The development of spatial location coding: Place learning and dead reckoning in the second and third years

There are two possible ways to code spatial location: viewer-referenced and externally referenced systems. The mature form of each system makes use of metric information (in forms of coding termed dead reckoning and place learning, respectively). Each system also exists in a simpler form not using metric information (termed response learning and cue learning, respectively). We report an experiment designed to examine the development of the two mature forms of spatial coding. Children aged 16 to 36 months were asked to search for objects hidden in a long rectangular sandbox, after they moved to the opposite side of the box, either with visual landmarks available or in a curtained environment. When moving in a curtained environment, and hence needing to rely primarily on dead reckoning, children across this age range performed at levels above chance but reliably less accurate than when they did not move. The size of the decrement due to movement did not decrease with age. Thus, the dead reckoning system may become available before 16 months but undergo no further improvement through 3 years. On the other hand, comparisons between children who moved in the curtained environment and children who moved with visual landmarks available showed that only children older than 21 months did better when external landmarks could be seen. This finding suggests a relatively late emergence of place learning.