Judgment & decision making

Reasoning

Music cognition & perception

Research Interest #1: Judgment and Decision Making

1. Goal specificity and its influence on Judgment

It has been difficult to define the scope of people¡¯s goals, because they have little insight into end states of the goals that drive behavior.  We provide evidence that the end states of goals may be specific by exploring patterns of preference change in the presence of active goals.  We have conducted experiments demonstrate that when a focal goal is active, items related to that goal increase in their preferability for a person (relative to the case where the focal goal is inactive) and that items unrelated to the goal decrease in their preferability.  Items intermediate in relatedness show neither a valuation or a devaluation in the presence of an active goal.  This pattern is consistent with a view of motivation that assumes there is a limited amount of motivational energy that is apportioned to active goals. 

We believe that patterns of valuation and devaluation can be used to explore the structure of people¡¯s goals.  The relationship between motivation and cognition has been difficult to illuminate, because people do not have insight into their own motivational states.  Thus, an indirect method is required.  Our current research is focused on exploring the many questions that arise from this research such as the relationship between goal activation, activation of concepts relating to goals, and how they influence preference and action. 

¡¡

2. Cultural differences in reasoning J&DM

Traditionally, the research of cultural differences in J/DM has been based on the well-known paradigm, Collectivism vs. individualism.  Such studies have provide beautiful pictures for the descriptive understanding of cultural difference.  However, this dichotomy-based approach, have some critical limitations.  We think one of the main reason is that they cannot explain the causal relationship between root variable and its influence on cognition and behavior.  It is because they did not have an detail/specific picture between independent and dependent variables.  Further this approach ultimately have no choice but to premise that there is a fundamental difference in "cognitive structure" between members of different cultures.  Considering that we already have a lot of general theory for the explanation of a variety of group difference, e.g., age or gender.  It is unlike that we need the assumption of the structural difference only when explaining cultural differences (I mean it's not parsimonious).  Rather the causal variable must be culture-common, and much more fundamental/basic than described in the paradigm.

In line with this idea, we have experimentally manipulated a target variable (e.g., fear of isolation) "within" a single culture.  And surprisingly we could find exactly the same pattern of difference between different levels (i.e., conditions such as high vs. low) of the variable with the results observed in the previous traditional culture studies.  That is manipulation of fear of isolation (FOI) reflects a difference in the relative preference for dialectical reasoning and relative sensitivity to contextual vs. target information (Kim and Markman, 2002;2003).  Furthermore, one of our ongoing project indicates that experimental manipulation of such variables can explain even much more complex human behavior like risk-taking. 

Research Interest #2: Reasoning

1. The relationship between theory and covariation information 

This study is attempting to understand the relationship between a reasoner¡¯s theory and processing covariation information in causal reasoning.  Basic hypothesis of this project is that knowledge about mechanisms influences attention to types of covariation information.  People¡¯s causality judgments are strongly influenced by their assumptions about the number of causes they believe underlie an effect.  The ultimate goal of this project is to know how individual pieces of covariation data are evaluated relative to a reasoner¡¯s hypothesis about the domain

Theories of causal reasoning have focused on the role of statistical relationships (e.g., covariation) and of knowledge of causal mechanism on learning of causal relations. Although both sources clearly inform causal judgments, most research examines these factors in isolation, or pits covariation-based data and theory-relevant information against each other as competing explanations for people¡¯s causal attributions. Few studies have explored the relationship between the two types of information in a direct way. In this paper we discuss how knowledge about mechanisms may influence attention to covariation information, and then test this view in two studies.
According to the covariation view, causal induction involves the acquisition of knowledge about contingencies (Wasserman, Chatlosh et al. 1983). However, a critical problem of the covariation view still remains. The statistical representation of an event is not as simple as described in covariation approaches (Spellman 1996; Cheng 1997; Spellman, Price et al. 2001). When a reasoner evaluates multiple causes, the number of possible combinations of contingencies increases factorially. It is unlikely that all of the covariation information are considered equally. Capacity of human cognition does not allow reasoner attend to and process all conditional contingencies within a set of events in environments.
The mechanism approach emerges from a viewpoint that causality is only indirectly reflected in covariation information. According to this approach, causal learning is guided mainly by prior knowledge about causal mechanisms that connect causes and effects (Ahn, Kalish et al. 1995; White 1995; Ahn, Kim et al. 2000). For example, Ahn et al. (1995) showed that reasoners do not spontaneously seek covariation information between potential causal candidates and effects, but, rather they gather further information regarding the specific target events in question to test hypotheses about possible underlying mechanisms. This view suggests that knowledge about causal mechanisms plays an important role in testing causal hypotheses and may even take priority over covariation-based data. Despite these descriptive differences between the covariation and mechanism views, the relationship between the two is quite poorly understood. One of purposes of the current study is to explore how covariation info is influenced by theory.
Although both theory and covariation inform causal judgments, the tendency of most research to date has been to examine one or the other factor in isolation, or to pit covariation-based data and causal theories against each other as competing explanations for people¡¯s causal attributions. One way to reconcile these the two views is to assume that beliefs about causal mechanisms constrain the statistical information to which we attend (Fugelsang and Thompson 2001; Waldmann and Hagmayer 2001). For example, Fugelsang and Thompson (2001) showed that reasoners weight covariation-based data more heavily for believable than for unbelievable candidates, but only when there is a causal mechanism that links the cause to the effect. Similarly, Spellman (2001) showed that judgments of causal strength are not always conditionalized on all covarying events but rather they are conditionalized when there is a reason to believe that a factor may be causally relevant to the effect. That reason may be a top-town causal theory or the bottom-up recognition of a covariation.
We suggest that people may use their beliefs about causal mechanisms to determine what statistical information to attend to. As described, a statistical representation for a cause is a collection of individual representations of covariation based on specific connections of the cause with the other causes and an effect. For example, Spellman (2001) discriminated CC for a cause with the other cause present (henceforth, CC-pairwise; because the two candidate causes appear together) and CC for a cause with the other absent (henceforth, CC-single; because the focal candidate cause appears alone) which are considered when controlling for alternative causes. However, the two CCs always had the same value in her experiments, making it difficult to determine out whether reasoners distinguished between these CCs. To demonstrate whether human covariation processing is specific to connections between causes (and an effect), we will construct multiple contingency organizations in which UC, CC-pairwise, and CC-single have different values within and between causes

¡¡