Comparative Perception and Attention Laboratory

TrI Toolbox



TrI Toolbox is a free software that derives predictions of reinforcement-based models of transitive inference (specifically, Wynne's and Siemann-Delius' models; Siemann & Delius, 1998; Wynne, 1995) for the n-term transitive inference task of any length on the basis of the data uploaded by a user. N-term transitive inference task is an increasingly popular behavioral task that presumably tests ability to reason deductively. In this task, a subject is presented with 4 (or more) overlapping pairs of stimuli: A+ B-, B+ C-, C+ D-, and D+ E- (where A, B, C, etc. indicate the stimuli, pluses indicate reinforcement, and minuses indicate non-reinforcement). During the test, a subject is given the critical testing pair is the pair BD. The ability to select a stimulus B in the pair BD is often interpreted as indicative of transitive inference (if B > C, and C > D, then B > D). 

Unfortunately, reinforcement-based models of transitive inference provide an alternative explanation to the choice of B over D in the BD test. According to those models, during standard TI training the associative values of the training stimuli begin to form an ordered series of associative values: A > B > C > D > E. Thus, at the end of the training a subject could choose stimulus B over stimulus D in the critical pair BD based on a direct comparison of their relative associative values instead of using inferential-like processes. Therefore, evaluating predictions of reinforcement-based models in non-verbal transitive inference (TI) tasks is critical for separating influences of reinforcement history from inferential processes. Due to their complexity, these models must be evaluated by performing iterative computations for several free-fitting parameters using subjects’ responses during the experiment. TrI Toolbox is the first software capable of performing such evaluations.

​TrI Toolbox is implemented as a MATLAB toolbox which makes it platform independent (Windows, Mac, Linux, etc.) if you have a recent version of MATLAB (R2011a or later). Running TrI Toolbox in MATLAB also provides an opportunity to customize the code for your project needs. A compiled version of TrI Toolbox that does not require MATLAB is currently available for Windows 7 and 8. Please contact us if you are interested in obtaining a compiled version for a different platform, and we'll try to help if we can.


Download:

Version

Requirements

Size

Instructions

User manual, ver. 1.0

Adobe PDF Reader or equivalent

800 KB​


TrI Toolbox 1.0 (MATLAB toolbox, zip archive)

All OS, requires MATLAB R2011a or later

15 KB

readme_toolbox.txt

TrI Toolbox 1.0 (full standalone executable with MCR, zip archive)

Windows 7 or 8; select if you do not currently have MCR 7.15 installed

190 MB

readme_standalone.txt

TrI Toolbox 1.0 (standalone executable without MCR, zip archive)

Windows 7 or 8; select if MCR 7.15 is already installed

120 KB

readme_standalone_noMCR.txt


Note: If you successfully installed MCR but are unable to run triToolbox.exe, then make sure that the file triToolbox.mat is located in the same folder as the executable.


Terms of use:


TrI Toolbox has been developed by David Goodman and Olga Lazareva of Psychology Department at Drake University (copyright 2014). This program is free software: you can redistribute it and/or modify it under the terms of the MIT license. The text of license is available in license.txt file.