Iversen Iver H, Matsuzawa Tetsuro
Department of Psychology, University of North Florida, Jacksonville, FL 32224, USA.
Anim Cogn. 2003 Sep;6(3):169-83. doi: 10.1007/s10071-003-0175-x. Epub 2003 May 22.
The experiments investigated how two adult captive chimpanzees learned to navigate in an automated interception task. They had to capture a visual target that moved predictably on a touch monitor. The aim of the study was to determine the learning stages that led to an efficient strategy of intercepting the target. The chimpanzees had prior training in moving a finger on a touch monitor and were exposed to the interception task without any explicit training. With a finger the subject could move a small "ball" at any speed on the screen toward a visual target that moved at a fixed speed either back and forth in a linear path or around the edge of the screen in a rectangular pattern. Initial ball and target locations varied from trial to trial. The subjects received a small fruit reinforcement when they hit the target with the ball. The speed of target movement was increased across training stages up to 38 cm/s. Learning progressed from merely chasing the target to intercepting the target by moving the ball to a point on the screen that coincided with arrival of the target at that point. Performance improvement consisted of reduction in redundancy of the movement path and reduction in the time to target interception. Analysis of the finger's movement path showed that the subjects anticipated the target's movement even before it began to move. Thus, the subjects learned to use the target's initial resting location at trial onset as a predictive signal for where the target would later be when it began moving. During probe trials, where the target unpredictably remained stationary throughout the trial, the subjects first moved the ball in anticipation of expected target movement and then corrected the movement to steer the ball to the resting target. Anticipatory ball movement in probe trials with novel ball and target locations (tested for one subject) showed generalized interception beyond the trained ball and target locations. The experiments illustrate in a laboratory setting the development of a highly complex and adaptive motor performance that resembles navigational skills seen in natural settings where predators intercept the path of moving prey.
这些实验研究了两只成年圈养黑猩猩如何在自动拦截任务中学会导航。它们必须捕捉在触摸显示器上可预测移动的视觉目标。该研究的目的是确定导致有效拦截目标策略的学习阶段。黑猩猩之前接受过在触摸显示器上移动手指的训练，并且在没有任何明确训练的情况下接触拦截任务。受试者可以用手指在屏幕上以任何速度将一个小“球”移向以固定速度在直线上来回移动或以矩形模式绕屏幕边缘移动的视觉目标。初始球和目标的位置在每次试验中都不同。当受试者用球击中目标时，会得到一小块水果作为奖励。在训练阶段，目标移动速度逐步提高，最高可达38厘米/秒。学习过程从仅仅追逐目标发展到通过将球移动到屏幕上与目标到达该点相重合的点来拦截目标。性能的提高包括运动路径冗余的减少和目标拦截时间的减少。对手指运动路径的分析表明，受试者甚至在目标开始移动之前就预测到了它将如何移动。因此，受试者学会了将试验开始时目标的初始静止位置用作预测信号，以确定目标开始移动后它随后会在哪里。在探测试验中，目标在整个试验过程中不可预测地保持静止，受试者首先会预期目标的预期移动而移动球，然后纠正移动方向，将球导向静止的目标。在新的球和目标位置的探测试验中（对一只受试者进行了测试）的预期球移动表明，拦截能力具有普遍性，超出了训练过的球和目标位置。这些实验在实验室环境中展示了一种高度复杂且适应性强的运动表现的发展，这种表现类似于在自然环境中捕食者拦截移动猎物路径时所看到的导航技能。
