How Can People Be so Good at Intercepting Accelerating Objects if They Are so Poor at Visually Judging Acceleration?

Abstract

People are known to be very poor at visually judging acceleration. Yet, they are extremely proficient at intercepting balls that fall under gravitational acceleration. How is this possible? We previously found that people make systematic errors when trying to tap on targets that move with different constant accelerations or decelerations on interleaved trials. Here, we show that providing contextual information that indicates how the target will decelerate on the next trial does not reduce such errors. Such errors do rapidly diminish if the same deceleration is present on successive trials. After observing several targets move with a particular acceleration or deceleration without attempting to tap on them, participants tapped as if they had never experienced the acceleration or deceleration. Thus, people presumably deal with acceleration when catching or hitting a ball by compensating for the errors that they made on preceding attempts.

Keywords: Interception; acceleration; catching; learning; motion; motor control; vision.

Figure 1.

The participant stood in front of a large screen and tried to hit rightward moving targets (shown in red) with their index finger. The targets appeared some time after the index finger was placed at a starting point (shown in white). (a) In Experiment 1, they could hit the target wherever they wanted. The background was one of three photographs (here that of a wooden surface). (b) In Experiment 2, they had to hit the target within an indicated region of the background (within a black disk on a grey background).

Figure 2.

Median horizontal errors in Experiment 1. The three different amounts of deceleration (simulated friction) were interleaved at random with the target always moving across a wooden surface (same image), interleaved at random with the target moving across a surface of ice when the deceleration was 0 cm/s², across a wooden surface when the deceleration was 20 cm/s², and across a plane of sand when the deceleration was 40 cm/s² (different image), or presented in separate blocks of trials for each deceleration with its associated image (separate blocks). The faint symbols represent additional results for six authors in the different image condition. The dotted line shows the errors that would arise from not considering the acceleration during the last 116 ms.

Figure 3.

Median horizontal errors in Experiment 2. Error as a function of the trial’s position within the sequence of 12 trials with the same condition and acceleration. Means and standard errors of the participants’ median horizontal errors. The grey area indicates the target’s maximal extent. The curves are fits of the simple learning model to the mean data. The dotted lines at Trial 1 show the errors one would expect if participants used the acceleration during the previous sequence to predict the target’s displacement during the last 169 ms.

Figure 4.

Participants who moved more did not benefit more from the red targets. The inset shows a top and side view of all of one participant’s finger movements during the first 800 ms after red targets appeared. These traces are for the participant indicated by the arrow in the main panel. In the top view (movements parallel to the screen), the grey disk represents the starting point and the dotted line indicates the target’s path. In the side view, the grey line represents the screen surface.