Optimally adapting heuristics: humans quickly abandon the constant bearing angle strategy
22.13, Saturday, May 11, 10:45 am - 12:30 pm, Royal Ballroom 1-3
Constantin Rothkopf1,2, Paul Schrater3; 1Institute of Cognitive Science, University of Osnabrueck, 2Frankfurt Institute for Advanced Studies, 3University of Minnesota
Animals ranging from dragonflies through teleost fish to humans all intercept moving targets using the same strategy of adjusting their speed so as to hold the angle pointing towards their target constant over time. This constant-bearing-angle strategy has been suggested as a fundamental visuomotor heuristic and as an instance of Darwininan intelligence that overcomes the need for complex and expensive computations involving multiple sources of uncertainty. We consider the task of intercepting a moving ball for which many previous studies have shown that humans use this constant bearing angle strategy. Here we manipulated the observation function in a virtual reality setup so as to change the uncertainty of the balls position parametrically. Specifically, the contrast of the ball changes as a function of the heading angle towards the ball along the subjects momentary trajectory. Subjects adjusted their interception strategy within an average of 26 trials and were consistently able to catch these balls. To gain insight into the adopted new interception strategy, we setup an approximate optimal control models, which is provided observation function governing the uncertainty in state variables given visual observations. Parameters of this model are based on previous literature. The approach utilizes a Monte Carlo sampling of smooth trajectories of increasing complexity in a low dimensional parameter space. This analyses shows that the ideal actor modifies its trajectories by executing controls that increase information gain, and that these changes mirror human behavior. Thus, we provide evidence that humans quickly abandon the constant bearing angle strategy in favor of more informative action sequences, if this allows catching moving targets more reliably. The constant-bearing-angle-strategy is not an invariant heuristic of Darwinian intelligence as humans employ near-optimal information seeking actions that violate the constant bearing angle strategy, but produce less uncertainty in the interception.