Learning in a changing environment.

Abstract

Multiple cue probability learning studies have typically focused on stationary environments. We present three experiments investigating learning in changing environments. A fine-grained analysis of the learning dynamics shows that participants were responsive to both abrupt and gradual changes in cue-outcome relations. We found no evidence that participants adapted to these types of change in qualitatively different ways. Also, in contrast to earlier claims that these tasks are learned implicitly, participants showed good insight into what they learned. By fitting formal learning models, we investigated whether participants learned global functional relationships or made localized predictions from similar experienced exemplars. Both a local (the Associative Learning Model) and a global learning model (the novel Bayesian Linear Filter) fitted the data of the first two experiments. However, the results of Experiment 3, which was specifically designed to discriminate between local and global learning models, provided more support for global learning models. Finally, we present a novel model to account for the cue competition effects found in previous research and displayed by some of our participants. Nothing endures but change Heraclitus (540 BC -480 BC).Predicting future events from past experience is a fundamental aspect of daily life. For instance, policy makers have to predict the outcome of interventions and sports coaches must

Testing adaptive toolbox models: A Bayesian hierarchical approach.

“…Likewise, von Helversen and argued that depending on the functional relationship between the cues and a criterion, different judgment processes take place. In contrast to these toolbox models, other theories assume that knowledge is not domain specific; the same process occurs independent of the context or the structure of the task (e.g., Busemeyer, Byun, Delosh, & McDaniel, 1997;Kelley & Busemeyer, 2008;Speekenbrink & Shanks, 2010). Because theories on function learning apply to continuous judgments, this example requires that we extend the proposed Bayesian toolbox framework beyond the discrete choice situations analyzed so far.…”

Persistence and plasticity in the human memory system: An empirical investigation of the overwriting hypothesis

“…It might seem ideal that whenever existing memory representations conflict with new information 17 in the environment we should simply replace the former with the latter, a process analogous to a computer overwriting information on its hard drive. However, a recent change in the environment may not necessarily be representative of more global regularities that emerge across longer time scales (Speekenbrink & Shanks, 2010). Local inconsistencies ('noise') could easily obscure useful patterns ('signals') that can only be detected when we gradually accumulate information over time.…”

Putting bandits into context: How function learning supports decision making.

“…While these kinds of tasks are ubiquitous in daily life, they are rarely studied within the psychological literature. This is unfortunate, as CMAB tasks encompass two important areas of cognition: experience-based decision making (Barron & Erev, 2003;Hertwig & Erev, 2009; and function learning (DeLosh, Kalish, Lewandowsky, & Kruschke, 2004;Speekenbrink & Shanks, 2010). Both topics have been studied extensively (see e.g., Newell, Lagnado, & Shanks, 2015, for an overview), but commonly in isolation.…”