Applying Occam’s razor in modeling cognition: A Bayesian approach

Myung, In Jae; Pitt, Mark A.

doi:10.3758/bf03210778

Cited by 372 publications

(433 citation statements)

References 64 publications

(76 reference statements)

Supporting

Mentioning

429

Contrasting

Unclassified

Order By: Relevance

“…As noted previously, GCM is more flexible than GRT for this design, and flexibility has become a topical issue in model selection (e.g., Kass & Raftery, 1995;Myung & Pitt, 1997;Pitt, Myung, & Zhang, 2003). Although this difference in flexibility makes model selection complicated in cases in which both GRT and GCM qualitatively account for the data, it is of little concern for cases in which one model clearly fails.…”

Section: Scale Construction and Model-based Analysesmentioning

confidence: 93%

Comparing categorization models.

Rouder¹,

Ratcliff²

2004

Journal of Experimental Psychology: General

View full text Add to dashboard Cite

Four experiments are presented that competitively test rule-and exemplar-based models of human categorization behavior. Participants classified stimuli that varied on a unidimensional axis into 2 categories. The stimuli did not consistently belong to a category; instead, they were probabilistically assigned. By manipulating these assignment probabilities, it was possible to produce stimuli for which exemplar-and rule-based explanations made qualitatively different predictions.F. G. Ashby and J. T. Townsend's (1986) rule-based general recognition theory provided a better account of the data than R. M. Nosofsky's (1986) exemplar-based generalized context model in conditions in which the to-be-classified stimuli were relatively confusable. However, generalized context model provided a better account when the stimuli were relatively few and distinct. These findings are consistent with multiple process accounts of categorization and demonstrate that stimulus confusion is a determining factor as to which process mediates categorization.In this article we present an empirical paradigm to test different theories of categorization behavior. One theory we test is the exemplar-based theory in which categories are represented by sets of stored exemplars. Category membership of a stimulus is determined by similarity of the stimulus to these exemplars (e.g., Medin & Schaffer, 1978;Nosofsky, 1986Nosofsky, , 1987Nosofsky, , 1991). An exemplar-based process relies on retrieval of specific trace-based information without further abstraction; for example, a person is judged as "tall" if he or she is similar in height to others who are considered "tall." The other theory we test is rule-based or decisionbound theory. Decisions are based on an abstracted rule. The relevant space is segmented into regions by bounds, and each region is assigned to a specific category (e.g., Ashby & Gott, 1988;Ashby & Maddox, 1992Ashby & Perrin, 1988;Ashby & Townsend, 1986;Trabasso & Bower, 1968). For example, a person might be considered tall if he or she is perceived as being over 6 ft (1.83 m). The essence of a rule-based process is that processing is based on greatly simplified abstractions or rules but not on the specific trace-based information itself.Both rule-and exemplar-based theories have gained a large degree of support in the experimental literature. There are both exemplar-based models (e.g., generalized context model; Nosofsky, 1986) and rule-based models (e.g., general recognition theory; Ashby & Gott, 1988;Ashby & Townsend, 1986) that can explain a wide array of behavioral data across several domains. Despite the many attempts to discriminate between these two explanations, there have been few decisive tests. Across several paradigms and domains, rule-and exemplarbased predictions often mimic each other.

show abstract

Section: Scale Construction and Model-based Analysesmentioning

confidence: 93%

Comparing categorization models.

Rouder¹,

Ratcliff²

2004

Journal of Experimental Psychology: General

View full text Add to dashboard Cite

show abstract

“…A hypothesis test based on the Bayes factor supports the model under which the observed data are most likely (for details see Berger & Pericchi, 1996;Bernardo & Smith, 1994, chap. 6;Klugkist, Laudy, & Hoijtink, 2002, cha 7;Klugkist et al, 2005a;Kass & Raftery, 1995;MacKay, 2003;Myung & Pitt, 1997;O'Hagan, 1995). Therefore, the Bayes factor represents ''the standard Bayesian solution to the hypothesis testing and model selection problems" (Lewis & Raftery, 1997, p. 648); in the following, we will use ''Bayesian hypothesis test" as a shortcut for ''a hypothesis test based on the Bayes factor".…”

Section: Bayesian Hypothesis Testingmentioning

confidence: 99%

“…This intuition is consistent with the law of parsimony or ''Ockham's razor" which states that, when everything else is equal, simple models are to be preferred over complex models (Jaynes, 2003, chap. 20;Myung & Pitt, 1997). Formal model selection methods try to quantify the tradeoff between goodness-of-fit and parsimony.…”

Section: Additional Advantages Of Bayesian Hypothesis Testingmentioning

confidence: 99%

Bayesian hypothesis testing for psychologists: A tutorial on the Savage–Dickey method

Wagenmakers

Lodewyckx

Kuriyal

et al. 2010

Cognitive Psychology

607

616

View full text Add to dashboard Cite

a b s t r a c tIn the field of cognitive psychology, the p-value hypothesis test has established a stranglehold on statistical reporting. This is unfortunate, as the p-value provides at best a rough estimate of the evidence that the data provide for the presence of an experimental effect. An alternative and arguably more appropriate measure of evidence is conveyed by a Bayesian hypothesis test, which prefers the model with the highest average likelihood. One of the main problems with this Bayesian hypothesis test, however, is that it often requires relatively sophisticated numerical methods for its computation. Here we draw attention to the Savage-Dickey density ratio method, a method that can be used to compute the result of a Bayesian hypothesis test for nested models and under certain plausible restrictions on the parameter priors. Practical examples demonstrate the method's validity, generality, and flexibility.

show abstract

“…One form of GMA that we have studied in prior work is model complexity (Myung, Balasubramanian & Pitt, 2000;Myung & Pitt, 1997;Pitt, Myung & Zhang, 2002). It is concerned with assessing the inherent flexibility of a model in fitting data.…”

Section: Landscaping: a Global Model Analysismentioning

confidence: 99%

“…It is well-known that, because they consider the behavior of the model across its entire parameter space (i.e., by adopting a global model analysis), Bayesian approaches are able to identify and compensate for model complexity (Myung & Pitt, 1997). Notice, however, that the landscaping approach also specifies a prior distribution p(θ | M), from which the parameters are sampled, and a likelihood function p(D | M, θ) from which the data are generated.…”

Section: B2 Landscaping and Bayesian Marginalsmentioning

confidence: 99%