Uncovering Mental Architecture and Related Mechanisms in Elementary Human Perception, Cognition, and Action

Abstract

This chapter is a tutorial on and review of a theory‐driven, quantitative approach to studying human information processing systems. Key properties in such systems comprise mental architecture, workload capacity, decisional stopping rules, and several varieties of independence, among others. We outline the early history of such interests, which began in the 19th century and then were reopened in the 1960s. We point out the hazards due to the frequent ability of even mathematically specified models to mimic each other's experimental predictions. Next, we show how the deepest properties of distinct psychological principles can, through what we term “meta‐theory,” be engaged to prove powerful theorems regarding model distinctions and transferred to incisive experimental designs that avoid the dilemma of model mimicry. This meta‐theory and its associated experimental designs have been implemented to address a wide variety of theoretical and empirical questions regarding attention, perception, psychophysics, memory processes, decision making, and categorization. This range of implementations—running from basic science to applied arenas such as clinical pathologies and human factors research—underscores the generality and potency of this approach.

Decomposing the composite face effect: Evidence for non-holistic processing based on the ex-Gaussian distribution

“…The emergence of a composite face effect does not necessarily imply that the information from the two face halves has been integrated. The effects can occur due to differences in relative discriminability between the top and bottom parts (Fitousi, 2016; Fitousi & Algom, 2006; Melara & Mounts, 1993), or due to decisional factors across conditions (Fitousi, 2016; Richler et al, 2011; Von Der Heide et al, 2018), which are only remotely related to holistic processing (Townsend et al, 2018). Second, when tested in related paradigms or against computational models of holism, the integration (holistic) hypothesis is not well supported (Bradshaw & Wallace, 1971; Donnelly et al, 2012; Ellison & Massaro, 1997; Fitousi, 2015, 2016; Gold et al, 2012; Loftus et al, 2004; Macho & Leder, 1998; Sergent, 1984; Tversky & Krantz, 1969; Wenger & Ingvalson, 2002, 2003).…”

Interactive Parallel Models: No Virginia, Violation of Miller's Race Inequality does not Imply Coactivation and Yes Virginia, Context Invariance is Testable

“…The Quantitative Methods for Psychology cific models, in the spirit of theoretical work by Dzhafarov (1993), Smith and Van Zandt (2000), and Townsend, Wenger, and Houpt (2018). In fact, to elaborate a bit further, our focus will be on the analysis of interactive parallel channels and comparison with stochastically independent parallel channels which do not vary in efficiency as workload is increased.…”

Extending Systems Factorial Technology to errored responses

“…The second approach focuses on developing theoretically-motivated methods for deriving qualitatively distinct predictions from large classes of information processing models. One meta-theoretic approach, Systems Factorial Technology (SFT, Townsend & Nozawa, 1995;Townsend, Wenger, & Houpt, 2018), has systematically built upon and generalized many of the key historical theories including, among others, the method of subtraction (Donders, 1868), additive factors methods (Sternberg, 1966(Sternberg, , 1969Schweickert, 1978;Townsend, 1971Townsend, , 1984, redundant target methods (Raab, 1962;Miller, 1982), and trichonometric theory . SFT has been reviewed in several recent tutorials (Algom & Fitousi, 2016;Altieri, Fifić, Little, & Yang, 2017;Houpt, Blaha, McIntire, Havig, & Townsend, 2013;Harding et al, 2016) and in the recent volume by Little, Altieri, Fifić, and Yang (2017) and a recent special issue of the Journal of Mathematical Psychology.…”

“…Broadly speaking, there are three major threads to SFT, accompanied by appropriate diagnostics, the Capacity Coefficient, which informs researchers about the system's ability to cope with increased load (typically an increase in the number of signals presented for processing), the Resilience Function and Conflict Contrast Function, which assesses how functions cope with distracting or conflicting information (Little, Eidels, Fific, & Wang, 2015, 2018Houpt & Little, 2017), and the Survivor Interaction Contrast (SIC), which informs the architecture and stopping rule. The theory underlying the measurement of workload capacity has already been extended to error RTs (Townsend & Altieri, 2012;Donkin, Little, & Houpt, 2014) and the form of the resilience function is highly similar to the capacity coefficient.…”