Academic journal article Memory & Cognition

The Importance of Decision Making in Causal Learning from Interventions

Academic journal article Memory & Cognition

The Importance of Decision Making in Causal Learning from Interventions

Article excerpt

Recent research has focused on how interventions benefit causal learning. This research suggests that the main benefit of interventions is in the temporal and conditional probability information that interventions provide a learner. But when one generates interventions, one must also decide what interventions to generate. In three experiments, we investigated the importance of these decision demands to causal learning. Experiment 1 demonstrated that learners were better at learning causal models when they observed intervention data that they had generated, as opposed to observing data generated by another learner. Experiment 2 demonstrated the same effect between self-generated interventions and interventions learners were forced to make. Experiment 3 demonstrated that when learners observed a sequence of interventions such that the decision-making process that generated those interventions was more readily available, learning was less impaired. These data suggest that decision making may be an important part of causal learning from interventions.

Causal knowledge plays an important role in everyday reasoning. Causal knowledge enables us to predict future outcomes, explain past events, control the present environment, and categorize novel information. A great deal of research suggests that both children and adults easily represent causal relations among events (Gopnik & Meltzoff, 1997; Murphy & Medin, 1985; Sperber, Premack, & Premack, 1995; Wellman & Gelman, 1992).

How are causal relations learned? Correlations among events can often be good indicators of the presence of some causal relations, but it is well known that observed associations are often insufficient to disambiguate causal structure (see, e.g., Bacon, 1620/1986). For this reason, much of causal learning in both formal (e.g., scientific) and informal settings takes place in the context of interventions. Interventions provide learners with conditional probability information that is often vital to causal learning (for formal accounts, see Pearl, 2000; Woodward, 2003). Interventions also provide learners with anticipatory information in the form of temporal priority: The variable intervened on should be considered a cause of any subsequent or concurrent event (see, e.g., Lagnado & Sloman, 2004). Several researchers have shown that people treat their own and others' intentional manipulative actions as interventions (in a formal sense) and learn causal relations better from these data than from observational data alone (Gopnik et al., 2004; Kushnir, Gopnik, Schulz, & Danks, 2003; Lagnado & Sloman, 2004; Steyvers, Tenenbaum, Wagenmakers, & Blum, 2003; Waldmann & Hagmayer, 2005).

Research on causal learning from interventions has often focused on how interventions lead to causal structure learning in simple scenarios involving two or three related events. Causal learning in the real world often involves learning a complex network of relations among many events. In such situations, there may be an additional benefit of interventions-particularly, a learner's own interventions-in that learners can be active in their learning process. Interventions can provide learners with certain decision-making demands: To learn from interventions, one must first decide what intervention to make. This decision-making process allows learners to use interventions to disambiguate particular causal structures-namely, those that they have in mind as potential models of the causal system.

The idea that active examination of data facilitates learning more than passive observation of the same data does is not novel. In perception, active observers-those who can control what aspects of the environment they perceive-make more accurate judgments about depth and changes to the environment than do those who simply view a scene (see, e.g., Larish & Andersen, 1995; see also Gibson, 1979). In developmental psychology, both infants and young children are more sensitive to information garnered from their own interventions than to that garnered from their observations when making inferences about intentions, problem solving, and causality (Fireman, Kose, & Solomon, 2003; Kushnir & Gopnik, 2005; Sommerville, Woodward, & Needham, 2005). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.