned regularities in an adult manner. Here, we examine the performance of 6‐ to 7‐year‐old children when tasked with localizing a ‘hidden’ target by combining uncertain sensory information with prior information learned over repeated exposure to the task. We demonstrate that 6‐ to 7‐year‐olds learn task‐relevant statistics at a rate on par with adults, and like adults, are capable of integrating learned regularities with sensory information in a statistically efficient manner. We also show that variables such as task complexity can influence young children's behavior to a greater extent than that of adults, leading their behavior to look sub‐optimal. Our findings have important implications for how we should interpret failures in young children's ability to carry out sophisticated computations. These ‘failures’ need not be attributed to deficits in the fundamental computational capacity available to children early in development, but rather to ancillary immaturities in general cognitive abilities that mask the operation of these computations in specific situations.