The primacy order effect in complex decision making.

The goal of the present study was to assess the role of information order in situations of complex decision making in which participants have to process a large amount of information (e.g., Dijksterhuis et al. Science 311(5763): 1005-1007, 2006). In two experiments, participants were presented with information about four cars, each characterized by 12 attributes. Immediately following the presentation of the 48 sentences describing these four cars, participants had to choose the one they would prefer to purchase. Two cars shared exactly the same positive and negative attributes, but they were displayed in a different order for each car. For one car, positive attributes were systematically displayed at the beginning while it was the reverse for the other car. The two remaining cars were used as fillers and had a lower number of positive attributes than the target cars in Experiment 1 and a higher number of positive attributes in Experiment 2. Results revealed a massive effect of information order with a clear preference for the car with positive information presented at the beginning. The second experiment further showed that this order effect was maintained and still strong even if the target cars did not have more positive attributes than the filler cars. Interestingly, in both experiments, participants never noticed that two cars were exactly characterized by the same list of attributes. These data clearly demonstrate that information order is a critical factor in complex decision-making situations involving a large amount of information.

Spelling performance on the web and in the lab.

Several dictionary websites are available on the web to access semantic, synonymous, or spelling information about a given word. During nine years, we systematically recorded all the entered letter sequences from a French web dictionary. A total of 200 million orthographic forms were obtained allowing us to create a large-scale database of spelling errors that could inform psychological theories about spelling processes. To check the reliability of this big data methodology, we selected from this database a sample of 100 frequently misspelled words. A group of 100 French university students had to perform a spelling-to-dictation test on this list of words. The results showed a strong correlation between the two data sets on the frequencies of produced spellings (r = 0.82). Although the distributions of spelling errors were relatively consistent across the two databases, the proportion of correct responses revealed significant differences. Regression analyses allowed us to generate possible explanations for these differences in terms of task-dependent factors. We argue that comparing the results of these large-scale databases with those of standard and controlled experimental paradigms is certainly a good way to determine the conditions under which this big data methodology can be adequately used for informing psychological theories.

Brain correlates of phonological recoding of visual symbols.

Learning to read involves setting up associations between meaningless visual inputs (V) and their phonological representations (P). Here, we recorded the brain signals (ERPs and fMRI) associated with phonological recoding (i.e., V-P conversion processes) in an artificial learning situation in which participants had to learn the associations between 24 unknown visual symbols (Japanese Katakana characters) and 24 arbitrary monosyllabic names. During the learning phase on Day 1, the strength of V-P associations was manipulated by varying the proportion of correct and erroneous associations displayed during a two-alternative forced choice task. Recording event related potentials (ERPs) during the learning phase allowed us to track changes in the processing of these visual symbols as a function of the strength of V-P associations. We found that, at the end of the learning phase, ERPs were linearly affected by the strength of V-P associations in a time-window starting around 200ms post-stimulus onset on right occipital sites and ending around 345ms on left occipital sites. On Day 2, participants had to perform a matching task during an fMRI session and the strength of these V-P associations was again used as a probe for identifying brain regions related to phonological recoding. Crucially, we found that the left fusiform gyrus was gradually affected by the strength of V-P associations suggesting that this region is involved in the brain network supporting phonological recoding processes.

The time course of visual influences in letter recognition.

This study builds on a specific characteristic of letters of the Roman alphabet-namely, that each letter name is associated with two visual formats, corresponding to their uppercase and lowercase versions. Participants had to read aloud the names of single letters, and event-related potentials (ERPs) for six pairs of visually dissimilar upper- and lowercase letters were recorded. Assuming that the end product of processing is the same for upper- and lowercase letters sharing the same vocal response, ERPs were compared backward, starting from the onset of articulatory responses, and the first significant divergence was observed 120 ms before response onset. Given that naming responses were produced at around 414 ms, on average, these results suggest that letter processing is influenced by visual information until 294 ms after stimulus onset. This therefore provides new empirical evidence regarding the time course and interactive nature of visual letter perception processes.

General or idiosyncratic item effects: What is the good target for models?

Recently, Adelman, Marquis, Sabatos-DeVito, and Estes (2013) formulated severe criticisms about approaches based on averaging item response times (RTs) over participants and associated methods for estimating the amount of item variance that models should try to account for. Their main argument was that item effects include stable idiosyncratic effects. In this comment, we provide supplementary empirical evidence that this assertion is indeed valid. However, the actual implications of this result are not those defended in Adelman et al. (2013), where there seems to be confusion about the precision of measures and the nature of target effects. Indeed, basic statistical considerations show that any arbitrary data precision level can be achieved in all cases using an appropriate number of observations per item, whereas general and idiosyncratic item effects are both targets of interest for modeling but in different objectives. (PsycINFO Database Record

The unbearable articulatory nature of naming: on the reliability of word naming responses at the item level.

Single-word naming is one of the most widely used experimental paradigms for studying how we read words. Following the seminal study by Spieler and Balota (Psychological Science 8:411-416, 1997), accounting for variance in item-level naming databases has become a major challenge for computational models of word reading. Using a new large-scale database of naming responses, we first provided a precise estimate of the amount of reproducible variance that models should try to account for with such databases. Second, by using an item-level measure of delayed naming, we showed that it captures not only the variance usually explained by onset phonetic properties, but also an additional part of the variance related to output processes. Finally, by comparing the item means from this new database with the ones reported in a previous study, we found that the two sets of item response times were highly reliable (r = .94) when the variance related to onset phonetic properties and voice-key sensitivity was factored out. Overall, the present results provide new guidelines for testing computational models of word naming with item-level databases.

Quick approximation of bivariate functions.

This paper presents two experiments where participants had to approximate function values at various generalization points of a square, using given function values at a small set of data points. A representative set of standard function approximation models was trained to exactly fit the function values at data points, and models' responses at generalization points were compared to those of humans. Then one defined a large class of possible models (including the best two identified predictors) and the class maximal possible prediction accuracy was evaluated. A new model of quick multivariate function approximation belonging to this class was proposed. Its prediction accuracy was close to the maximum possible, and significantly better than that of all other models tested. The new model also provided a significant account of human response variability. Finally, it was shown that this model is more particularly suitable for problems in which the visual system can perform some specific structuring of the data space. This model is therefore considered as a suitable starting point for further investigations into quick multivariate function approximation, which is to date an inadequately explored question in cognitive psychology.

Missing data imputation and corrected statistics for large-scale behavioral databases.

This article presents a new methodology for solving problems resulting from missing data in large-scale item performance behavioral databases. Useful statistics corrected for missing data are described, and a new method of imputation for missing data is proposed. This methodology is applied to the Dutch Lexicon Project database recently published by Keuleers, Diependaele, and Brysbaert (Frontiers in Psychology, 1, 174, 2010), which allows us to conclude that this database fulfills the conditions of use of the method recently proposed by Courrieu, Brand-D'Abrescia, Peereman, Spieler, and Rey (2011) for testing item performance models. Two application programs in MATLAB code are provided for the imputation of missing data in databases and for the computation of corrected statistics to test models.

Validated intraclass correlation statistics to test item performance models.

A new method, with an application program in Matlab code, is proposed for testing item performance models on empirical databases. This method uses data intraclass correlation statistics as expected correlations to which one compares simple functions of correlations between model predictions and observed item performance. The method rests on a data population model whose validity for the considered data is suitably tested and has been verified for three behavioural measure databases. Contrarily to usual model selection criteria, this method provides an effective way of testing under-fitting and over-fitting, answering the usually neglected question "does this model suitably account for these data?"

Item performance in visual word recognition.

Standard factorial designs in psycholinguistics have been complemented recently by large-scale databases providing empirical constraints at the level of item performance. At the same time, the development of precise computational architectures has led modelers to compare item-level performance with item-level predictions. It has been suggested, however, that item performance includes a large amount of undesirable error variance that should be quantified to determine the amount of reproducible variance that models should account for. In the present study, we provide a simple and tractable statistical analysis of this issue. We also report practical solutions for estimating the amount of reproducible variance for any database that conforms to the additive decomposition of the variance. A new empirical database consisting of the word identification times of 140 participants on 120 words is then used to test these practical solutions. Finally, we show that increases in the amount of reproducible variance are accompanied by the detection of new sources of variance.

Density codes and shape spaces.

This paper presents an algorithm that allows for encoding probability density functions associated to samples of points of R(n). The resulting code is a sequence of points of R(n) whose density function approximates that of the set of data points. However, contrarily to sampled data points, code points associated to two different density functions can be matched, which allows to efficiently compare such functions. Moreover, the comparison of two codes can be made invariant to a wide variety of geometrical transformations of the support coordinates, provided that the Jacobian matrix of the transformation be everywhere triangular, with a strictly positive diagonal. Such invariances are commonly encountered in visual shape recognition, for example. Thus, using this tool, one can build spaces of shapes that are suitable input spaces for pattern recognition and pattern analysis neural networks. Moreover, a parallel neural implementation of the encoding algorithm is available for 2D image data.

Function approximation on non-Euclidean spaces.

This paper presents a family of layered feed-forward networks that is able to uniformly approximate functions on any metric space, and also on a wide variety of non-metric spaces. Non-Euclidean input spaces are frequently encountered in practice, while usual approximation schemes are guaranteed to work only on Euclidean metric spaces. Theoretical foundations are provided, as well as practical algorithms and illustrative examples. This tool potentially constitutes a significant extension of the common notion of 'universal approximation capability'.

Straight monotonic embedding of data sets in Euclidean spaces.

This paper presents a fast incremental algorithm for embedding data sets belonging to various topological spaces in Euclidean spaces. This is useful for networks whose input consists of non-Euclidean (possibly non-numerical) data, for the on-line computation of spatial maps in autonomous agent navigation problems, and for building internal representations from empirical similarity data.