Henrik Singmann

MPTinR: Analysis of multinomial processing tree models in R

We introduce MPTinR, a software package developed for the analysis of multinomial processing tree (MPT) models. MPT models represent a prominent class of cognitive measurement models for categorical data with applications in a wide variety of fields. MPTinR is the first software for the analysis of MPT models in the statistical programming language R, providing a modeling framework that is more flexible than standalone software packages. MPTinR also introduces important features such as (1) the ability to calculate the Fisher information approximation measure of model complexity for MPT models, (2) the ability to fit models for categorical data outside the MPT model class, such as signal detection models, (3) a function for model selection across a set of nested and nonnested candidate models (using several model selection indices), and (4) multicore fitting. MPTinR is available from the Comprehensive R Archive Network at http://cran.r-project.org/web/packages/MPTinR/.

Algorithmic complexity for psychology: a user-friendly implementation of the coding theorem method

Kolmogorov-Chaitin complexity has long been believed to be impossible to approximate when it comes to short sequences (e.g. of length 5-50). However, with the newly developed coding theorem method the complexity of strings of length 2-11 can now be numerically estimated. We present the theoretical basis of algorithmic complexity for short strings (ACSS) and describe an R-package providing functions based on ACSS that will cover psychologists’ needs and improve upon previous methods in three ways: (1) ACSS is now available not only for binary strings, but for strings based on up to 9 different symbols, (2) ACSS no longer requires time-consuming computing, and (3) a new approach based on ACSS gives access to an estimation of the complexity of strings of any length. Finally, three illustrative examples show how these tools can be applied to psychology.

Does Logic Feel Good? Testing for Intuitive Detection of Logicality in Syllogistic Reasoning

Recent research on syllogistic reasoning suggests that the logical status (valid vs. invalid) of even difficult syllogisms can be intuitively detected via small changes in affective state (Morsanyi & Handley, 2012). In a series of 6 experiments, we replicated effects of logical status on liking ratings of difficult syllogisms (although their shape differs from that reported by Morsanyi and Handley), and we tested 2 alternative accounts of our and Morsanyi and Handleys findings in terms of surface features accidentally confounded with logical status: the partial-repetition hypothesis and the content-effects hypothesis. The results support the content-effects hypothesis, according to which the effects of logical status reflect differences in mean liking for the presented conclusions rather than effects of logical status itself.

Intuitive Logic Revisited: New Data and a Bayesian Mixed Model Meta-Analysis

Recent research on syllogistic reasoning suggests that the logical status (valid vs. invalid) of even difficult syllogisms can be intuitively detected via differences in conceptual fluency between logically valid and invalid syllogisms when participants are asked to rate how much they like a conclusion following from a syllogism (Morsanyi & Handley, 2012). These claims of an intuitive logic are at odds with most theories on syllogistic reasoning which posit that detecting the logical status of difficult syllogisms requires effortful and deliberate cognitive processes. We present new data replicating the effects reported by Morsanyi and Handley, but show that this effect is eliminated when controlling for a possible confound in terms of conclusion content. Additionally, we reanalyze three studies () without this confound with a Bayesian mixed model meta-analysis (i.e., controlling for participant and item effects) which provides evidence for the null-hypothesis and against Morsanyi and Handleys claim.

Deductive and inductive conditional inferences: Two modes of reasoning

A number of single- and dual-process theories provide competing explanations as to how reasoners evaluate conditional arguments. Some of these theories are typically linked to different instructions—namely deductive and inductive instructions. To assess whether responses under both instructions can be explained by a single process, or if they reflect two modes of conditional reasoning, we re-analysed four experiments that used both deductive and inductive instructions for conditional inference tasks. Our re-analysis provided evidence consistent with a single process. In two new experiments we established a double dissociation of deductive and inductive instructions when validity and plausibility of conditional problems were pitted against each other. This indicates that at least two processes contribute to conditional reasoning. We conclude that single-process theories of conditional reasoning cannot explain the observed results. Theories that postulate at least two processes are needed to account for our findings.

Probabilistic conditional reasoning: Disentangling form and content with the dual-source model ☆

The present research examines descriptive models of probabilistic conditional reasoning, that is of reasoning from uncertain conditionals with contents about which reasoners have rich background knowledge. According to our dual-source model, two types of information shape such reasoning: knowledge-based information elicited by the contents of the material and content-independent information derived from the form of inferences. Two experiments implemented manipulations that selectively influenced the model parameters for the knowledge-based information, the relative weight given to form-based versus knowledge-based information, and the parameters for the form-based information, validating the psychological interpretation of these parameters. We apply the model to classical suppression effects dissecting them into effects on background knowledge and effects on form-based processes (Exp. 3) and we use it to reanalyse previous studies manipulating reasoning instructions. In a model-comparison exercise, based on data of seven studies, the dual-source model outperformed three Bayesian competitor models. Overall, our results support the view that people make use of background knowledge in line with current Bayesian models, but they also suggest that the form of the conditional argument, irrespective of its content, plays a substantive, yet smaller, role.

Concerns with the SDT approach to causal conditional reasoning: a comment on Trippas, Handley, Verde, Roser, McNair, and Evans (2014).

Signal Detection Theory (SDT; Wickens, 2002) is a prominent measurement model that characterizes observed classification responses in terms of discriminability and response bias. In recent years, SDT has been increasingly applied within the psychology of reasoning (Rotello and Heit, 2009; Dube et al., 2010; Heit and Rotello, 2010, 2014; Trippas et al., 2013). SDT assumes that different stimulus types (e.g., valid and invalid syllogisms) are associated with different (presumably Gaussian) evidence or argument-strength distributions. Responses (e.g., “Valid” and “Invalid”) are produced by comparing the argument-strength of each syllogism with a set of established response criteria (Figure ​(Figure1A).1A). The response profile associated to each stimulus type can be represented as a Receiver Operating Charateristics (ROC) function by plotting performance pairs (i.e., hits and false-alarms) along different response criteria, which Gaussian SDT predicts to be curvilinear (Figure ​(Figure1B1B). Figure 1 (A) A graphical representation of the SDT model for a syllogistic reasoning task. (B) ROC curve representing the cumulative probabilities for hypothetical pairs of hits and false-alarms (“valid” responses to valid and invalid syllogisms, ... Trippas et al. (2014; henceforth THVRME) applied SDT to causal-conditional reasoning and make two points: (1) that SDT provides an informative characterization of data from a reasoning experiment with two orthogonal factors such as believability and argument validity; (2) that an inspection of the shape of causal-conditional ROCs provides insights on the suitability of normative theories with the consequence to consider affirmation and denial problems separately. The goal of this comment is to make two counterarguments: First, to point out that the SDT model is often unable to provide an informative characterization of data in designs as discussed by THVRME as it fails to unambiguously separate argument strength and response bias. THVRMEs conclusion that “believability had no effect on accuracy […] but seemed to affect response bias” (p. 4) solely hinge on arbitrary assumptions. Second, that THVRMEs reliance on ROC shape to justify a separation between affirmation and denial problems is unnecessary and misguided.

Turn around to have a look? Spatial referencing in dorsal vs. frontal settings in cross-linguistic comparison

When referring to an object in relation to another, speakers of many languages can adopt a relative frame of reference (FoR). Following Levinson (2003), this kind of FoR can be established by projecting an observers perspective onto the ground object either by translation, reflection, or rotation. So far, research on spatial FoRs has largely ignored the extent of variation in which of these projections are preferred generally, and specifically what kind of FoR is established for spatial arrays in ones back. This may seem justified by assumptions on “natural” preferences: for reflection in frontal settings (Canonical Encounter Hypothesis), and for converting dorsal into frontal situations by a turn of the observer before a reference is made (Turn Hypothesis). We scrutinize these assumptions by comparing the FoRs adopted for small-scale, static spatial arrays by speakers of four languages (German, US-English, Mandarin Chinese, and Tongan). Addressing the problem of inherent ambiguities on the item level when assessing FoRs from spatial prepositions, we use a multinomial processing tree (MPT) model for estimating probabilities of referencing strategies across sets of items. Substantial differences in frontal settings, both between and within languages, disprove the Canonical Encounter Hypothesis—translation occurs as frequently as reflection across samples. In dorsal settings, in contrast, the same type of response dominates in all samples. We suggest that this response is produced by a backward projection of the observers coordinate system in correspondence with the two main FoR preferences for frontal settings. However, none of these strategies involves a turn of the observer, thus also disproving the Turn Hypothesis. In conclusion, we discuss possible causes of the observed variability, explore links between the domains of space and time, and reflect the relation between language, communication, and culture.

Does Global and Local Vision Have an Impact on Creative and Analytic Thought? Two Failed Replications.

According to GLOMOsys (the GLObal versus LOcal processing MOdel, a systems account), an important distinction is that between a local and a global processing system: The former processes information in parts, the latter processes it globally. These systems can be activated by perceptual processing and carry over to subsequent conceptual processing, in particular to analytical and creative thought. A conceptual and a high-powered close replication of previously reported studies test predictions of GLOMOsys for analytical thought and for analytical and creative thought, respectively. The present studies found no evidence that processing style primed via the Navon letter task has an impact on creative or analytic thought.

On the measurement of criterion noise in signal detection theory: reply to Benjamin (2013).

Kellen, Klauer, and Singmann (2012) questioned whether possible criterion noise would contribute significantly to modeling recognition memory. Our arguments were based on a reanalysis of the data by Benjamin, Diaz, and Wee (2009) as well as on new experimental data. In a comment, Benjamin (2013) questioned some of Kellen et al.s conclusions and raised important issues regarding the new experimental data. In this reply, we revisit our arguments and provide new analyses in response to Benjamins questions and issues.