Ability, breadth, and parsimony in computational models of higher-order cognition

Computational models will play an important role in our understanding of human higher‐order cognition. How can a model's contribution to this goal be evaluated? This article argues that three important aspects of a model of higher‐order cognition to evaluate are (a) its ability to reason, solve problems, converse, and learn as well as people do; (b) the breadth of situations in which it can do so; and (c) the parsimony of the mechanisms it posits. This article argues that fits of models to quantitative experimental data, although valuable for other reasons, do not address these criteria. Further, using analogies with other sciences, the history of cognitive science, and examples from modern‐day research programs, this article identifies five activities that have been demonstrated to play an important role in our understanding of human higher‐order cognition. These include modeling within a cognitive architecture, conducting artificial intelligence research, measuring and expanding a model's ability, finding mappings between the structure of different domains, and attempting to explain multiple phenomena within a single model.     

Current status of the Rorschach Inkblot Method

The current scientific, clinical, and professional status of the Rorschach Inkblot Method (RIM) is reviewed with respect to its psychometric properties, the applied purposes it can be expected to serve, the extent of its use, and the nature of prevailing attitudes toward it. Available evidence indicates that the RIM is a psychometrically sound measuring instrument that provides valid assessments of personality characteristics and can facilitate differential diagnosis and treatment planning and evaluation. The RIM continues as in the past to be widely used by both clinicians and researchers. However, the esteem in which it is held by practitioners, who are generally agreed that clinical psychologists should be competent in Rorschach assessment, is not universally shared by academicians, many of whom presently question the future place of Rorschach training in graduate education.     

The Importance of Complexity in Model Selection

Model selection should be based not solely on goodness-of-fit, but must also consider model complexity. While the goal of mathematical modeling in cognitive psychology is to select one model from a set of competing models that best captures the underlying mental process, choosing the model that best fits a particular set of data will not achieve this goal. This is because a highly complex model can provide a good fit without necessarily bearing any interpretable relationship with the underlying process. It is shown that model selection based solely on the fit to observed data will result in the choice of an unnecessarily complex model that overfits the data, and thus generalizes poorly. The effect of over-fitting must be properly offset by model selection methods. An application example of selection methods using artificial data is also presented. Copyright 2000 Academic Press.     

Color appearance of surfaces viewed through fog

How do the colors of surfaces seen through fog depend on the chromatic properties of the fog? Prior work (e.g. Chen and D'Zmura, 1998 Perception 27 595-608) shows that the colors of surfaces seen through a transparent filter can be described by a convergence model. The convergence model takes into account color shift and change in contrast. Whether the convergence model can also be applied to fog was tested experimentally with an asymmetric matching task. In computer graphic simulation, observers adjusted the color of a surface seen through fog in order to match the color of a surface seen in the absence of fog. The convergence model fits the data well. The results suggest that the color constancy revealed in this task with fog is described best by a model that takes into account both shift in color and change in contrast.     

Relative damping improves linear mass-spring models of goal-directed movements

A limitation of a simple linear mass-spring model in describing goal directed movements is that it generates rather slow movements when the parameters are kept within a realistic range. Does this imply that the control of fast movements cannot be approximated by a linear system? In servo-control theory, it has been proposed that an optimal controller should control movement velocity in addition to position. Instead of explicitly controlling the velocity, we propose to modify a simple linear mass-spring model. We replaced the damping relative to the environment (absolute damping) with damping with respect to the velocity of the equilibrium point (relative damping). This gives the limb a tendency to move as fast as the equilibrium point. We show that such extremely simple models can generate rapid single-joint movements. The resulting maximal movement velocities were almost equal to those of the equilibrium point, which provides a simple mechanism for the control of movement speed. We further show that peculiar experimental results, such as an 'N-shaped' equilibrium trajectory and the difficulties to measure damping in dynamic conditions, may result from fitting a model with absolute damping where one with relative damping would be more appropriate. Finally, we show that the model with relative damping can be used to model subtle differences between multi-joint interceptions. The model with relative damping fits the data much better than a version of the model with absolute damping.     

A multinomial model for short-term priming in word identification.

A simple multinomial model for short-term priming in perceptual word identification is presented. In the experiments to which the model is applied, prime words are presented just prior to a flashed target word, and subjects must decide which of 2 alternative words matches the target. The model assumes that on some proportion of trials, confusion among the words leads to the decision being based on 1 of the prime words instead of the target. In addition, it is assumed that subjects sometimes discount a prime that matches 1 of the test alternatives and so choose the alternative that does not match. With these assumptions, the model fits the data from 5 experiments (including 4 used to develop the model known as ROUSE [responding optimally with unknown sources of evidence]; D. E. Huber, R. Shiffrin, K. Lyle, & K. Ruys, 2001). The multinomial model fits the data about as well as the ROUSE model and so should lead to further development and critical testing of both models.     

Fundamental measurement of the difficulty of test items

Two different procedures were used to measure the difficulties of Scholastic Aptitude Test items: a new distribution-free procedure that uses periodic functions and LOGIST, a well-developed optimization procedure that fits a logistic model. Despite the fact that the two procedures used different data in very different ways, they obtained virtually the same numbers. The new method exemplifies “fundamental” psychological measurement in as much as it is based on counting and ordinal properties of data only.     

Goodness-of-fit and confidence intervals of approximate models

If the model for the data are strictly speaking incorrect, then how can one test whether the model fits? Standard goodness-of-fit (GOF) tests rely on strictly correct or incorrect models. But in practice the correct model is not assumed to be available. It would still be of interest to determine how good or how bad the approximation is. But how can this be achieved? If it is determined that a model is a good approximation and hence a good explanation of the data, how can reliable confidence intervals be constructed? In this paper, an attempt is made to answer the above questions. Several GOF tests and methods of constructing confidence intervals are evaluated both in a simulation and with real data from the internet-based daily news memory test.     

The EZ diffusion method: Too EZ?

The diffusion model (Ratcliff, 1978) for fast two-choice decisions has been successful in a number of domains. Wagenmakers, van der Maas, and Grasman (2007) proposed a new method for fitting the model to data (“EZ”) that is simpler than the standard chisquare method (Ratcliff & Tuerlinckx, 2002). For an experimental condition, EZ can estimate parameter values for the main components of processing using only correct response times (RTs), their variance, and accuracy, not error RTs or the shapes of RT distributions. Wagenmakers et al. suggested that EZ produces accurate parameter estimates in cases in which the chi-square method would fail-specifically, experimental conditions with small numbers of observations or with accuracy near ceiling. In this article, I counter these claims and discuss EZ’s limitations. Unlike the chi-square method, EZ is extremely sensitive to outlier RTs and is usually less efficient in recovering parameter values, and it can lead to errors in interpretation when the data do not meet its assumptions, when the number of observations in an experimental condition is small, or when accuracy in an experimental condition is high. The conclusion is that EZ can be useful in the exploration of parameter spaces, but it should not be used for meaningful estimates of parameter values or for assessing whether or not a model fits data.     

An evaluation of a computational model of lexical access: comment on Dell et al. (1997)

The computational model of lexical access proposed by G. S. Dell, M. F. Schwartz, N. Martin, E. M. Saffran, and D. A. Gagnon (1997) is evaluated. They argued that fits of their model to naming data obtained from normal and brain-damaged patients support assumptions regarding interactivity in the lexicon, global damage in aphasia, and continuity between normal and aphasic naming behavior. Additional analysis reveals that the model fits the empirical data poorly and that the claims Dell et al. made on the basis of the model's performance would not follow even if the model were accurate. Although use of a novel automatic regression procedure improved the model's fit, it cannot account for 5 of Dell et al.'s 21 patients (24%), and its limitations were found to be inherent in its design. It is argued that claims such as those made by Dell et al. can only be addressed by considering evidence from multiple related tasks and by comparing multiple computational models.     

A semiparametric approach for item response function estimation to detect item misfit

When scaling data using item response theory, valid statements based on the measurement model are only permissible if the model fits the data. Most item fit statistics used to assess the fit between observed item responses and the item responses predicted by the measurement model show significant weaknesses, such as the dependence of fit statistics on sample size and number of items. In order to assess the size of misfit and to thus use the fit statistic as an effect size, dependencies on properties of the data set are undesirable. The present study describes a new approach and empirically tests it for consistency. We developed an estimator of the distance between the predicted item response functions (IRFs) and the true IRFs by semiparametric adaptation of IRFs. For the semiparametric adaptation, the approach of extended basis functions due to Ramsay and Silverman (2005) is used. The IRF is defined as the sum of a linear term and a more flexible term constructed via basis function expansions. The group lasso method is applied as a regularization of the flexible term, and determines whether all parameters of the basis functions are fixed at zero or freely estimated. Thus, the method serves as a selection criterion for items that should be adjusted semiparametrically. The distance between the predicted and semiparametrically adjusted IRF of misfitting items can then be determined by describing the fitting items by the parametric form of the IRF and the misfitting items by the semiparametric approach. In a simulation study, we demonstrated that the proposed method delivers satisfactory results in large samples (i.e., N ≥ 1,000).     

广告代言人参与度研究:深层代言还是浅层代言?

有关广告代言人效果、消费者卷入的研究都曾引起学界的重视并已取得了丰硕的成果,然而,代言人在品牌代言时实际参与(卷入)的程度是否影响消费者对其代言效果的认识,当代言不同类别的产品时这种代言效果是否依然存在,此外,消费者的个性特点是否也会影响这种代言效果等仍需要理论上的深入探讨。本文通过2(代言人高参与度/代言人低参与度)×2(高认知需求/低认知需求)×2(享乐型产品/功能型产品)的实验设计,全面、系统地探讨了代言人参与的深浅程度对消费者品牌态度和品牌文化感知的影响,以及消费者认知需求对两者之间的调节作用。结果表明,无论是代言享乐型还是功能型产品,代言人参与程度都会对消费者品牌态度和品牌文化感知有显著的影响,而且消费者对于明星代言享乐型产品的品牌态度显著高于功能型产品。特别地,这种作用还受到消费者认知需求水平的影响。即当消费者认知需求水平高时,名人参与代言的程度会对其品牌态度造成积极的正向的影响;而当消费者认知需求水平低时,名人参与代言程度则不会对其品牌态度造成显著影响。     

Reconsidering the lessons of the lottery for knowledge and belief

In this paper, I propose that one can have reason to choose a few tickets in a very large lottery and arbitrarily believe of them that they will lose. Such a view fits nicely within portions of Lehrer??s theory of rational acceptance. Nonetheless, the reasonability of believing a lottery ticket will lose should not be taken to constitute the kind of justification required in an analysis of knowledge. Moreover, one should not accept what one takes to have a low chance of being true. Accordingly, one should take care not to believe of too many tickets that they are to lose. Finally, while arbitrariness is no absolute barrier to epistemic reasonability, one may not be able to believe that one??s lottery ticket will lose if one cannot regard oneself as knowing it will lose.     

Estimating parameters of the diffusion model: approaches to dealing with contaminant reaction times and parameter variability

Three methods for fitting the diffusion model (Ratcliff, 1978) to experimental data are examined. Sets of simulated data were generated with known parameter values, and from fits of the model, we found that the maximum likelihood method was better than the chi-square and weighted least squares methods by criteria of bias in the parameters relative to the parameter values used to generate the data and standard deviations in the parameter estimates. The standard deviations in the parameter values can be used as measures of the variability in parameter estimates from fits to experimental data. We introduced contaminant reaction times and variability into the other components of processing besides the decision process and found that the maximum likelihood and chi-square methods failed, sometimes dramatically. But the weighted least squares method was robust to these two factors. We then present results from modifications of the maximum likelihood and chi-square methods, in which these factors are explicitly modeled, and show that the parameter values of the diffusion model are recovered well. We argue that explicit modeling is an important method for addressing contaminants and variability in nondecision processes and that it can be applied in any theoretical approach to modeling reaction time.     

Multivariate analysis with linearizable regressions

We study the class of multivariate distributions in which all bivariate regressions can be linearized by separate transformation of each of the variables. This class seems more realistic than the multivariate normal or the elliptical distributions, and at the same time its study allows us to combine the results from multivariate analysis with optimal scaling and classical multivariate analysis. In particular a two-stage procedure which first scales the variables optimally, and then fits a simultaneous equations model, is studied in detail and is shown to have some desirable properties.     

Recovering Substantive Factor Loadings in the Presence of Acquiescence Bias: A Comparison of Three Approaches

Researchers are often advised to write balanced scales (containing an equal number of positively and negatively worded items) when measuring psychological attributes. This practice is recommended to control for acquiescence bias (ACQ). However, little advice has been given on what to do with such data if the researcher subsequently wants to evaluate a 1-factor model for the scale. This article compares 3 approaches for dealing with the presence of ACQ bias, which make different assumptions: an ipsatization approach based on the work of Chan and Bentler (CB; 1993), a confirmatory factor analysis (CFA) approach that includes an ACQ factor with equal loadings (Billiet & McClendon, 2000; Mirowsky & Ross, 1991), and an exploratory factor analysis (EFA) approach with a target rotation (Ferrando, Lorenzo-Seva, & Chico, 2003). We also examine the “do nothing” approach which fits the 1-factor model to the data ignoring the presence of ACQ bias. Our main findings are that the CFA method performs best overall and that it is robust to the violation of its assumptions, the EFA and the CB approaches work well when their assumptions are strictly met, and the “do nothing” approach can be surprisingly robust when the ACQ factor is not very strong.     

A Confirmatory Approach to Examining the Factor Structure of the Strengths and Difficulties Questionnaire (SDQ): A Large Scale Cohort Study

The aim of this study was to examine the factor structure of the Strengths and Difficulties Questionnaire (SDQ) using a Structural Confirmatory Factor Analytic approach. The Danish translation of the SDQ was distributed to 71,840 parents and teachers of 5–7 and 10–12-year-old boys and girls from four large scale cohorts. Three theoretical models were examined: 1. a model with five first order factors (i.e., hyperactivity/inattention, conduct, emotional, peer problems and prosocial), 2. a model adding two internalising and externalising second order factors to model 1, and 3. a model adding a total difficulties second order factor to model 1. Model fits were evaluated, multi-group analyses were carried out and average variance extracted (AVE) and composite reliability (CR) estimates were examined. In this general population sample, low risk sample models 1 and 2 showed similar good overall fits. Best model fits were found when two positively worded items were allowed to cross load with the prosocial scale, and cross loadings were allowed for among three sets of indicators. The analyses also revealed that model fits were slightly better for teachers than for parents and better for older children than for younger children. No convincing differences were found between boys and girls. Factor loadings were acceptable for all groups, especially for older children rated by teachers. Some emotional, peer, conduct and prosocial subscale problems were revealed for younger children rated by parents. The analyses revealed more internal consistency for older children rated by teachers than for younger children rated by parents. It is recommended that model 1 comprising five first order factors, or alternatively model 2 with additionally two internalising/externalising second order factors, should be used when employing the SDQ in low risk epidemiological samples.     

The consumer psychology of customer–brand relationships: Extending the AA Relationship model

The Attachment–Aversion (AA) Relationship model offers a unifying model of customer–brand relationships. To develop it further as a relevant consumer-psychology model, future research should examine three key factors: how brand perception differs from person perceptions; what role brand experiences play as determinants of customer–brand relationships, and how the AA Relationship model fits with other brand frameworks. The author offers insights and suggestions on how to address these three tasks.     

Moorean pluralism as a solution to the incommensurability problem

Several prominent ethical philosophers have attempted to demonstrate that there exist instances or types of value that are of crucial moral significance but which cannot legitimately be compared with one another. Bernard Williams and Michael Stocker, for example, argue that it can sometimes be rational to regret having chosen the all-things-considered better of two alternatives, and that this sense of regret entails that the goodness of the worse option is not made up for by and is therefore incommensurable with that of the better. Joseph Raz and others have made similar points. In this paper, I propose a theory of value that is monistic in that it countenances just one sort of morally crucial value, but pluralistic in that several distinct properties bearer this value. I then explain how this view avoids incommensurable values without doing violence to the core intuitions that seemed to necessitate them, and how it fits into a larger conception of morality, right conduct, and moral psychology.