Abstract
To avoid collision with an obstacle in the path of motion, the deceleration required to stop must not exceed the brake's maximum deceleration. Maximum deceleration thus defines a boundary separating possible from impossible stops. To keep required deceleration below the boundary, one must detect information about required deceleration. A combination of global optic flow rate and tau specifies required deceleration, but observers may rely on non-specifying variables, especially during early stages of learning. We developed an “emergency” braking task in which participants waited until the last possible moment to initiate maximum deceleration. If observers use a specifying variable, then the required deceleration at which braking is initiated should be unaffected by target size and initial speed. In Experiments 1 and 2, participants viewed computer simulations of approaches to a stop sign, and used a joystick to initiate braking at the last possible moment to avoid collision. Trials stopped when speed dropped to zero, so final distance to the sign provided feedback about the timing of brake initiation on each trial. Effects of target size and initial speed were significant, but were weaker at the end of the session, suggesting that novices rely on non-specifying variables and learn to use more effective variables with practice. To test whether similar learning effects occur during natural braking, blocks of regulated braking trials were interleaved between blocks of emergency braking trials in Experiment 3. Displays terminated at the onset of deceleration in emergency braking blocks to ensure that feedback was only provided during regulated braking. Analyses focused on effects of target size and initial speed on required deceleration at the onset of emergency braking both early and late in practice. We conclude that learning to perform a visually guided action such as braking involves attunement to more effective optical variables with practice.
Supported by the National Science Foundation BCS 0236734