Hypothesis generation, sparse categories, and the positive test strategy.

Navarro, Daniel J.; Perfors, Amy

doi:10.1037/a0021110

Cited by 74 publications

(102 citation statements)

References 112 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thus, uncertainty sampling may be a general-purpose strategy that is less demanding but often consistent with normative principles. A similar point has been made in recent analyses showing that positive testing can be equivalent to information gain under certain conditions (Austerweil & Griffiths, 2011;Klayman & Ha, 1987;Navarro & Perfors, 2011). For example, Austerweil and Griffiths (2011) showed that when hypotheses make deterministic predictions about the next event in a sequence, testing the prediction of the most probable hypothesis (e.g., asking "is the next event A?"…”

Section: Relation To Other Modeling Approachesmentioning

confidence: 80%

“…For example, for certain kinds of hypothesis spaces, confirmatory sampling is consistent with normative goals, precluding the need to consider more than one alternative (Austerweil & Griffiths, 2011;Klayman & Ha, 1987;Navarro & Perfors, 2011;Nelson & Movellan, 2001). Learners who correctly account for these constraints might be expected to show more or less confirmatory sampling depending on the problem space.…”

Section: An Intermediate Positionmentioning

confidence: 99%

See 1 more Smart Citation

Self‐Directed Learning Favors Local, Rather Than Global, Uncertainty

2015

View full text Add to dashboard Cite

Collecting (or "sampling") information that one expects to be useful is a powerful way to facilitate learning. However, relatively little is known about how people decide which information is worth sampling over the course of learning. We describe several alternative models of how people might decide to collect a piece of information inspired by "active learning" research in machine learning. We additionally provide a theoretical analysis demonstrating the situations under which these models are empirically distinguishable, and we report a novel empirical study that exploits these insights. Our model-based analysis of participants' information gathering decisions reveals that people prefer to select items which resolve uncertainty between two possibilities at a time rather than items that have high uncertainty across all relevant possibilities simultaneously. Rather than adhering to strictly normative or confirmatory conceptions of information search, people appear to prefer a "local" sampling strategy, which may reflect cognitive constraints on the process of information gathering.

show abstract

Section: Relation To Other Modeling Approachesmentioning

confidence: 80%

Section: An Intermediate Positionmentioning

confidence: 99%

Self‐Directed Learning Favors Local, Rather Than Global, Uncertainty

2015

View full text Add to dashboard Cite

show abstract

“…A diagnostic test would be one in which the likelihood of a positive result differs between competing hypotheses (e.g., an elevated white-blood-cell count might be highly likely for appendicitis, but relatively unlikely for cracked ribs), providing the potential for revising beliefs about relevant hypotheses. Although some recent work has argued that confirmation testing can be a reasonable strategy once the complexity of realworld domains is accounted for Navarro & Perfors, 2011), diagnostic search has been observed under conditions that facilitate the consideration of alternatives (Mynatt, Doherty, & Dragan, 1993).…”

Section: Hypothesis Generation Influences Hypothesis-testing Behaviormentioning

confidence: 99%

Memory Constraints on Hypothesis Generation and Decision Making

Thomas

Dougherty

Buttaccio

2014

Curr Dir Psychol Sci

View full text Add to dashboard Cite

Hypothesis generation is the process people use to generate explanations for patterns of data, which is an act vital to everyday problem solving. It is the basis for decision making in many professions, such as medicine, intelligence and reconnaissance analysis, auditing, and fault detection in nuclear power plants. Even laypeople's impressions of acquaintances' personalities based on behavioral patterns can be considered a case of hypothesis generation. This article provides an overview of research elucidating the cognitive processes that underlie hypothesis generation and decision making.

show abstract

“…Thus, the choice of an effective testing strategy in a situation is a question of ecological rationality in that it depends to a large degree on the structure of the environment Gigerenzer, Todd, & the ABC Research Group, 1999). Past work has shown that people behave in an ecologically rational fashion when sparsity is varied in a number of non-causal hypothesis testing tasks (where sparsity CAUSAL EXPERIMENTATION 8 applied to hypotheses or events, e.g., McKenzie, Chase, Todd, & Gigerenzer, 2012; Hendrickson, Navarro, & Perfors, 2016;Oaksford & Chater, 1994;Langsford, Hendrickson, Perfors, & Navarro, 2014;Navarro & Perfors, 2011). For example, Hendrickson et al (2016) showed that people switch from requesting positive to negative examples of a concept when the overall proportion of positive cases increases.…”

Section: Sparsity Determines Effectiveness Of Learning Strategiesmentioning

confidence: 99%

“…Here, too, the optimal strategy (in terms of expected information gain, see next section) is to ask questions targeting features that apply to half the possibilities under consideration (e.g., "Is the person female? ", if the hypotheses are people and half are each sex), because it reduces the number of alternatives more rapidly than asking about specific identities directly (e.g., Navarro & Perfors, 2011;Mosher & Hornsby, 1966).In analogy to the switch testing example presented above, inquiring about a feature means asking if any of the individuals sharing that feature is the target, just as turning on multiple switches asks if any of those switches has an effect on the outcome (the fan). An alternative approach to asking about shared features would be to test each person individually (e.g., "Is it Bob?").…”

mentioning

confidence: 99%

Testing one or multiple: How beliefs about sparsity affect causal experimentation

Coenen¹,

Ruggeri²,

Bramley³

et al. 2018

Preprint

View full text Add to dashboard Cite

What is the best way of discovering the underlying structure of a causal system composed of multiple variables? One prominent idea is that learners should manipulate each candidate variable in isolation to avoid confounds (sometimes known as the "Control of Variables" or CV strategy). We demonstrate that CV is not always the most efficient method for learning. Using an optimal actor model which aims to minimize the average number of tests, we show that when a causal system is sparse (i.e., when the outcome of interest has few or even just one actual cause among the candidate variables) it is more efficient to test multiple variables at once. Across a series of behavioral experiments, we then show that people are sensitive to causal sparsity and adapt their strategies accordingly. When interacting with a non-sparse causal system (high proportion of actual causes among candidate variables), they use a CV strategy, changing one variable at a time. When interacting with a sparse causal system they are more likely to test multiple variables at once. However, we also find that people sometimes use a CV strategy even when a system is sparse. IntroductionTo develop a causal understanding of the world, we often need to find out how multiple candidate variables affect an outcome of interest. This problem arises in everyday situations (e.g., "Which switch(es) control the bathroom fan?"), during scientific exploration ("Which of these treatments can affect disease x?"), and plays an important part in answering economic and social questions ("What is the impact of these policies on Gross Domestic Product?"). Often, the quickest and most effective method of resolving causal relationships is to conduct experiments that manipulate variables of a system (e.g., turning switches on or off) and to observe the resulting outcome. This kind of causal experimentation is often (but not always) required to decouple causation and correlation (Pearl, 2009;Woodward, 2005;Sloman, 2005).Both children and adults can systematically leverage the outcomes of interventions to test causal hypotheses that would be indistinguishable based on observation alone (Lagnado & Sloman, 2004; Lagnado, Waldmann, Hagmayer, & Sloman, 2006;Rottman & Keil, 2012;Schulz, Gopnik, & Glymour, 2007;Sloman & Lagnado, 2005;Waldmann & Hagmayer, 2005). Furthermore, people are sometimes able to come up with highly efficient experiments that optimize information gained per intervention or minimize the total number of tests needed on average to discover the true causal structure (Bramley, Lagnado, & Speekenbrink, 2015;Bramley, Dayan, Griffiths, & Lagnado, 2017; Coenen, Rehder, & Gureckis, 2015;Steyvers, Tenenbaum, Wagenmakers, & Blum, 2003).Here we consider a specific learning situation in which people are asked to explore a causal system consisting of a number of independent variables (switches) and a dependent outcome (turning on a fan). These types of problems have played a central role in research on science education and cognitive development and are common in ...

show abstract

Hypothesis generation, sparse categories, and the positive test strategy.

Cited by 74 publications

References 112 publications

Self‐Directed Learning Favors Local, Rather Than Global, Uncertainty

Self‐Directed Learning Favors Local, Rather Than Global, Uncertainty

Memory Constraints on Hypothesis Generation and Decision Making

Testing one or multiple: How beliefs about sparsity affect causal experimentation

Contact Info

Product

Resources

About