Theory and Simulation in Decision Making

A central goal of contemporary cognitive science is the explanation of cognitive abilities or capacities. The dominant explanatory strategy for doing this assumes an internally represented knowledge structure which serves to guide the execution of the capacity to be explained. This knowledge structure is often described as the agents "theory" of the domain in question. Recently, however, a view appeared that holds that, rather than relying on theory, we interpret, explain, and predict human behavior by using a special sort of a mental simulation in which we take ourselves as a model and imagine being in those circumstances. According to simulation theory a fundamental role is played by our ability to identify with another person in imagination and to replicate or re-enact aspects of the other person`s mental life. It is a matter of heated dispute whether the faculty of decision making is driven by the use of theory or by simulation. We use a method developed by Perner et al. (1999) to tackle this issue. The basis for the method is that a theory is easier to apply if the critical factor is being highlighted. In contrast, simulation should work better the more similar the task conditions for the simulator are to the conditions of the simulated person. This means that a methodological rule can be used to differentiate between theory and simulation. We use the between-within contrast to test experimentally whether various cognitive illusions are due to theory or due to simulation. We use 4 focal phenomena: the framing effect (highlighting the positive aspects of formally identical problems leads to risk aversion and highlighting their equivalent negative aspects leads to risk seeking), the if-only effect (the nature and strength of emotions that follow from a particular outcome depends on the ease with which they can be mentally undone), the endowment effect (people become more attached to an object in their possession than they predict they would on the basis of their desire to possess the object before it belonged to them), and the fairness-effect in ultimatum games (people offer more than game rationality predicts). In addition to distinguishing between theory and simulation, the proposed experiments test how robust the effects in the decision literature are with respect to procedure; and they evaluate the fit between real and hypothetical decisions.

A central goal of contemporary cognitive science is the explanation of our ability to have an idea about the content of other people`s thoughts. We can do this by relying on our lay psychology of what people will think in a specific situation, or by imagining being in their situation. The former is reading other people`s mind by using theory the latter is doing this by using mental simulation. According to simulation theory a fundamental role thus is played by our ability to identify with another person in imagination and to replicate or re-enact aspects of the other person`s mental life. The present project undertook it to apply the theory-simulation distinction to decision processes. The idea was that just like putting oneself in others` shoes to understand them we could project ourselves into hypothetical situations, in order to understand our actual decisions. Using a method originally developed by Perner et al. (1999) we investigated whether simulation is possible in decision making. We found that some decisions were clearly simulated (e.g., the immediate endowment effect; gain-loss framing), while others were based on theory (e.g., the long term endowment effect; the influence of payoff size for risk taking), and still others were based on a mixture of theory and simulation (e.g., the if-only effect). Accordingly, some initial principles were formulated as to when and how simulation will lead to correctly predicting decisions. The findings offer clues as to how we predict decisions. Moreover, they enable the judgment of the frontiers of hypothetical prediction of own and other`s choices. For instance, the prediction of influences that develop over time is doomed to be incorrect, when based on simulation. These findings can be applied to a variety of personal, social, and economical situations where the prediction of ones own preferences, as well as the preferences of other people, is of crucial importance.