Dynamic models of choice

Heathcote, Andrew; Lin, Yi-Shin; Reynolds, Angus; Strickland, Luke; Gretton, Matthew; Matzke, Dora

doi:10.3758/s13428-018-1067-y

Cited by 165 publications

(220 citation statements)

References 63 publications

Supporting

Mentioning

206

Contrasting

Order By: Relevance

“…We used the Dynamic Models of Choice (DMC; Heathcote et al, 2018) software implemented in R (R Core Team, 2018) to estimate model parameters via Bayesian hierarchical modelling. In hierarchical modelling, the population-level mean and standard deviation parameters characterise the population-level distribution for each model parameter.…”

Section: Modelling Of Response Inhibition the Ex-gaussian Distributimentioning

confidence: 99%

“…Weakly informative uniform priors were set for the population-level parameters, which are identical to those of Skippen et al (2019). Posterior distributions of the parameters were obtained using Differential Evolution Markov Chain Monte Carlo (MCMC) sampling (Ter Braak, 2006), with steps closely mimicking Heathcote et al (2018).…”

Section: Modelling Of Response Inhibition the Ex-gaussian Distributimentioning

confidence: 99%

See 1 more Smart Citation

Reconsidering electrophysiological markers of response inhibition in light of trigger failures in the stop-signal task

Skippen

Fulham

Michie

et al. 2019

Preprint

View full text Add to dashboard Cite

Although advancements in electrophysiological methods to explore response inhibition have been substantial, the methods to describe behavioural differences in response inhibition have remained relatively unchanged. Here we use a model-based neuroscience approach to understand the neural correlates underpinning response inhibition as estimated using a recently developed ex-Gaussian hierarchical Bayesian model of stop-signal task performance. In a large healthy sample (N=156) of community drawn participants, we show the model-based estimates of stop-signal reaction time (SSRT) and a "trigger failure" parameter reflecting lapses of attention to task goals alter previously held interpretations of relationships between inhibition and event-related potential (ERP) component measures.Our results show clear attenuation of SSRT by ≈65ms when substantial levels of trigger failure are accounted for. This attenuation casts doubt on previous interpretations of the P3 as a manifestation of response inhibition. Instead, the N1, which reflects attentional processes, provides a better description of both SSRT and trigger failure than the P3. In particular, the peak N1 latency both correlated and coincided with ex-Gaussian estimated SSRT. Furthermore, participants with higher rates of trigger failure do not show a dissociation between N1 latency and the outcome of a stop trial. Our results show sufficient attentional control to elicit an inhibitory process is just as important, if not more so, than the speed of the process itself and that early ERPs provide a rich account of individual differences in both the speed and reliability of the inhibitory process.

show abstract

Section: Modelling Of Response Inhibition the Ex-gaussian Distributimentioning

confidence: 99%

Section: Modelling Of Response Inhibition the Ex-gaussian Distributimentioning

confidence: 99%

Reconsidering electrophysiological markers of response inhibition in light of trigger failures in the stop-signal task

Skippen

Fulham

Michie

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…As many evidence-accumulation models have analytic likelihoods, and so are amenable to MCMC methods for obtaining posterior distributions, Warp-III sampling is not limited to the LBA, but may be readily applied to other models, such as the diffusion decision model (DDM; Ratcliff, 1978;Ratcliff & McKoon, 2008). Heathcote et al's (2018) DMC software enables the hierarchical MCMC-based estimation of not only the LBA and the DDM, but also a variety of other models including single-boundary and racing diffusion models (Leite & Ratcliff, 2010;Logan, Van Zandt, Verbruggen, & Wagenmakers, 2014), lognormal race models (Heathcote & Love, 2012;Rouder, Province, Morey, Gómez, & Heathcote, 2015), as well as race models of the stop-signal paradigm (Matzke et al, 2013;Matzke, Love, & Heathcote, 2017). Our easyto-use R-implementation of the Warp-III sampler enables the computation of the marginal likelihood of any model implemented in the DMC software.…”

Section: Discussionmentioning

confidence: 99%

“…The priors in Eq. 8 were taken from Heathcote et al (2018). Although we believe that these priors provide a reasonable setup based on our experience with the LBA parameter ranges, they may be replaced by empirically informed priors in future applications.…”

Section: Prior Distributionsmentioning

confidence: 99%

Computing Bayes Factors for Evidence-Accumulation Models Using Warp-III Bridge Sampling

Gronau¹,

Heathcote²,

Matzke³

2018

Preprint

Self Cite

View full text Add to dashboard Cite

Over the last decade, the Bayesian estimation of evidence-accumulation models has gained popularity, largely due to the advantages afforded by the Bayesian hierarchical framework. Despite recent advances in the Bayesian estimation of evidence-accumulation models, model comparison continues to rely on suboptimal procedures, such as posterior parameter inference and model selection criteria known to favor overly complex models. In this paper, we advocate model comparison for evidence-accumulation models based on the Bayes factor obtained via Warp-III bridge sampling. We demonstrate, using the linear ballistic accumulator (LBA), that Warp-III sampling provides a powerful and flexible approach that can be applied to both nested and non-nested model comparisons, even in complex and high-dimensional hierarchical instantiations of the LBA. We provide an easy-to-use software implementation of the Warp-III sampler and outline a series of recommendations aimed at facilitating the use of Warp-III sampling in practical applications.

show abstract

“…We found that TIDE provided an approximation of the marginal likelihood that closely matched TI for models with a single subjects. However, when extending the models hierarchically, we found that certain assumptions about the dependence between the individual and group level parameter samples resulted in large differences in the TI approximated marginal likelihood, where the standard dependent sampling results in higher marginal likelihoods than the recently implemented (e.g., Heathcote et al, 2018; implemented within their DMC package) independent sampling. We extended TIDE to these two different situations, with dependent sampling only requiring a natural extension of TIDE, and independent sampling adding the use of past iterations in a manner similar to "Z updating" from an extension of DE-MCMC, DE-MCz (ter Braak & Vrugt, 2008).…”

Section: Introductionmentioning

confidence: 90%

Thermodynamic integration via differential evolution: A method for approximating marginal likelihoods

Evans¹,

Annis²

2019

Preprint

View full text Add to dashboard Cite

A typical goal in cognitive psychology is to select the model that provides the best explanation of the observed behavioral data. The Bayes factor provides a principled approach for making these selections, though the integral required to calculate the marginal likelihood for each model is intractable for most cognitive models. In these cases, Monte Carlo techniques must be used to approximate the marginal likelihood, such as thermodynamic integration (TI; Friel & Pettitt, 2008; Lartillot & Philippe, 2006), which relies on sampling from the posterior at different powers (called power posteriors). TI can become computationally expensive when using population Markov chain Monte Carlo (MCMC) approaches such as differential evolution MCMC (DE-MCMC; Turner, Sederberg, Brown, & Steyvers, 2013) that require several interacting chains per power posterior. Here, we propose a method called thermodynamic integration via differential evolution (TIDE), which aims to reduce the computational burden associated with TI by using a single chain per power posterior. We show that when applied to non-hierarchical models, TIDE produces an approximation of the marginal likelihood that closely matches TI. When extended to hierarchical models, we find that certain assumptions about the dependence between the individual and group level parameters samples (i.e., dependent/independent) have sizable effects on the TI approximated marginal likelihood. We propose two possible extensions of TIDE to hierarchical models, which closely match the marginal likelihoods obtained through TI with dependent/independent sampling in many, but not all, situations. Based on these findings, we believe that TIDE provides a promising method for estimating marginal likelihoods, though future research should focus on a detailed comparison between the methods of approximating marginal likelihoods for cognitive models.

show abstract

Dynamic models of choice

Cited by 165 publications

References 63 publications

Reconsidering electrophysiological markers of response inhibition in light of trigger failures in the stop-signal task

Reconsidering electrophysiological markers of response inhibition in light of trigger failures in the stop-signal task

Computing Bayes Factors for Evidence-Accumulation Models Using Warp-III Bridge Sampling

Thermodynamic integration via differential evolution: A method for approximating marginal likelihoods

Contact Info

Product

Resources

About