Methods of comparing associative models and an application to retrospective revaluation

Witnauer, James E.; Hutchings, Ryan; Miller, Ralph R.

doi:10.1016/j.beproc.2017.08.004

Cited by 6 publications

(5 citation statements)

References 62 publications

(109 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…When different versions of the ADM were compared, they were fit independently, and residuals were converted into a model selection metric, the corrected Akaike Information Criterion (AICc). Models were selected if they had the fewest number of free parameters among those with a ΔAICc <4 (per recommendations of Avila et al, 2009; Burnham, Anderson, & Huyvaert, 2011; Posada & Buckley, 2004; and Witnauer, Hutchings, & Miller, 2017). This ensured that the selected model provided the best balance between parsimony (number of free parameters) and fit (in this case, minimized residuals)…”

Section: Testing the Adm Of Paradoxical Choicementioning

confidence: 99%

See 1 more Smart Citation

An associability decay model of paradoxical choice.

Daniels

Sanabria²

2018

Journal of Experimental Psychology: Animal Learning and Cogniti

View full text Add to dashboard Cite

Paradoxical choices in human and nonhuman animals represent substantial deviations from rational models of behavior; such deviations often demand models that incorporate multiple perspectives, including psychology, biology, and economics. The past couple of decades have seen an increased interest in the paradoxical choice of pigeons in 2-armed bandit tasks (2ABT) developed by Zentall and colleagues. In these 2ABTs, pigeons, but not rats, systematically choose an alternative that yields less reward over multiple trials but provides more information on events within a trial, over an alternative that yields more reward over trials but provides less information on events within a trial. Although current computational models account for much of the extant data generated in studies on 2ABT choice, they do not explain, in a trial-by-trial basis, how pigeons learn to ignore some stimuli and not others in 2ABTs. To address the provenance of this differential allocation of attention, a simple model composed of Bush-Mosteller linear operators and a Pearce-Hall-like associability mechanism is proposed. This model, referred to as the Associability Decay Model (ADM) of paradoxical choice, adequately accounts for the performance of pigeons and rats in 2ABTs. The ADM yields an untested prediction that is inconsistent with other computational models of 2ABT performance, and offers potential explanations for why differences in motivation, social enrichment, and impulsivity alter the degree to which pigeons systematically choose information despite earning fewer rewards. The success of the ADM shows that a relatively simple dynamic trial-by-trial model can account for much of the extant paradoxical-choice data while identifying opportunities for further study and refinement of models of 2ABT performance. (PsycINFO Database Record

show abstract

Section: Testing the Adm Of Paradoxical Choicementioning

confidence: 99%

“…ASSOCIABILITY DECAY IN PARADOXICAL CHOICE al., 2009;Burnham, Anderson, & Huyvaert, 2011;Posada & Buckley, 2004;and Witnauer, Hutchings, & Miller, 2017). This ensured that the selected model provided the best balance between parsimony (number of free parameters) and fit (in this case, minimized residuals).…”

mentioning

confidence: 99%

An associability decay model of paradoxical choice.

Daniels

Sanabria²

2018

Journal of Experimental Psychology: Animal Learning and Cogniti

View full text Add to dashboard Cite

show abstract

“…Having more parameters clearly gives an intrinsic flexibility in data fitting that more rigid theories do not have, and one should balance this factor against the actual capabilities of the new models to accommodate observations. This is the typical situation where comparative statistics, such as the Bayesian (or Schwarz) Information Criterion (BIC) and the Akaike Information Criterion (AIC) can be extremely useful, as advocated by Witnauer et al (2017).…”

Section: Data Analysis: Comparison With the Rw Modelmentioning

confidence: 99%

Behavior Stability and Individual Differences in Pavlovian Extended Conditioning

2020

View full text Add to dashboard Cite

How stable and general is behavior once maximum learning is reached? To answer this question and understand post-acquisition behavior and its related individual differences, we propose a psychological principle that naturally extends associative models of Pavlovian conditioning to a dynamical oscillatory model where subjects have a greater memory capacity than usually postulated, but with greater forecast uncertainty. This results in a greater resistance to learning in the first few sessions followed by an over-optimal response peak and a sequence of progressively damped response oscillations. We detected the first peak and trough of the new learning curve in our data, but their dispersion was too large to also check the presence of oscillations with smaller amplitude. We ran an unusually long experiment with 32 rats over 3,960 trials, where we excluded habituation and other well-known phenomena as sources of variability in the subjects' performance. Using the data of this and another Pavlovian experiment by Harris et al. (2015), as an illustration of the principle we tested the theory against the basic associative single-cue Rescorla-Wagner (RW) model. We found evidence that the RW model is the best non-linear regression to data only for a minority of the subjects, while its dynamical extension can explain the almost totality of data with strong to very strong evidence. Finally, an analysis of short-scale fluctuations of individual responses showed that they are described by random white noise, in contrast with the colored-noise findings in human performance.

show abstract

“…All of the examples in the Journal of the Experimental Analysis of Behavior, Journal of Experimental Psychology: Animal Behavior Processes, and Journal of Experimental Psychology: Animal Cognition and Learning were examined, as were selected papers in the other journals. Some variant of I-T analysis was used to examine models of choice or response strength (Cowie, Davison, & Elliffe, 2014;Hunter & Davison, 1982;Hutsell & Jacobs, 2013;Lau & Glimcher, 2005;MacDonall, 2009;McLean, Grace, & Nevin, 2012;Mitchell, Wilson, & Karalunas, 2015;Navakatikyan, 2007), discounting (Białaszek, Marcowski, & Ostaszewski, 2017;DeHart & Odum, 2015;Jarmolowicz et al, 2018;Rung & Young, 2015;Young, 2017), peak shift (Krägeloh, Elliffe, & Davison, 2006), behavioral momentum (Hall, Smith, & Wynne, 2015;Killeen & Nevin, 2018), associative learning (Cabrera, Sanabria, Shelley, & Killeen, 2009;Hall et al, 2015;Witnauer, Hutchings, & Miller, 2017), the partitioning of behavior into response bouts (Brackney, Cheung, Neisewander, & Sanabria, 2011;Smith, McLean, Shull, Hughes, & Pitts, 2014;Tanno, 2016), navigation strategies (Anselme, Otto, & Güntürkün, 2018), instructional techniques (Warnakulasooriya, Palazzo, & Pritchard, 2007), problem solving by corvids (Cibulski, Wascher, Weiss, & Kotrschal, 2014), models of punishment (Klapes, Riley, & McDowell, 2018), and timing (Beckmann & Young, 2009;Ludvig, Conover, & Shizgal, 2007). This is just a partial, not exhaustive, list to convey the range of applications.…”

Section: Use In Behavior Sciencementioning

confidence: 99%

An Information Theoretic Approach to Model Selection: A Tutorial with Monte Carlo Confirmation

Newland

2019

Perspect Behav Sci

View full text Add to dashboard Cite

A reliance on null hypothesis significance testing (NHST) and misinterpretations of its results are thought to contribute to the replication crisis while impeding the development of a cumulative science. One solution is a data-analytic approach called Information-Theoretic (I-T) Model Selection, which builds upon Maximum Likelihood estimates. In the IT approach, the scientist examines a set of candidate models and determines for each one the probability that it is the closer to the truth than all others in the set. Although the theoretical development is subtle, the implementation of IT analysis is straightforward. Models are sorted according to the probability that they are the best in light of the data collected. It encourages the examination of multiple models, something investigators desire and that NHST discourages. This article is structured to address two objectives. The first is to illustrate the application of IT data analysis to data from a virtual experiment. A noisy delay-discounting data set is generated and seven quantitative models are examined. In the illustration, it is demonstrated that it is not necessary to know the "truth" is to identify the one that is closest to it and that the most likely models conform to the model that generated the data. Second, we examine claims made by advocates of the IT approach using Monte Carlo simulations in which 10,000 different data sets are generated and analyzed. The simulations showed that 1) the probabilities associated with each model returned by the single virtual experiment approximated those that resulted from the simulations, 2) models that were deemed close to the truth produced the most precise parameter estimates, and 3) adding a single replicate sharpens the ability to identify the most probable model.

show abstract

Methods of comparing associative models and an application to retrospective revaluation

Cited by 6 publications

References 62 publications

An associability decay model of paradoxical choice.

An associability decay model of paradoxical choice.

Behavior Stability and Individual Differences in Pavlovian Extended Conditioning

An Information Theoretic Approach to Model Selection: A Tutorial with Monte Carlo Confirmation

Contact Info

Product

Resources

About